<%BANNER%>

Envisioning Urban Growth Patterns That Support Long-Range Transportation Goals - A Comparative Analysis of Two Methods o...

Permanent Link: http://ufdc.ufl.edu/UFE0042190/00001

Material Information

Title: Envisioning Urban Growth Patterns That Support Long-Range Transportation Goals - A Comparative Analysis of Two Methods of Forecasting Future Land Use Change
Physical Description: 1 online resource (186 p.)
Language: english
Creator: Thompson, Elizabeth
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Anticipating the future underlies much of the work of urban and regional planners who often rely on forecasts created by models attempting to simulate the real world. For planners concerned with the coordination of land use and transportation the complex and interrelated nature of the two presents many challenges to the development of models that realistically capture the intricacies of the land use and transportation relationship. Because land use change effects transportation demand, land use forecasting methods may thus have a significant effect on forecasts of transportation demand and ultimately influence the extent to which land use and transportation plans can be successfully coordinated. This study is based upon the principle that the method used to forecast future land use change in a region is influential to the achievement of long-range transportation planning goals. A set of interrelated land use and transportation planning goals are used as guidelines for the creation of three future land use scenarios for the study area, Lake County, Florida for the period 2007 to 2025. These goals focus on the discouragement of urban sprawl through a compact development pattern, and an increase in energy conservation through a reduction in single-occupant vehicle trips, vehicle-miles of travel (VMT) and increased transit ridership. Two different methods of forecasting future land use change are used in this study to model three future land use scenarios for the study area. One method uses the Florida Land Use Allocation Method (FLUAM) and is based on historical development trends, comprehensive plan policies and a mathematically derived gravity model to allocate future population and employment growth. The other method uses the Land Use Conflict Identification Strategy-Planning Land Use Scenarios method (LUCIS-plus) and is one based in land use suitability analysis and scenario planning techniques. The transportation demand of each scenario is modeled and several transportation variables are compared to determine the effect each future land use forecasting method has in the achievement of long-range planning goals. The results of this study show the LUCIS-plus method can create a future land use scenario where 11% more population reside within 3-miles of future transit routes and at a density 71% greater than in a scenario created using FLUAM. A more compact LUCIS-plus land use scenario can result in greater energy conservation through a potential decrease in vehicle-miles of travel of 5.5% when compared to the FLUAM scenario. Using the LUCIS-plus method to forecast future land use change can result in a regional urban form that is more supportive of long-range transportation goals than a land use pattern forecasted using the FLUAM method. The results of this study highlight the need for planning professionals and responsible government agencies in Florida to recognize the effect different land use forecasting methods can have on the coordination of long-range land use and transportation plans.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Elizabeth Thompson.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2010.
Local: Adviser: Zwick, Paul D.
Local: Co-adviser: Steiner, Ruth L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042190:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042190/00001

Material Information

Title: Envisioning Urban Growth Patterns That Support Long-Range Transportation Goals - A Comparative Analysis of Two Methods of Forecasting Future Land Use Change
Physical Description: 1 online resource (186 p.)
Language: english
Creator: Thompson, Elizabeth
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Anticipating the future underlies much of the work of urban and regional planners who often rely on forecasts created by models attempting to simulate the real world. For planners concerned with the coordination of land use and transportation the complex and interrelated nature of the two presents many challenges to the development of models that realistically capture the intricacies of the land use and transportation relationship. Because land use change effects transportation demand, land use forecasting methods may thus have a significant effect on forecasts of transportation demand and ultimately influence the extent to which land use and transportation plans can be successfully coordinated. This study is based upon the principle that the method used to forecast future land use change in a region is influential to the achievement of long-range transportation planning goals. A set of interrelated land use and transportation planning goals are used as guidelines for the creation of three future land use scenarios for the study area, Lake County, Florida for the period 2007 to 2025. These goals focus on the discouragement of urban sprawl through a compact development pattern, and an increase in energy conservation through a reduction in single-occupant vehicle trips, vehicle-miles of travel (VMT) and increased transit ridership. Two different methods of forecasting future land use change are used in this study to model three future land use scenarios for the study area. One method uses the Florida Land Use Allocation Method (FLUAM) and is based on historical development trends, comprehensive plan policies and a mathematically derived gravity model to allocate future population and employment growth. The other method uses the Land Use Conflict Identification Strategy-Planning Land Use Scenarios method (LUCIS-plus) and is one based in land use suitability analysis and scenario planning techniques. The transportation demand of each scenario is modeled and several transportation variables are compared to determine the effect each future land use forecasting method has in the achievement of long-range planning goals. The results of this study show the LUCIS-plus method can create a future land use scenario where 11% more population reside within 3-miles of future transit routes and at a density 71% greater than in a scenario created using FLUAM. A more compact LUCIS-plus land use scenario can result in greater energy conservation through a potential decrease in vehicle-miles of travel of 5.5% when compared to the FLUAM scenario. Using the LUCIS-plus method to forecast future land use change can result in a regional urban form that is more supportive of long-range transportation goals than a land use pattern forecasted using the FLUAM method. The results of this study highlight the need for planning professionals and responsible government agencies in Florida to recognize the effect different land use forecasting methods can have on the coordination of long-range land use and transportation plans.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Elizabeth Thompson.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2010.
Local: Adviser: Zwick, Paul D.
Local: Co-adviser: Steiner, Ruth L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042190:00001


This item has the following downloads:


Full Text





ENVISIONING URBAN GROWTH PATTERNS THAT SUPPORT LONG-RANGE
PLANNING GOALS -A COMPARATIVE ANALYSIS OF TWO METHODS OF
FORECASTING FUTURE LAND USE CHANGE














By

ELIZABETH A. THOMPSON


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS IN URBAN AND REGIONAL PLANNING

UNIVERSITY OF FLORIDA

2010




























2010 Elizabeth A. Thompson




























To Amelie Helen Thompson, my beloved daughter and guiding light









ACKNOWLEDGMENTS

My sincere thanks and gratitude are bestowed upon my generous and supportive

committee members Dr. Paul Zwick and Dr. Ruth Steiner. As chair of my committee,

Dr. Zwick has been instrumental in nurturing and guiding my research journey in a way

that continually challenges me to think "outside of the box" so that I might learn to see

more than what is first apparent. If not for Dr. Zwick's support and guidance I could not

have reached this important personal goal, and so to him I am truly indebted. As my

committee co-chair Dr. Steiner has been equally important by helping me navigate the

often murky waters of understanding the relationship between land use and

transportation. Without her vast knowledge, advice and guidance I would have certainly

stumbled along the way far more than I did, so to her I am also very thankful. Thanks

also to Dr. Zhong-Ren Peng for introducing me to the world of transportation demand

modeling and for gladly giving assistance whenever it was needed.

To my fellow LUCIS-ists, Iris Patten, Emily Stallings, Naser Arafat, Christen Hutton

and Yuyang Zou, I owe a large measure of thanks, not only for their technical support

and advice at all hours of the night and day but also for their sincere friendship which

has meant a great deal to me.

The love and support of my family made this whole journey possible, without their

encouragement I surely would not have had the strength to reach this goal. I shall

always be indebted to my parents Marshall and JoAn Wrightstone, and my parents-in-

law, Keith and Marilyn Thompson who have helped me in more ways than I can ever

mention here.

And above all others I owe the most thanks to my adoring husband Kevin and

precious daughter Amelie who have endured my lengthy distractedness, frequent









frustrations and occasional bad temper, but love me nonetheless. I could not have

done this without you both, nor would I ever want to.









TABLE OF CONTENTS

page

A C K N O W LE D G M E N T S .................................................................................. ... .... 4

LIST OF TABLES ...................................................................... ......... 10

LIST O F FIG URES........................................... ............... 12

LIS T O F A B B R EV IA T IO N S ............................................................ ............... 13

ABSTRACT .................................... ................................... ........... 15

CHAPTER

1 INTR O D U CT IO N ................................................................. .. ......... 17

The C challenge of Forecasting the Future....................................... .................... 17
Research Argument and Study Objective ............... ................................... 18
S tudy O outline ...................................................................... ......... 19

2 LITERATURE REV IEW .......................................................................... 21

Introd uctio n ............... .............. ....... .... ............ ...... ... ...... 2 1
The Land Use Transportation Relationship .................... ............................... 23
The Spatial Evolution of Urban Am erica................................. .................... 24
Walking-Horsecar era (1800 1890).......................... .............. 24
Electric streetcar era (1890 1920) ........... .. ........................ ............. .. 24
Recreational automobile era (1920 1945)................................... 25
Freeway era (1945 present)......... .... ........ .................... ............. 26
M obility and Accessibility Key Concepts............................... .................... 27
Trends in Planning Approach and Research......... ............. ............... 29
Meta-Analysis of Selected Literature Reviews .......... ............... ................. 34
B adoe and M iller (2000)............................................................. 36
Ew ing and C ervero (2001) ........... .... ..... .. .......................... ............... 37
Handy (2005) ................... .. ......... ........................... 41
Transportation Research Board (2009)............................. ... ............ 43
Sum m ary .................................................... .... ....................... 45
Transportation and Land Use Modeling .............. ......... ........................... 46
Transportation Modeling The Four Step Model ................ .................. 47
O verview ........... .............. ..................................... ..... 47
S ub -m o de ls ........................................... ...... ........... 4 8
Land Use M modeling ............. .......... ................... ..... ........... 53
Model types and operational characteristics ................... ............... 53
Meta-analysis of selected literature reviews..................... ......... 55
S scenario P planning ............................................ ........ .............. 59
History and Planning Applications ....................... .......... .. ............. 61


6









Strengths and Lim stations ........ ....................................................... ........ 63
Land Use Suitability Analysis ............. ............ ........ ............... 66
O overview ............. ........... ....... .. .. ....... .. ..... ............... 66
Land Use Conflict Identification Strategy (LUCIS).................. ........... 69
S u m m a ry .............. ..... ............ ................. ............................................... 7 4

3 STUDY AREA AND TIME FRAME ............................... .................. 77

Study Area Selection ....... .................. ........ ......... 77
Study Tim e Fram e ............ .................... ............ .............. ................ 86

4 METHODOLOGY ............................ ....... ................. 88

Methodology Overview ................ ......... ................. 88
Step O ne .......................... ............ .... ....................... 90
Florida Land Use Allocation Method Scenario Data ..................................... 90
Land Use Conflict Identification Strategy Planning Land Use Scenarios
D a ta .............. ..... ............ ................. ............................................. 9 1
S te p T w o ............................. ......................................... ...... ..... 9 2
Transportation Analysis Zone Data Overview .............. ........ ...... ......... 92
FLUAM the Florida Land Use Allocation Method................... ............... 93
Develop ent of control totals........................................... ............... 93
Determ nation of developable lands..................................... .................... 94
Input of approved developments, manual adjustments and overrides....... 94
Allocation of growth to TAZs ....... .................... ............... 96
LUCIS-plus The LUCIS Planning Land Use Scenarios Method .................... 97
Creation of combine raster to identify areas suitable for growth .............. 100
Establishment of existing gross densities ............................................ 110
Establishment of control totals for growth ......... ............... ................. 114
Allocation of existing population and employment using gross densities. 114
Establishment of guidelines for distribution of growth .............................. 115
Allocation of growth to suitable areas based on established guidelines... 120
Step Three ......... ...... .. .. ... .... ....... ....................... 126
FLUAM Scenario TAZ Data and Transit System ..................................... 127
LUCIS-Plus Scenarios TAZ Data and Transit System.............................. 128
Central Florida Regional Planning Model Model Run for Each Scenario .... 129

5 F IN D IN G S............................................... ....... .......... ...... 13 2

Com pact Developm ent .......... .......... ......... ................ ........ ....... 132
Energy Conservation .................. ..... ......... ... ....... ........ 135
Single-Occupant Vehicle and Transit Trips............ ................... ... ............. 139
S u m m a ry ......... ...... ............ ................................ ........................... 14 0

6 DISCUSSION .............. ......... ... ...... ...... ................... .......... 142

Discussion of Findings and Methods ...... ...................... ............. 142
Compact Development ............ ............ ...................... .... 142









Increased Transit Ridership.............................. ............... 144
Decreased Dependency on SOV .......................... .............. 146
Energy Conservation Reduction in VMT .............. ..... .................. 148
Regional Accessibility and the Challenges of DRIs............... ...... .......... 150
Proximity to Transit and Employment Densities ........... ... .......... ......... 151
Criticisms of the FSM and the CFRPM .................................................... 151
Complex Land Use Models vs. Deterministic Approaches to Forecasting
Land Use C change ........... ............................ ........... .. ............. 152
S tudy Lim stations .......... ......... ......... .......... ............... .. .......... 153

7 C O N C LU S IO N ............. ...... ... ....... ...... ... .......... ........ ............ 155

APPENDIX

A LAND USE MODELS REVIEWED BY AUTHOR AND MODEL TYPE ................. 161

B STRENGTHS AND LIMITATIONS OF LAND USE MODEL TYPES BY
A U T H O R ......... ................ ................................ ........................... 162

C DATA DICTIONARY FOR LUCIS-PLUS SCENARIOS..................................... 164

D ZDATA1 AND ZDATA2 DESCRIPTIONS .......................................................... 165

E ALLOCATION BUFFERS 1 AND 2 ................................................... ............. 166

F A LLO C A T IO N B U F F E R 3 ................................................................................... 16 7

G A LLO CATIO N BUFFER 4 ........................................................... .............. 168

H A LLO C A T IO N B U F F E R 5 ................................................................................... 169

I AGGREGATION OF FDOR LAND USE CODES IN THE PARCEL DATA TO
T A Z LA N D U S E C O D E S ...................................... ..................... ......................... 170

J AGGREGATION OF ECFRPC EMPLOYMENT CATEGORIES TO TAZ LAND
USE CODES............................................ .......... 171

K LAKE COUNTY 2009 2020 TRANSIT DEVELOPMENT PLAN MAP AND
DESCRIPTION OF MODES AND ROUTES................................................... 172

L TOD DESIGN STANDARDS USED AS STUDY GUIDELINES............................ 173

M GENERAL SELECTION CRITERIA FOR ALLOCATION INTO BUFFERS 1 TO
5 FOR LOW AND HIGH LUCIS-PLUS SCENARIOS ........................ ................. 174

N STUDY AREA ROADS USED IN CFRPM BY AREA TYPE .............................. 175

O LUCIS-PLUS LOW SCENARIO POPULATION DENSITY ................................... 176









P LUCIS-PLUS HIGH SCENARIO POPULATION DENSITY................................ 177

Q LUCIS-PLUS LOW SCENARIO EMPLOYMENT DENSITY .......................... 178

R LUCIS-PLUS HIGH SCENARIO EMPLOYMENT DENSITY................................ 179

LIST O F R EFER EN C ES ............................................... ........................ ........ 180

B IO G R A P H IC A L S K E T C H ......................................................................... .. ....... 186









LIST OF TABLES


Table page

2-1 Number of journal articles and book chapters resulting from literature search of
ScienceDirect database ....................... ............. ........... 30

2-2 Selected federal policies related to urban transportation planning .................... 32

2-3 Meta-analysis of selected literature reviews summary of findings.................... 45

2-4 Selected literature reviews for meta-analysis of land use models............. .......... 55

2-5 The six step LUC IS process ...................... .... ............................ ............... 70

2-6 Example of a subset of goals, objectives and sub-objectives for the agriculture
ca te g o ry ................................................ ....... .......... ...... 7 1

3-1 Florida's population growth rates by region percentage change per decade,
1900 to 2000........................................ ............... 77

3-2 Florida's projected population increase (per thousand people) by region, 2010
to 2 0 3 5 ............ .......... ................ ...... .......... .................................... 7 9

3-3 Florida's central region, gross population densities and increases from 2010 to
2 0 3 5 ... ... .......................................... ........ .......... ...... 8 1

3-4 Residence and work counties for workers in Lake, Osceola and Orange
C counties for 1990 and 2000 ...................... .... .......................... .... ........... 82

3-5 Median values of owner-occupied houses in 2000 and 2008.............................. 83

4-1 Total acreage by TAZ land use categories............................ ... ............ 111

4-2 Census Bureau estimates for 2007 for Lake County population by residential
TAZ land use categories........................................... ............... 111

4-3 TAZ estimates for 2007 for Lake County population by residential TAZ land use
categories ................ .... ..... .. ................ ........ ............. 112

4-4 Existing gross residential densities for study area ................... ...... ............ 112

4-5 ECFRPC estimates for 2007 for Lake County employment by TAZ land use
categories ................ .... ..... .. ................ ........ ............. 113

4-6 TAZ estimates for 2007 for Lake County employment by TAZ land use
categories ................ .... ..... .. ................ ........ ............. 113

4-7 Existing gross employment densities for study area ............... .................. 114









4-8 Population and employment control totals ................................ ............... 114

4-9 NEW LU codes ... .................................................................. 115

4-10 Transit systems A and B routes, modes, and headways............................... 127

5-1 Population and employment distribution FLUAM and LUCIS-plus scenarios
compared to 2007 TAZ data ................. ............. ... ............... 134

5-2 Average TAZ population and employment densities by buffers ........................ 135

5-3 Daily volume of vehicle-miles traveled in Lake County comparison of transit
systems by land use scenario............................... ............... 136

5-4 Daily volume of vehicle-miles traveled in Lake County comparison of all
scenarios for transit system B............................... ..... ............... 136

5-5 Vehicle-miles traveled in Lake County by transit system B comparison of
LUCIS-plus low and FLUAM scenarios by area type....... .............. ........ 137

5-6 Vehicle-miles traveled in Lake County by transit System B comparison of the
FLUAM and LUCIS-plus low scenarios by area type....... .............. ........ 137

5-7 Vehicle-miles traveled in Lake County by transit System B comparison of the
FLUAM and LUCIS-plus high scenarios by area type ................................ 138

5-8 Vehicle-miles traveled in Lake County by transit System B comparison of
LUCIS-plus low and LUCIS-plus high scenarios by area type....................... 138

5-9 Daily volume of home-based work and non-home-based work trips in Lake
County by mode all scenarios.............................. ............... 140









LIST OF FIGURES


Figure page

2-1 Example of mutually supportive land use and transportation planning goals....... 22

2-2 Methods of interpreting planning goals and their relationship ............................. 23

2-3 The Four-Step Model. ............ .................................................. 48

2-4 Exam ple of sim ple raster analysis............................................... .... ................. 68

2-5 Combining stakeholder preferences to show where conflict occurs.................... 73

3-1 Florida's population distribution by region, 1900 to 2000. ................ .......... 78

3-2 Florida's central region, projected population growth from 2010 to 2035.............. 80

3-3 Municipality, MPO and TPO boundaries for Florida's central region .............. .. 85

4-1 D iagram of study m methodology ......... ........................................... .. ............ ..... 89

4-2 Determination of developable lands in FLUAM ............................................ ..... 95

4-3 The 'com bine' process. .................... ............... ............................. 102

4-4 Model of combine raster for allocation buffer 1 ................................ 104

4-5 The region group process ...................... .. .. ............................... ............... 106

4-6 Calculation of mixed use densities for the low LUCIS-plus scenario ............... 117

4-7 Calculation of mixed use densities for the high LUCIS-plus scenario. ................. 118

4-8 Guidelines for allocation of growth for low LUCIS-plus scenario....................... 119

4-9 Guidelines for allocation of growth for high LUCIS-plus scenario ................... 120

4-10 Calculation of remaining growth to be allocated .......................................... 121

4-11 Calculation of school enrolment for 2025 for the low LUCIS-plus scenario........ 126

4-12 Transit system A routes and stops ........ ............. ............................ 128

4-13 Transit system B routes and stops ........ ............. ............................ 130

5-1 TAZs within and outside a 3-mile distance of future transit.............................. 133









LIST OF ABBREVIATIONS

BRT Bus Rapid Transit

CAA Clean Air Acts of 1970 and 1977

CAAA Clean Air Act Amendments of 1990

DOT United States Department of Transportation

DRI Development of Regional Impact(s)

ECFRPC East Central Florida Regional Planning Council

EIS Environmental Impact Statement(s)

FDOR Florida Department of Revenue

FDOT Florida Department of Transportation

FGDL Florida Geographic Data Library

FLUAM Future Land Use Allocation Model

FSM Four-Step Model

FSUTMS Florida Standard Urban Transportation Model Structure

GHG Greenhouse Gas

GIS Geographic Information System(s)

GUI Graphical User Interface(s)

ISTEA Intermodal Surface Transportation Efficiency Act of 1991

LRT Light Rail Transit

LRTP Long-range Transportation Plan

LSMPO Lake-Sumter Metropolitan Planning Organization

MPO Metropolitan Planning Organization

NEPA National Environmental Policy Act of 1969

PUD Planned Unit Development(s)

SOV Single-occupant vehicles)









TAZ Traffic Analysis Zone(s)

TOD Transit-Oriented Development

TDM Transportation Demand Model

TDP Transit Development Plan

TPO Transportation Planning Organization









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Master of Arts in Urban and Regional Planning

ENVISIONING URBAN GROWTH PATTERNS THAT SUPPORT LONG-RANGE
PLANNING GOALS -A COMPARATIVE ANALYSIS OF TWO METHODS OF
FORECASTING FUTURE LAND USE CHANGE

By

Elizabeth A. Thompson

August 2010

Chair: Paul D. Zwick
Cochair: Ruth L. Steiner
Major: Urban and Regional Planning

Anticipating the future underlies much of the work of urban and regional planners

who often rely on forecasts created by models attempting to simulate the real world.

For planners concerned with the coordination of land use and transportation the

complex and interrelated nature of the two presents many challenges to the

development of models that realistically capture the intricacies of the land use and

transportation relationship. Because land use change effects transportation demand,

land use forecasting methods may thus have a significant effect on forecasts of

transportation demand and ultimately influence the extent to which land use and

transportation plans can be successfully coordinated.

This study is based upon the principle that the method used to forecast future

land use change in a region is influential to the achievement of long-range

transportation planning goals. A set of interrelated land use and transportation planning

goals are used as guidelines for the creation of three future land use scenarios for the

study area, Lake County, Florida for the period 2007 to 2025. These goals focus on

the discouragement of urban sprawl through a compact development pattern, and an









increase in energy conservation through a reduction in single-occupant vehicle trips,

vehicle-miles of travel (VMT) and increased transit ridership.

Two different methods of forecasting future land use change are used in this study

to model three future land use scenarios for the study area. One method uses the

Florida Land Use Allocation Method (FLUAM) and is based on historical development

trends, comprehensive plan policies and a mathematically derived gravity model to

allocate future population and employment growth. The other method uses the Land

Use Conflict Identification Strategy-Planning Land Use Scenarios method (LUCIS-plus)

and is one based in land use suitability analysis and scenario planning techniques. The

transportation demand of each scenario is modeled and several transportation variables

are compared to determine the effect each future land use forecasting method has in

the achievement of long-range planning goals.

The results of this study show the LUCIS-plus method can create a future land use

scenario where 11% more population reside within 3-miles of future transit routes and at

a density 71% greater than in a scenario created using FLUAM. A more compact

LUCIS-plus land use scenario can result in greater energy conservation through a

potential decrease in vehicle-miles of travel of 5.5% when compared to the FLUAM

scenario. Using the LUCIS-plus method to forecast future land use change can result in

a regional urban form that is more supportive of long-range transportation goals than a

land use pattern forecasted using the FLUAM method. The results of this study

highlight the need for planning professionals and responsible government agencies in

Florida to recognize the effect different land use forecasting methods can have on the

coordination of long-range land use and transportation plans.









CHAPTER 1
INTRODUCTION

The Challenge of Forecasting the Future

Much of the work of urban and regional planners involves anticipating the future,

not simply how it will look but specifically how the actions of the present might change

what is yet to come. Invariably planners rely on forecasts of the future that most often

are created using models that attempt to simulate the real world. For planners

concerned with the coordination of land use and transportation, forecasting models are

integral tools for the creation and assessment of long-range plans aimed at minimizing

the negative impacts of urban growth. The relationship between land use and

transportation is complex however and its interrelated nature creates many challenges

to the development of models that can accurately capture its complexity. The

dimension of time adds to these challenges. Considering that the ramifications of

change in both land use and transportation systems often takes many years before

becoming apparent, the difficulty of creating realistic forecasts using models grows

exponentially.

The ongoing development of future land use and transportation demand models

has taken place over the past five decades. During this period several federal policies,

recognizing the need to anticipate the impacts of major transportation investments, have

mandated the use of forecasting models. Over time forecasting models have been

increasingly complex and scientific in their calculations, lending an air of inevitability to

their outputs with associated unintended consequences for the coordination of land use

and transportation plans. When a forecast of future land use change is regarded as a

certainty, alternative future scenarios with perhaps greater potential for minimizing long-









term transportation impacts may be overlooked in the planning process and

opportunities for better coordination are missed.

Research Argument and Study Objective

The idea that the method of forecasting future land use change is influential in the

achievement of long-range land use and transportation planning goals is the principle

upon which this study is based. To test this idea, a set of interrelated land use and

transportation planning goals will be used as guidelines for the creation of three future

land use scenarios. Two different methods of forecasting future land use change will

be used to model three future scenarios. The transportation demand of each scenario

will then be modeled and several transportation variables will be compared to determine

the effect each future land use forecasting method had in the achievement of the long-

range planning goals under investigation.

Current research interest in the field of land use and transportation planning has

centered on the relationship between urban form and transportation demand. This

study extends that research interest by testing forecasting methods against the

achievement of long-range goals that specifically relate to the land use and

transportation relationship. Lake County, in the rapidly growing central region of

Florida, was chosen as a study area, and selected goals from its comprehensive and

long-range transportation plans are used as guidelines for the creation of the scenarios

to be tested in this study. The two methods of forecasting future land use change differ

on several levels. One is the forecasting method historically used in Lake County, and

is based on historical development trends, comprehensive plan policies and a

mathematically derived gravity model to allocate future population and employment

growth. The other method is one based in land use suitability analysis and scenario









planning techniques that allocates future growth according to user-defined guidelines

and considerations of land use suitability. It is the contention of this study that the latter

of these two methods will result in a more satisfactory achievement of the long-range

planning goals under investigation in this study than the former method.

The results of this study will highlight the need for planning professionals and

responsible government agencies to recognize the effect different land use forecasting

methods can have on the coordination of long-range land use and transportation plans.

Should the difference in methods be significant, it may be necessary to re-evaluate the

current methodologies for forecasting land use at state, regional and local planning

levels. In the event that the results are inconclusive however, this study will serve as a

starting point for further research into ways in which the coordination of land use and

transportation plans can be improved.

Study Outline

The following chapters detail the research path taken in this study. Beginning in

Chapter Two, background information is assembled in a review of the literature on

topics relating to the relationship between land use and transportation, models of

forecasting transportation demand and land use change, scenario planning and land

use suitability analysis. The choice of study area and time frame are described in

Chapter Three, followed by an explanation of the study methodology detailing the

planning goals, forecasting methods, and transportation model used in testing the

research argument in Chapter Four. Chapter Five describes the findings of the study

and Chapter Six discusses those results in light of the literature reviewed, opportunities

for further research and limitations of the study. Chapter Six concludes the study with a









summary of its overall findings and a discussion of this study's significance in improving

the coordination of land use and transportation planning in the state of Florida.









CHAPTER 2
LITERATURE REVIEW

Introduction

The interdependent nature of the relationship between land use and transportation

is a central underlying theme in this study. Both land use and transportation are

demand-driven human activities and each has the potential to effect change in the

other. This is of particular importance to urban and regional planners as they formulate

and amend long-range plans that aim to minimize negative social, economic and

environmental impacts of urban growth. To fulfill that aim planners endeavor to create

land use and transportation plans that include complementary goals and objectives that

support mutually positive change in each of these key elements of their long-range

plans. This study focuses on such a set of complimentary goals; the encouragement of

a compact urban development pattern and a reduction of single-occupant vehicle (SOV)

use, increased transit ridership and increased energy conservation. Successful

achievement of any of these goals has the potential to positively impact the

achievement of any of the remainder. Figure 2-1 represents these interactions. The

effective coordination of land use and transportation goals thus has a significant bearing

on the success or failure of long-range planning initiatives and highlights the importance

of better understanding the relationship between land use and transportation.

Of similar importance to the effective coordination of planning goals are the

methods used to translate, quantify and represent them. For example, the method used

to translate a land use goal, such as encouraging a compact development pattern, into

a spatial representation like a map or geographic dataset may have a direct bearing on

how effectively the goal is interpreted and in turn the extent to which interrelated goals











like increasing transit ridership may be achieved (Figure 2-2). The effectiveness of

methods used to translate, quantify, and represent planning goals thus also plays an

important role in the attainment of long-range planning goals and the coordination of

transportation and land uses.




Increased densities = more potential riders proximal to transit

Compact Increased
Development Transit
Pattern Incern r.:I riiriase derinjtlC rourd Irar. l, Ridership



S Reduced land Fewer vehicle
consumption trips
o6
C- I
12 2 IncreasedM
E Energy
0 Conservation




S" Fewer vehicle
S E M trips




SSingle-Occupant
_F Vehicle Use





Figure 2-1. Example of mutually supportive land use and transportation planning goals.
Note. Diagram courtesy of author.



The following literature review addresses these two broad topics; the relationship

of land use and transportation, and methods used to represent land use and

transportation goals in long-range plans. Beginning with the land use and transportation

relationship the review will narrow its focus to literature relevant to methods of









forecasting future land use change and transportation demand. Trends in both land use

and transportation modeling will be reviewed as these tools are used in methods of

translating planning goals and are specifically used in this study. Finally literature

concerning scenario planning and suitability analysis will be reviewed as they form the

basis of one of the methods of forecasting land use change this study compares.


Figure 2-2. Methods of interpreting planning goals and their relationship. Note.
Diagram courtesy of author.


The Land Use Transportation Relationship

Land use and transportation have evolved together to produce the urban form and

travel patterns evident in the United States today. Muller (2004) describes this evolution

as taking place in four stages each predominantly driven by technological









advancements that have enhanced human mobility. The following overview broadly

highlights this interaction between transportation and land use and provides some

historical background to the discussion that follows.

The Spatial Evolution of Urban America

Walking-Horsecar era (1800 1890)

The introduction of the horse-drawn streetcar in the mid 19th century enabled

middle-income city residents to move to the urban fringes away from the overcrowding

and pollution that accompanied the industrialization of American cities (Muller, 2004, p.

64-67). Prior to this transportation innovation the urban area was limited to the extent to

which people could comfortably walk resulting in the clustering of people and activities

within close proximity of each other (Muller, 2004, p. 64). The horse-drawn streetcars

moved along rails which slightly improved their speed compared to moving along

unpaved roadways, attracting city residents that could afford the fare the opportunity to

move to the narrow strip of land on the cities periphery (Muller, 2004, p. 66). Initially the

streetcars followed radial routes but the demand for housing on the urban fringe saw the

construction of cross-town lines and infill development soon followed.

Electric streetcar era (1890 1920)

The invention of the electric traction motor was to revolutionize mobility and Muller

(2004) considers it one of the most important innovations in US history. By attaching

motors to streetcars speeds of up to 15 miles- per-hour could be achieved bringing a

much wider area on the urban fringe into commuting distance to the city (Muller, 2004,

p. 67). As streetcar lines ran further away from downtown areas development along

them changed the overall shape of cities from a circular to a star-shaped pattern.









Commercial development occurred alongside trolley tracks with residential streets

forming a grid pattern in between lines (Muller, 2004, p. 67).

The relatively low fare charged by streetcar companies coupled with extensive

networks of lines increased mobility for city residents and resulted in a greater

separation of land uses as people could now live further from centers of commerce and

industry (Muller, 2004, p.69). Specialized land use zones quickly emerged with the

central business district becoming the predominant location for such activities. The

invention of the elevator also enabled greater clustering as buildings grew in height

further concentrating commercial development downtown (Muller, 2004, p. 69). The

later years of this era saw the introduction of electric commuter trains in larger cities

such as New York that either superseded the streetcar systems or in the case of new

cities such as Los Angeles were adopted outright (Muller, 2004, p. 69).

Recreational automobile era (1920 1945)

The advent of the automobile was to have the most drastic impact on urban form

beginning slowly in the interwar period. Initially restricted to the wealthy, Henry Ford's

mass production techniques quickly made the car an affordable mode of transportation

for the majority of Americans (Muller, 2004, p. 70). Many of the earliest roads were

constructed in rural areas where farmers badly needed better access to local services.

City dwellers initially used their cars for recreational trips but quickly came to realize

their potential for personal daily travel. As early as 1922 the number of car dependent

households had grown to 135,000 across 60 cities (Muller, 2004, p. 70). The increasing

mobility that cars provided caused an increase in possible commuting distances and

spurred further development on urban fringes and between suburban rail axes. The

subsidization of the streetcar system by home-building companies was no longer









necessary as potential new homeowners provided their own means of transportation.

The demise of the suburban transit system soon began and was heightened during the

Depression years of the 1930s (Muller, 2004, p. 71-72).

Freeway era (1945 present)

In contrast to the preceding eras that were spurred by some innovation in

transportation technology, this period has been driven by a different force, the freeway.

Enabled by the massive highway-building building effort of the post-war economic

boom, the pattern of urban development rapidly dispersed as mobility increased along

the high-speed expressways that allowed even further separation of land uses (Muller,

2004, p. 75-81). Car ownership was no longer a luxury; it was a household necessity

for working, shopping and entertainment. Just as the streetcar system produced a

network-shaped development pattern, so too did the freeway system (Muller, 2004, p.

76); extending development into increasingly more distant locations and resulting in the

widely dispersed patterns of urban development common across the United States

today.

The regional advantage that city central business districts had in attracting

business and employment was mostly eliminated by the network of expressways that

now connected any location along its path (Muller, 2004, p. 76). Lower cost locations

for business along expressway routes attracted commercial, retail and light industrial

development to highway intersections in the outer city areas; creating suburban

downtown which in turn attracted residential development (Muller, 2004, p. 79). With

the ever increasing distance and separation of land uses, mobility for most Americans

had become dependent upon car ownership and the ability of the road network to take

them to the places the needed to go.









Mobility and Accessibility Key Concepts

The above overview highlights the affect transportation innovations have had on

urban form in the United States and the patterns of development that have evolved

within urban areas. Two key concepts emerge from this historical background that are

helpful in understanding the relationship between land use and transportation; mobility

and accessibility. The following section provides an overview of these concepts, so the

changing emphasis each has received in planning theory and practice may be better

observed.

At a very broad level the link between land use and transportation can be

understood in simple terms. Transportation can be described as the network of routes

taken to move from one location to another using various modes of transport (Meyer

and Miller, 2001, p. 128). How this network is organized is largely dependent on the

configuration of the land uses it serves as these dictate where people and the activities

they undertake are located and directly affects their mobility, or in other words their

ability to move from place to place. Similarly, land use patterns are influenced by the

transportation system which determines how accessible one location is to another

(Meyer and Miller, 2001, p. 128) and is dependent upon the available routes and modes

for travel.

As the above description highlights mobility and accessibility are two concepts

central to understanding the land use and transportation relationship. In this light

mobility can be seen as being principally a function of the characteristics of the

transportation network and the modes it employs, or more simply the means by which

people can travel. Accessibility however is linked to location and is characterized by

how easily a location can be reached using the transportation network as well as by the









number and type of activities a location offers and when these are available (Hanson,

2004, p. 5-6). Thus accessibility is dependent on mobility because it determines how

easily a location can be reached and perhaps more importantly by what modes.

Hanson (2004) points out that accessibility can be measured in different ways.

For example personal accessibility often refers to the number of activity sites within a

given distance of a person's home and how easily they can be reached, whereas

location accessibility refers to the number of activity sites within a given distance of a

specific place (p. 6). Both are measured in similar ways however there is an important

distinction between them. Measures of location accessibility treat all people within the

zone of measurement as being the same, so that those without access to a private

vehicle are considered to have the same accessibility as those who do (p. 6). Thus a

person living in walking-horsecar era town where people and activities were clustered

closely together would have had higher levels of personal accessibility than someone

living in a freeway era home without owning a private vehicle because the majority of

activity sites can only be reached by car. Although such measurements and examples

are overly simplified they do highlight the role mobility, and particularly available modes

of transportation, play in how accessible people and places are to each other.

The historical background of the spatial evolution of urban America highlighted

that as the demand for mobility increased the need for mobility increased

simultaneously as land uses became more segregated and people more car dependent.

In keeping with this demand, increasing mobility has been a paramount goal for

transportation planners who have often equated an increase in mobility with an increase

in accessibility (Hanson, 2004, p. 4-5). As Handy (2005) points out this planning









approach is problematic as it perpetuates a cycle of continually planning for mobility,

where increasing the capacity of road systems to reduce congestion and increase

mobility eventually leads to more travel, mounting congestion and a need to further

increase mobility (p. 11). While accessibility may be increased for those who own a car,

those unable to drive due to their age, disability or economic hardship are significantly

disadvantaged (p. 11). The likelihood of increasing transportation costs in the relatively

near future due to declining global oil reserves may create economic hardship and

declining accessibility for large numbers of people living in urban fringe areas where

driving distances are typically longer than average commutes.

Trends in Planning Approach and Research

Interest in understanding the relationship between land use and transportation is

not new. As early as 1930, research was undertaken to investigate the effect of a new

subway line on land values in surrounding areas (Meyer and Miller, 2001, p. 133). The

past several decades however have seen an increase in research interest in better

understanding the relationship. Meyer and Miller (2001) summarized 36 studies that

examined the land use and transportation relationship, spanning the years 1930 to

1997; 6 were from prior to 1970, 12 were from the 1970s, 3 from the 1980s, and 15 from

the 1990s. To substantiate whether this increase in studies represents an increase in

research interest in studying the relationship between land use and transportation, a

literature search was undertaken using the online ScienceDirect database. The

ScienceDirect database contains approximately 9.5 million journal articles and book

chapters from over 2,500 peer-reviewed journals and books (Elsevier, 2010). Several

different parameters regarding land use and transportation were queried and the results

are summarized in Table 2-1. The results of this literature search reinforce the review









undertaken by Meyer and Miller (2001) by highlighting that since 1970 research interest

has grown increasingly over the past four decades. Depending on the search

parameters used, the increase in the number of published journal articles and book

chapters on the topic of land use and transportation and related issues increased from

five to ten-fold between 1970 and 2010.

Table 2-1. Number of journal articles and book chapters resulting from literature search
of ScienceDirect database
1970
1961 1971 1981 1991 2001 to
to to to to to 2010
1970 1980 1990 2000 2010 %
Title, Abstract or Keyword = incr.
"land use" & "transportation" 2 67 70 129 398 494
"land use" & "transportation" & "transit" 1 18 22 42 134 644
"land use" & "transportation" & "density" 1 43 36 69 247 474
"land use" & "transp." & "accessibility" 0 19 25 35 108 468
"land use" & "transp." & "mobility" 0 8 11 35 105 1213
"land use" & "transp." & "VMT" 1 11 7 23 67 509
Note. The ScienceDirect online database was searched via the University of Florida Library
system at http://www.sciencedirect.com.lp.hscl.ufl.edu/ on March 9, 2010. The search included
all journals and books, from all sources and all subject areas.

The research trend demonstrated in Table 2-1 may be explained in the context of

changes to the land use and transportation planning process that began in the late

1960s with the introduction of several key federal policies that dictated a need for a

better understanding of the land use and transportation relationship. Table 2-2 lists

these policies and some of their key points that were to effect the transportation

planning process. The passage of these Acts collectively reflects changing American

attitudes towards their society and the environment.

Issues of social justice within urban areas came to the forefront of planning

discussions in the US during the 1960s and 1970s, with an emphasis on creating

redistributive policies that favored minorities and those under-represented in the









planning process becoming popular (Deka, 2004, p. 334). Urban unrest in many

American cities was the driving force behind this movement, with the Watts riot in Los

Angeles in 1965 typifying the tensions during this period. The McCone Commission

that investigated the Watts riot concluded that the social problems that led to the rioting

were in part due to a lack of adequate transportation services restricting the mobility and

accessibility of inner-city residents (Deka, 2004, p. 335). Three key policies reflect a

government response to these issues, the Federal-Aid Highway Acts of 1970 and 1973,

and the National Mass Transportation Assistance Act of 1974, all of which direct

government spending on mass transit. The observed increase in literature published on

the topics of land use, transportation, transit, density, accessibility and mobility (Table 2-

1) may be attributed to the introduction of these three pieces of legislation and the effect

they were to have on the planning process, spurring a need to better understand the

interrelationship of these topics.

The 1960s and 1970s was also the period at which environmentalism was at the

forefront of planning discussions, with environmental planning emerging as a profession

in its own right around this time (Deka, 2004, 345). Of greatest concern to

environmental planners was the protection of the natural environment from polluting

industries (Deka, 2004, 345). Such concern is reflected in the enactment of several

important pieces of legislation that were to have significant impacts on transportation

planning. They were the National Environmental Policy Act of 1969 (NEPA), the Clean

Air Acts of 1970 and 1977 (CAA) and the National Energy Act of 1978 (NEA). Each of

these acts was to place requirements on responsible agencies to monitor environmental

impacts of transportation investments.













Table 2-2. Selected federal policies related to urban transportation planning
Year Act Key Points


National Environmental
1969
Policy Act (NEPA)



Clean Air Act
1970
Amendments


1970 Federal-Aid Highway Act



1973 Federal-Aid Highway Act


National Mass
1974 Transportation
Assistance Act


Clean Air Act
Amendments



1978 National Energy Act



Clean Air Act
1990
Amendments

Intermodal Surface
1991 Transportation Efficiency
Act


Req'd preparation of env. Impact statements (EIS) for
major federal actions to include direct and indirect
effects (present & future) induced development
impacts of h'way projects are considered an indirect
effect and must be forecasted
Created Env. Protection Agency authorized to set
ambient air-quality standards req'd development of
state implementation plans (SIPs)
Req'd US Dept. of Transportation regulations to
assure adverse economic, social and env. effects are
fully considered in h'way projects
Allowed expenditure of federal-aid h'way urban
systems money to be spent on mass transp. projects
- allowed withdrawal of interstate segments and
substitution of mass transit projects

Authorized federal operating assistance for urban
transit systems

Req'd revisions to (SIPs) for areas not in attainment
of national air-quality standards SIPs were req'd to
develop transp. control plans to reduce mobile (from
transp. sources) emissions
All phases of transp. planning and project
development were to encourage fuel conservation -
states req'd to undertake conservation actions such
as car-pooling programs
Projected emissions associated with transp. projects
and programs must be reconciled with the req'd
emission reductions of SIPs
Req'd consideration of 15 planning factors in metro
transp. planning, relating to mobility and access for
people and goods, system performance and
preservation and environment and quality of life


Note. Adapted from "Urban Transportation Planning", by Meyer and Miller, 2001, pp. 619-629.









The substantial air quality problems being experienced in many US cities during

the 1960s and 1970s led to the amendments of the Clean Air Act (see Table 2-2) which

had originally been passed in 1955 in response to growing concerns about the health

problems associated with vehicle emissions (Wachs, 2004, 141). Requirements to

meet air-quality standards placed greater emphasis on understanding how the

interaction between land use and transportation influence travel behavior, the major

driver of the demand for mobility and automobile use. Increased interest in transit as a

means to reduce congestion, improve air quality and save energy is reflected in the

enactment of the Federal-Aid Highway Act amendments of 1970 and 1973 and the

National Mass Transportation Assistance Act of 1974. Seen together, the enactment of

the federal policies from the 1960s and 1970s listed in Table 2-2 help explain the

increased research interest during that period that the literature search details in Table

2-1 depicts.

The early 1990s saw two influential acts of federal legislation also have marked

affects on how transportation investments were planned. The Clean Air Act of 1970

(CAA) had already brought transportation planners into the sphere of environmental

planning by linking the automobile to the nation's air pollution problems (Hanson,

2004:24). The 1990 Clean Air Act Amendments (CAAA) strengthened that link by

requiring the integration of clean air planning and transportation planning at the regional

level, mandating that transportation programs must meet air quality standards. The

need to forecast future travel demand, or predict vehicle-miles of travel (VMT), as a

means of assessing impacts to air quality for a transportation project thus became a

legislative requirement for local, regional and state transportation planning agencies.









One particular conformity rule (40 CFR 93.122[b][1]) required Metropolitan Planning

Organizations (MPOs) to adopt some kind of land use forecasting model or committee

that could account for the regional impacts of transportation plans on land development

(Johnston, 2004:119).

Similarly the 1991 Inter-modal Surface Transportation Efficiency Act (ISTEA) had

an impact on the quantitative assessment of transportation plans. The Act required

State Departments of Transportation (DOTs) and MPOs to have coordinated long-range

transportation plans (LRTPs) and transportation improvement programs (TIPs) that

were now to include as considerations land use, inter-modal connectivity and transit

service enhancement methods (Bureau of Transportation Statistics, unknown). In

particular, the requirement to consider land use in transportation planning would compel

agencies to better understand the relationship between the two.

When analyzed in conjunction with an overview of several key federal policies

relating to urban transportation (Table 2-2), the increased research interest observed in

the literature search (Table 2-1) suggests that a trend exists that reflects firstly a gap in

knowledge about how land use and transportation interact, and secondly a recognition

that a better understanding of it may be necessary to address the social and

environmental problems urban planners seek to resolve.

Meta-Analysis of Selected Literature Reviews

The two-fold increase in literature published on topics relating to land use and

transportation between 1990 and 2010 (Table 2-1) can be furthered explained in the

context of two global concerns that have a direct relevance to these topics; adverse

climate change due to global warming and uncertainties about future global oil supplies.

For a country like United States, which had over two billion cars in 2005 (United Nations









Economic Commission for Europe, 2008), these issues have serious implications for the

future mobility of a population who depend almost entirely on personal vehicles for

transportation. According to the US Environmental Protection Agency (EPA)

transportation is the second largest contributor to greenhouse gas (GHG) emissions

(US EPA, 2009). Transportation is also the largest consumer of petroleum products in

the United States according to the US Government Accountability Office (GAO) (p. 9)

which also reported in 2007 that global oil supplies are predicted to begin declining

within the next 30 to 40 years (p. 4). Strategies to reduce GHG emissions and the

consumption of petroleum products involve technological advancements to improve fuel

efficiency in cars, increased use of alternative energy sources, and reducing fuel

consumption. A reduction in fuel consumption, or VMT, has thus become a serious goal

of transportation planners in the past decade and provides the motivation for many of

the more recent research studies in land use and transportation.

In response to the recognized need to reduce fuel consumption the US

Department of Energy requested the Transportation Research Board (TRB), a private

non-profit institution that provides services to the government and public on

transportation matters (TRB, 2010), to undertake a study into the relationships between

VMT, development patterns and energy consumption (TRB, 2009, p. xi). This report and

the literature reviews of three other authors will be examined and a meta-analysis of

their findings will be reported in the remainder of this section. Table 2-3 details the

authors being reviewed, the year of their publication, and a broad summary of their

findings.









Badoe and Miller (2000)

In their study into the transportation-land use interaction, Badoe and Miller (2000)

identify the current state of knowledge with a particular focus on potential policy impacts

and also to identify gaps in the knowledge to guide future research (p. 236). Their

overall goal is to generate discussion for the development and application of integrated

models of land use and transportation (p. 236). Their empirical review focused mostly

on studies published after 1994 and analyzed the data based on two categories;

literature on the impact of urban form on travel behavior, and literature on the impact of

transportation (transit in particular) on urban form (p. 236).

Their review of the literature on the impact of urban form on travel behavior was

further categorized according to five main topics; residential density, accessibility,

neighborhood design, car ownership, and transit supply. Their analysis of findings

related to residential density was that the evidence was very mixed. Although some

studies showed that density had a significant effect in increasing transit use or

decreasing VMT, other studies found that the significance decreased when other factors

such as socio economic status and car ownership were considered (p. 248).

Findings related to accessibility were also mixed due to the fact that most studies

assess its importance relative to other factors and these factors varied across the

studies (p. 251). Most studies overlooked the significance of connectivity (of the road

network) which is often a major determinant of accessibility (p. 252). Studies

concerning neighborhood design provided a mixture of findings with the principal flaw of

most relating to the scale of their investigation. Because the geographic range of

activities for most people extends far beyond the neighborhood scale the complex









relationships between neighborhoods and regions may be overly simplified by this type

of analysis (p. 253).

Findings related to car ownership were consistent across the studies reviewed,

with the common result showing that higher residential densities are associated with

lower car ownership and households with fewer cars tend to use transit more than those

with more cars (p. 253). Few studies investigated the impact of transit use on urban

form but those that did showed transit played a significant role in explaining mode

choice, VMT, and effects of residential density (p. 254).

Studies investigating the impacts of transit on urban form focused mostly on rail

transit (subway, light and commuter rail) and resulted in variety of findings. Most

studies concentrated on transit's influence on land values with the overall conclusion

that rail development facilitates rather than generates new development (p. 259).

The overall finding of the literature review by Badoe and Miller (2000) was that

there was a wide variety of conclusions being drawn across various studies about the

strengths and weakness of the relationship between land use and transportation (p.

260-261). Methodological and data weaknesses were largely responsible for a lack of

clarity with respect to potential policy impacts. They conclude that an integrated urban

model that accounts for all actors and factors in the urban system should be an area for

further research (p. 261).

Ewing and Cervero (2001)

In their study of travel and the built environment Ewing and Cervero (2001) set out

to provide a generalization across a large number of previous studies into the

relationship between travel behavior and urban form. Many existing studies focus

mainly on research findings without elaborating on methodological details making it









difficult to assess their reliability and validity (p. 87). Ewing and Cervero aimed to

collectively assess the body of literature with a focus on providing more detail of how

studies were done so differences between them could be identified (p. 87). They take

an empirical approach by reviewing 46 studies focusing on how each explains four

types of travel variables; trip frequencies, trip lengths, mode choices, and cumulative

person-miles traveled or vehicle-miles traveled or vehicle-hours traveled (p. 87). The

studies are categorized according to the general characteristics of the built environment

they investigate; neighborhood and activity center designs, land use patterns,

transportation networks, urban design features, and composite transit- or pedestrian-

oriented design indices. A table for each category summarizes the studies it includes

and the land use-transportation relationships identified.

Studies that compared neighborhood and activity center designs were furthered

categorized as being contemporary, or traditional, car or pedestrian oriented, and urban

or suburban (p. 88). Across all these sub-categories of the built environment trip

frequencies differed very little, with socio-economic characteristics of households being

a major determinant. Although evidence was limited trip lengths appeared to be shorter

in traditional neighborhoods which would be expected due to the finer land use mixes

and grid networks characteristic of this type of development (p. 88). Walking and transit

use are also more prevalent in traditional settings but this could be due to self-selection;

those who prefer these modes choose to live in settings where it is possible (p. 88).

Studies that tested land use variables were far more prevalent than any other type

of study (p. 92). These have generally focused on residential and employment

densities, measures of land use mix, and measures of accessibility that reflect the









number of attractions with a specific distance of households (p. 92). Overall these

studies showed vehicular travel is mainly a function of regional accessibility, even when

the effects of local density and land use mixed are controlled. This indicates that higher

density, mixed use developments that are regionally isolated from other activity sites

may be of only modest regional travel benefit (p. 92). Trip frequencies were mostly

dependent upon socio-economic characteristics rather than land use variables (p. 92).

Mode choice was the travel variable that was most affected by local land use variables

with transit use being firstly dependent on local densities and secondly on land use mix,

and walking equally dependent on both (p. 92). Employment densities at trip

destinations was possibly more important at destinations than population densities at

origins for both transit and walking modes, meaning the preoccupation with residential

densities by proponents of transit-oriented development (TOD) may be misguided (p.

92). Many studies focused on density but whether the impact of density on travel

behavior is due to density itself or other variables is still not determined (p. 92-93).

Many transportation network variables can affect travel times by different modes

and can potentially affect travel decisions. Some include street connectivity, routing

directness, block size and sidewalk continuity (p. 100). Walking and transit access are

improved by grid patterns of streets but so is car access, so it is difficult to determine

which mode gains the most advantage from this configuration (p. 100). The

attractiveness of network types to particular modes depends on design and scale, with

grid patterns of narrow streets being more attractive to walking than car travel (p. 100-

101). Evidence showing a relationship between network design and travel are mixed so

firm conclusions could not be drawn (p. 101).









There were relatively few studies that tested urban design variables such as

crosswalks, sidewalks, parking supply and building orientation. Individually design

variables were seldom significant in impacting travel and those that appeared to such as

crosswalks near bus stops more likely were capturing some other unmeasured feature

of the built environment (p. 102). Collectively however urban design variables may

have some impact on travel. Studies of composite land use design features, those

focusing on transit or pedestrian-oriented design may show an interactive effect

between land use and transportation variables (p. 106). For example a high traffic area

with no sidewalks may lower accessibility whereas individually each of these factors has

little impact on travel (p. 106). Different studies used different composite measures with

some being more subjective than others, and some arbitrarily weighting variables and

others using statistical estimates to base weights according to their associations with

other variables (p. 106). Overall these disparate approaches to measuring, for example

transit friendliness or walking quality, have resulted in inconclusive results about the

relationship between composite design measures and the impact on travel (p. 106).

Generalizing across all the studies Ewing and Cervero conclude that trip

frequencies are mainly a function of socio-economic characteristics and then the built

environment, whereas trip lengths are a function of the reverse, firstly the built

environment and then socio-economic characteristics (p. 106). Mode choices are

dependent equally on both the built environment and socio-economic characteristics,

and vehicle or person miles traveled is most significantly affected by the built

environment (p. 107). The authors call for more transparent ways of reporting results of

studies on the relationship between land use and transportation and suggest as an









example an approach involving the measurement of the elasticity of VMT in relation to

land use and design variables (p. 107).

Handy (2005)

Handy (2005) sets out to test four often stated propositions of advocates for Smart

Growth approaches to planning that encourage urban development to be compact in

form, pedestrian friendly, accessible to transit and diverse in land uses (p. 147-148).

Handy reviews the studies of the relationship between land use and transportation to

uncover evidence that support the following propositions:

Building more highways will contribute to sprawl (p. 148)

Building more highways will lead to more driving (p. 148)

Investing in light rail will increase densities (p. 148)

Adopting new urbanism design strategies will reduce automobile use (p.

148),

With regard to the first proposition, evidence from the literature shows that

highway building does contribute to sprawl by influencing where growth occurs. Rather

than generating growth the research showed the effects are redistributive and overall

highway building influences where and at what densities growth occurs (p. 152).

Handy points out that while this appears to be true the converse is probably not. In

other words, not building highways will not slow sprawl, (p. 153). For example the

expectation of building a highway may be sufficient to induce new development.

Studies in the 1990s showed that, with regard to the second proposition, a

statistically significant relationship between increasing highway capacity and increased

travel demand existed (p. 154). Their results suggested that in economic terms the









travel time savings gained by increased road capacity caused an increase in travel

consumption (p. 154). However more recent studies using disaggregate approaches

and modeling techniques that better identify causalities suggest that the earlier studies

overestimated impacts and that highway building has only a limited affect on induced

travel demand (p. 154). Handy concludes that it has yet to be concluded as to whether

or not increasing highway capacity contributes to a growth in VMT (p. 154).

Just as the literature showed that highway building had a redistributive rather than

a generative effect on new development, the evidence in support of the third proposition

that light rail transit (LRT) will increase densities was similar (p. 156-157). The

literature did show however that under certain circumstances LRT may increase

densities, although it was not assured. These circumstances included significant growth

occurring in a region, a transit system that significantly increases accessibility, station

locations sited in areas conducive to development, and supportive land use policies and

capital investments (p. 159).

Handy points out that with regard to the fourth proposition that new urbanism

design strategies will reduce car use, researchers are challenged by the difficulty of

separating out the relative importance of socio-economic characteristics from the effects

of design (p. 161). The problem of self-selection, that people who prefer to drive less

seek out neighborhoods where it is possible to drive less, was addressed by a few

researchers but overall they failed to capture people's motivation for choosing the

residences (p. 162). Handy concludes that new urbanism design strategies may reduce

car use by a small amount by addressing the unmet needs of these wanting to live in

such neighborhoods (p. 162).









In conclusion Handy finds that for all the propositions questions remain as to how

strong the link is between land use and transportation and the direction of causality in

factors affecting the relationship (p. 164. She summarizes her conclusion as follows:

New highway capacity will influence the location of growth (p. 163)

New highway capacity may slightly increase travel (p. 163)

LRT can encourage higher densities under certain conditions (p. 163)

New urbanism strategies make driving less easier for those wish to drive

less (p. 163).

Transportation Research Board (2009)

The TRB report investigating the effects the built environment has on driving

devotes a chapter to a review of the literature on the impacts of land use patterns on

VMT and draws on the studies covered by the three literature reviews summarized

above, Badoe and Miller (2000), Ewing and Cervero (2001) and Handy (2005) (p. 39).

As such focus here will be given to the more recent studies, published after 2005, that

TRB investigates in their literature review.

In an effort to account for the problem of self-selection, some of the more recent

studies have carefully controlled for a wide range of socio-economic variables to test for

their effect on VMT (p. 43). While some have investigated the effect of only one factor,

density, on VMT, a thorough study by Bento et al. (2005) controlled for several

variables, population centrality, jobs-housing balance, city shape, road density and rail

supply and found each to have a significant albeit statistically small effect (p. 43). The

work of Bento et al. suggests however that when these factors are considered

simultaneously VMT could be lowered by as much as 25 percent (p. 46).









The effects of TOD on travel were investigated by several recent studies which

indicate that transit supply and accessibility in combination with land use are important

variables affecting mode choice and thus VMT (p. 46). Of particular importance

appears to be the location of a TOD within a region because this affects its accessibility

to desired locations, as does the quality of connecting transit services which were found

to be more influential in travel patterns than the actual design characteristics of the TOD

itself (p. 46). With regard to land use and design features, proximity to transit and

employment densities at trip ends appears to have a stronger influence on transit use

than urban design features to enhance walkability and land use factors such as mixed

uses and increased residential densities (p. 47).

The literature review of this report concludes that few of the studies consider the

potential a group of policies that combine increased density with higher concentrations

of employment, improved accessibility to a variety of land uses and a good transit

network can have on reducing VMT (p. 31). The authors consider two case studies

where such a combination of factors has been fostered by different policy approaches,

Portland Oregon, and Arlington County Virginia (p. 51-53). They conclude that the

dramatic changes to the built environment and travel patterns require a significant

political commitment and substantial transportation investments over a long period of

time which will be challenging for many metropolitan areas (p. 54).









Table 2-3. Meta-analysis of selected literature reviews summary of findings
Year Author(s) Summary of Findings
Findings of studies very mixed with some showing a
2000 B e Mir strong link between urban form and travel behavior
2000 Badoe & Miller
while others show a weak link similarly mixed
results for transit impacts on urban form.


2001 Ewing & Cervero





2005 Handy






2009 TRB


Trip frequencies are a function of 1) socio-economic
characteristics and 2) built environment trip lengths
are a function of 1) built environment and 2) socio-
economic characteristics mode choices are a
function of both for VMT the built environment is
most significant.
New highway capacity will influence the location of
growth new highway capacity may slightly increase
travel light rail can encourage higher densities under
certain conditions new urbanism strategies makes it
easier to drive less for those wish to drive less.
Widely varying results across studies prevent any
conclusive finding on the importance of changes in
land use and the magnitude of their effects on travel -
recent studies controlling for confounding variables
find modest effects of the built environment on VMT -
however there is some evidence that multiple factors
such as population centrality, jobs-housing balance,
regional transit supply when implemented together
may achieve greater reductions in VMT.


Summary

The relationship between land use and transportation is complex with each having

the ability to influence change in the other. The land use patterns and transportation

systems seen in the United States today are a result of over one hundred years of

urban development that has been shaped by several innovations in transportation

technology. Of these the automobile has by far been the most significant in effecting

change in development patterns by affording Americans an unprecedented level of

personal mobility that has had positive and negative outcomes for people and their

environments. In attempts to control some of the negative impacts of transportation

government policies have been introduced that have placed specific requirements on









planning, implementation and management of the transportation system. The literature

reveals that while research into better understanding the land use and transportation

relationship has been influenced by government policies and public concerns about

uncontrolled urban development, recent global concerns about climate change and

crude oil shortages have lead to even more concerted research in this area. While

much has already been learned the literature shows that there is still a considerable gap

in the knowledge of how land use and transportation systems should be managed in

order to avoid negative social, economic and environmental impacts and still provide an

adequate level of mobility and accessibility for everyone.

Transportation and Land Use Modeling

The use of computer models in the preparation of transportation plans has been

common practice since the 1960s following the introduction of Federal Aid Highway Act

of 1962, which required all federally funded highway projects to be based on a

continuing, comprehensive and cooperative planning process involving state and local

planning agencies (Meyer and Miller, 2001, p. 619). An integral part of that process is

the forecasting of transportation demand and how that will impact the transportation

network. This is typically performed using computer models, as are forecasts of land

use change that determine the location of people and activities upon which

transportation demand is estimated. Both model types form the basis of a framework

generally used by transportation planners in the creation of long rang transportation

plans. This section will review the transportation and land use models commonly used

in this framework and how that has changed over time, with special attention focused on

the model types used in this study.









Transportation Modeling The Four Step Model

A model can be described as a simplification or abstraction of a real world system

that can be used to test what might happen to the system if changes occur within it

(Meyer and Miller, 2001, p. 256). In the history of transportation planning one model

type has been dominant since the early 1950s (McNally, 2007b, p. 35 -36). The urban

transportation modeling system, most commonly referred to as the four-step model

(FSM) was originally developed for evaluating large-scale infrastructure projects at the

regional and sub-regional scale (McNally, 2007b, p. 38). Despite its limitations which

will be reviewed later it remains the primary tool used for forecasting future

transportation demand and performance (McNally, 2007a, p. 54).

Overview

The FSM is made up of four sub-models that use data inputs concerning the

characteristics of the transportation system and the characteristics of the land use

system (Figure 2-3). The transportation system is represented graphically in the model

as a network of links and nodes which have information such as length, speed, and

capacity for links, and turn prohibitions and penalties for nodes associated with them

(McNally, 2007b, p. 38). The land use system is aggregated to traffic analysis zones

(TAZs) containing socio-economic and land use data typically comprised of census data

and/or data collected from household and workplace travel surveys, vehicle intercept

station surveys or onboard transit surveys (Meyer and Miller, 2001, p. 191-199). The

data typically reflects activities performed over a 24-hour period and includes activity

type, location, duration, and arrival and departure times (McNally, 2007b, p. 40).

Following is an overview of the four sub-models that typically make up the FSM.













































Figure 2-3. The Four-Step Model. Note. Adapted from "The Four-Step Modef' by M.G.
McNally in "Handbook of Transport Modelling", Button and Hensher (eds),
2007b, p. 39.

Sub-models

Trip generation. In the first sub-model the amount of total daily travel is

determined for various activities or trip purposes. These are typically categorized as

home-based work (HBW) trips, home-based other (non-work) trips (HBO), or non-home-

based (NHB) trips. Each trip has two ends, an origin and a destination and each trip


48









end is classified as a production or an attraction. Productions are considered to be the

home-end of a home-based trip, or the origin of a NHB trip, while attractions are the

non-home end of a home-based trip, or the destination of a NHB trip (Meyer and Miller,

2001, p. 271). Separate models are used to define productions and attractions using

socio-economic data at the TAZ level. For the calculation of productions data such as

household income, size and employment, car ownership, residential densities and

distance to major activity centers are used. Workforce data such as the number of

employees, floor-space areas and accessibility to workplaces are used for the

calculation of attractions (Meyer and Miller, 2007, p. 271). The production and

attraction models basically calculate a measure of attractiveness for each TAZ which

serves to scale the trips for the next stage of the modeling process, trip distribution.

Where certain land uses generate untypical activities within a TAZ, such as a university

or hospital, special generator factors are included in the production and attraction

models to account for this activity (McNally, 2007b, p. 43).

Trip distribution. The objective of this sub-model is to distribute or connect the

zonal trip ends, the productions and attractions estimated for each zone in the trip

generation model, in order to predict the flow of trips from production zones and to

attraction zones (Meyer and Miller, 2001, p. 278). While there are many types of trip

distribution models, the most commonly used type is the gravity model. This type of

model estimates a travel impedance or friction factor (in terms of travel time or cost)

between two zones as a function of both the land use and transportation systems

characteristics (McNally, 2007b, p. 45). The transportation model used in this study, the

Central Florida Regional Planning Model (CFRPM) follows the Florida Standard Urban









Transportation Model Structure (FSUTMS) which specifies the use of a gravity model

for calculating trip distribution (Gannett Fleming, unknown, p. 45). In this gravity model

the desirability of traveling to a zone is directly related to its potential as a destination

based on the activities within that zone, and is inversely related to the spatial separation

(friction factor) between production and attraction zones (Gannett Fleming, unknown, p.

45). The final stage of trip distribution involves calibrating the model to the observed

data and often involves the manual adjustment of friction factors so a best-fit can be

made between the observed and estimated trips (Meyer and Miller, 2001, p. 280).

Mode choice. The purpose of the this sub-model is to factor the trip tables from

the trip distribution model to produce trip tables specific to each mode being modeled

(McNally, 2007b, p. 48). The most commonly used model type for this step is the

nested logit model, which is used in the CFRPM. Zone to zone person-trips are

allocated by trip purpose across the various modes and coefficients are used that

quantify the sensitivity or elasticity of each mode choice to changes in service (Gannett

Fleming, unknown, p. 71).

Traffic assignment. In the final sub-model the predicted flows between origin-

destination pairs for each mode are assigned to actual routes on the transportation

network. In determining route selection most models assume that individuals seek out

the most cost effective route, in time or money. The CFRPM employs an equilibrium

technique that runs several iterations of a minimum-path capacity constraint assignment

with travel time delays for each iteration being incorporated into subsequent ones

(Gannett Fleming, unknown, p. 77). The final result of this model step is an estimate of

trip volumes and speeds on the network links. In the CFRPM highway and transit trips









are assigned in two different sub-models. The outputs from the highway assignment

include statistics on the number of lane and system miles, VMT, VHT (vehicle-hours

traveled), as well as original and congested speeds. The transit assignment outputs

include transit boards by mode, route and operator, as well as boarding and departure

statistics for each stop by route (Gannett Flemming, unknown, pp. 85 88).

Once the model is calibrated to reflect observed data it can be run for future year

estimates based on projected population and employment data and new land use

configurations. This study involves the modeling of three future land use scenarios

using the CFRPM.

Limitations of the FSM. An often cited criticism of the FSM is that it was

originally designed for evaluating large-scale highway projects and is policy sensitive

only with regard to major capacity improvements (McNally, 2007a, p. 54). As such the

FSM is not considered to be very effective in analyzing policy sensitivity because it fails

to adequately capture the complexities of travel, particularly at smaller scales. The FSM

focus on trips rather than activities results in the two not being linked during the

modeling process which is problematic because it is the underlying activity behavior that

generates trips (McNally, 2007a, p. 55). Also the FSM does not reflect temporal

constraints and other dependencies of activity scheduling such as trip-chaining,

meaning the sequencing of several trips to maximize time and cost savings (McNally,

2007a, p. 55). These limitations, among others, have led to continued research and

development of alternative modeling approaches.

Alternative approaches to the FSM. The FSM is an aggregate level model,

meaning it uses zonal data to model predictions about individual trips. An alternative









method is the disaggregate approach that begins by simulating the decision process of

individuals and summing them up to calculate aggregate travel demand (Meyer and

Miller, 2001, p. 290). Often referred to as discrete choice models, these typically are

based on the assumption that trip makers always seek to maximize utility, that is, they

rank trips according to their relative desirability, time/money cost, comfort and/or

convenience (Meyer and Miller, p. 290). Discrete choice models may utilize different

concepts to assume how trip decisions are made, such as random utility or nested logit

models like those commonly used in the mode choice step of the FSM.

Activity-based models are another approach which seeks to account more for the

behavioral aspects of trip making decisions. Models of this type typically fall into two

broad categories, econometric models and hybrid simulation models (Meyer and Miller,

2001, p. 306). Econometric models contain systems of equations that calculate activity-

travel choices as probabilities of possible outcomes or decisions. Hybrid simulation

models typically consist of rule-based algorithms that attempt to replicate actual

decision making processes (Meyer and Miller, 2001, pp. 301-307).

Integrated land use and transportation models are another approach to improving

the predictive capabilities of travel demand forecasting. Such an approach involves the

creation of feedback mechanisms between both model types within a modeling

framework which acknowledges the influence that land use, or human activities, has on

travel behavior, which determines the transportation network and in turn influences the

location and intensities of land uses (Miller et al., 1998, p. 1). Different land use models

and how they have been incorporated into the transportation modeling framework will

be discussed in more detail in the following section.









Land Use Modeling

At the basis of integrated land use and transportation modeling frameworks is the

idea that the relationship between the two can be simulated in a way that meaningfully

represents reality. Since the late 1950s a number of theoretical frameworks have been

developed by researchers that have significantly influenced the current research in land

use and transportation modeling (lacono et al., 2008, p. 323). In a broad way these

frameworks can be categorized into three main areas, spatial interaction models,

econometric models and micro-simulation models.

Model types and operational characteristics

Spatial interaction models. Many early land use models employed the theory of

gravity to explain the distribution of people and activities across regions. In 1964 Lowry

was the first to use a gravity model as a basis for explaining the spatial interaction of

regions and their land uses (lacono et al., 2008, p. 325). The main assumption of

gravity models is that the larger a city or region the greater its ability to attract

population, employment and commerce (Koomen et al., 2007, p. 7). A large number of

spatial interaction models have been developed over the past four decades that

continue to use gravity theory. One of the two methods of forecasting future land use

compared in this study uses a spatial interaction model based in gravity theory to

allocate projected population and employment growth. It will be reviewed in more detail

in the Chapter 3.

Econometric models. Models that employ econometric frameworks generally

contain two model types to predict locations of population, employment, commercial

activities and at times transportation flows. These models typically contain land market

and regional economic sub-models models which form the model's foundation. A wide









variety of models have developed using this framework, with many using different

theories to explain spatial distributions and actor behaviors. Random utility theory and

discrete choice theory are two methods used to improve the capabilities of these

models to simulate individual behavior which has resulted in improved measures of

accessibility and travel demand forecasting (lacono et al., 2008, p. 328-331).

Micro-simulation models. As the name infers, micro-simulation models attempt

to model at a much finer resolution than the other model-types. They focus on the level

of the individual and typically include all the actors who influence land use changes.

Although some econometric models incorporate individual behavior they differ from

micro-simulation models in the way they estimate these behaviors. Rather than using a

cross-sectional or average of a population or market to estimate an individual behavior,

micro-simulation models take a bottom-up approach and estimate the behavior of

populations and markets based on the cumulative behavior of individuals (Koomen et

al., 2007, p. 12). Like econometric models, micro-simulation models employ a variety of

theories to explain the location of people, activities and commerce, including random

utility, discrete choice, multi-agent rule-based, and cellular automata approaches to

name a few (lacono et al., 2007, pp. 332-335).

Operational characteristics. Apart from theoretical differences, models can also

differ by their operational characteristics. For example, a model may be characterized

as static, meaning it calculates a set of circumstances at a given point in time; or

dynamic, meaning it contains intermediate time steps that become beginning points for

subsequent occurrences (Koomen et al., 2007, p. 3). The variety of theoretical and

operational characteristics used to describe models often makes it difficult to categorize









them into uniform groups. The three frameworks described above however are a

starting point to understanding land use modeling and provide a wide context within

which the following meta-analysis of literature reviews can be analyzed.

Meta-analysis of selected literature reviews

Four literature reviews on land use models used in transportation modeling

frameworks were selected for analysis to determine whether any trends in the literature

could be observed. The selection of reviews was based on several criteria. As the

United States and Florida in particular are the locations of the study area for this

research, literature reviews that were relevant to both these geographic regions were

selected. To offer an international perspective it was also considered important to

include at least one review from outside the United States. Additionally, it was also

considered important to include reviews that were written from an academic or

theoretical standpoint, as well as reviews that aimed to fulfill practical purposes or

specific project goals, and to focus on most recently published works. Table 2-4 lists

the reviews by region, year, author and title.

Table 2-4. Selected literature reviews for meta-analysis of land use models
Region Year Author(s) Title
Integrated Urban Models for Simulation of Transit and
USA 1998 Miller, Kriger and Hunt Land-Use Policies
Land-Use Policies
0U.S. Environmental Projecting Land-Use Change: A Summary of Models
USA 2000 Pfor Assessing the Effects of Community Growth and
Protection Agency
Change on Land-Use Patterns
FL 2006 Zhao and Chung A study of alternative land use forecasting models
Models of the Relationship Between Transport and
KOREA 2006 Chang
SLand-use: A Review


In the literature models are typically characterized by the different methodologies

they employ. Often models use multiple methodologies which can make categorization









difficult. In this meta-analysis the method used to categorize model types is the same

as that used in the literature review by Zhao and Chung (2006) which reviews the

largest number of models of all the selected reviews. Their approach was to classify

models based upon the methodology most emphasized by each model (Zhao and

Chung, 2006, p.3). A table listing these categories and which models were reviewed by

each author is listed in Appendix A.

Model strengths and weaknesses. In each of the selected literature reviews the

authors assess the strengths and weaknesses of each model type. Appendix B

provides a table summarizing these by author and model type. Over the time period

covered by these literature reviews (1998 to 2006) there appears to be a trend toward

favoring the more dynamic and disaggregate model types capable of a finer spatial

resolution, over the static and aggregate ones which typically have much coarser spatial

resolution. Disaggregate models attempt to simulate the behavior of individuals then

make assumptions about the behavior of large groups such as markets or populations.

Conversely, aggregate models simulate the behavior of markets or populations and

make assumptions about the behavior of individuals. Additionally, a model is

considered to be dynamic if the outputs of an earlier iteration of the model form part of

the inputs of subsequent iterations, in a way that reflects the cyclical nature of

transportation and land use. Static models on the other hand are not cyclical and

require a stepped process to occur if a feedback is required between iterations of the

model.

With the exception of the review by Zhao and Chung (2006) the literature reviews

did not intend to identify a particular model or type that is superior to the rest. In all of









the reviews however strengths and limitations are mentioned for the majority of model

types. When these factors are summarized (Appendix B) a general trend appears that

correlates with a characterization of the aggregate and static model types being limited

by those factors (i.e. by being static and aggregate), and the disaggregate and dynamic

model types as being strengthened by those characteristics (i.e. by being dynamic and

disaggregate). In addition, the majority of models reviewed in the earliest study by

Miller et al. (1998) were static and aggregate (spatial interaction and spatial input/out

models), with only one dynamic and disaggregate model (UrbanSim a micro-simulation

model) being reviewed. The more recent study by Zhao and Chung (2006) reviewed

the most models, including the largest number of dynamic and disaggregate models of

all the literature reviews, finding those to be the least limited of all model types.

Observed trend in the literature. A trend that favors the dynamic, disaggregate

model types over the static, aggregate ones has been recognized by others (Johnston

2004, Waddell 2004, lacono et al. 2008, and Koomen et al. 2007) who have put forth

explanations for this development. The literature suggests that two influential factors

have contributed to this trend, firstly two key federal policies concerning the

environment and transportation, and secondly the emergence of Smart Growth

initiatives. The influence that the CAAA (1990) and the ISTEA (1991) had on the land

use and transportation planning process has already been discussed in an earlier

section, Trends in Planning Approach and Research, so only the topic of Smart Growth

initiatives will discussed here.

Smart Growth Initiatives. Beginning in the mid 1980s with increasing numbers of

states in the US overhauling their growth management legislation in an attempt to









combat sprawling development patterns (Weitz, 1999, p. 268), consensus within and

outside the planning community was building in support of a new approach to land use

planning. Although the interdependent nature of land use and transportation had long

been recognized (Kelly, 1994, p. 129) the idea that transportation investments might be

used as a tool to manage growth in a 'smarter' way was just beginning to be realized

(Knaap, unknown, p. 12). The passage of ISTEA in 1991 led the US EPA to create the

Urban Economic Development Division (UEDD) which formed the Smart Growth

Network, to provide funding to a variety of Smart Growth projects (Knaap, unknown, p.

9).

Of the many principles that characterize Smart Growth development, three stand

out as most relevant to this study. By advocating mixed land uses, a provision of a

variety of transportation choices, and directing development toward existing

communities, Smart Growth initiatives have changed the way in which transportation

planners have traditionally viewed land use and development. Until this time

transportation planners and the land use models they used assumed land use change

resulted in changed transportation demands but that the transportation system itself did

not influence land use change (Borning, 2006, p. 2). These particular Smart Growth

concepts introduced a new level of complexity to travel demand modeling, and in turn

land use change forecasting, that highlighted the spatial, temporal and behavioral

limitations of the earlier models. The trend towards more dynamic and disaggregate

model types identified in this meta-analysis can be seen as an outcome associated with

the rise of Smart Growth initiatives.









Criticisms of complex models. The more complex models favored in the trend

observed by this review are not without some criticisms and drawbacks. Their

complexity often requires expert technical assistance for their implementation and

operation, making them potentially expensive and not accessible to those who may

benefit from them most, in particular metropolitan planning organizations. Additionally,

they often lack transparency, with the underlying assumptions that form critical model

parameters being embedded within the model process in such a way that the non-

expert user may not fully understand the information the model produces. These

challenges however present opportunities for alternative directions in the development

of land use models. One such direction is to pursue a middle ground between the

simple and the complex. Rather than "dumbing-down" the more complex models

simpler models, such as rule-based scenario-analysis models, can be created that

incorporate more sophisticated modeling techniques. Such an approach is investigated

in this study which uses a land use forecasting method based in scenario-analysis and

suitability analysis for the creation of two of the three land use scenarios this study

compares. These two techniques will be reviewed in more detail in the following

sections.

Scenario Planning

Much of what planners do involves planning for the future which invariably requires

some estimation of what the future might look like. As noted in the previous section

transportation planners use models as tools to forecast the performance of

transportation systems and an integral part of that process involves the use of models to

forecast land use change. Also, as reported in previous sections several federal

policies, such as NEPA and CAAA require state and local government agencies to









assess impacts of transportation-related initiatives using forecasts of future urban

growth and transportation demands. Over the past five decades forecasts have come

to play a more prominent role in land use and transportation planning than in previous

eras when plans had a more visionary purpose as opposed to their contemporary focus

of mitigating possible future problems (Wachs, 2001, p. 369). Some are critical of this

changing emphasis, such as Isserman who charged in 1985 that "We make plans as if

the role of planning were simply to accommodate what is forecast and ignore the fact

that planning can affect the future" (p. 485) (as cited in Dalton, 2001, p. 397). More

recently Handy (2008) has also been critical of the emphasis on forecasts because

often they are viewed in the planning process as a prediction of the future rather than

what they really are, just one possible outcome among many (p. 122).

The literature reviewed in the previous section revealed over the past 15 years a

trend in land use modeling that has seen more complex, dynamic and disaggregate

models being favored over simpler modeling techniques. Simpler land use models such

as rule-based or deterministic models, which produce future scenarios based on user

inputted data, preferences and assumptions to reflect a particular hypothetical future,

are often considered to be lacking in theory and sophisticated modeling techniques

(USEPA, 2000; Zhao and Chung, 2006). Conflicting views exist about the strengths

and weaknesses of each approach, leading to the question of which method is better

complex/technical or simple/hypothetical. To present a context for understanding this

argument and provide a background for this research project that compares a scenario-

based forecasting approach with a standard land use forecasting method based in









gravity theory, a review of the literature on scenario planning and its strengths and

weaknesses follows.

History and Planning Applications

A scenario is simply a version of what the future might be based on a particular set

of assumptions (Bartholomew, 2005 p. 5; Myers and Kitsuse, 2000, p. 228). Scenario

planning typically involves the conception of a range of possible scenarios followed by a

process of analysis and evaluation to narrow down possibilities so that an appropriate

course of action can be identified (Bartholomew and Ewing, 2009, p. 14). Scenario

planning was first used in strategic warfare with some dating its origins as far back as

the sixth century (Bartholomew, 2005, p. 4-5). In contemporary times its use was first

popularized for military purposes by the RAND Corporation in the 1950s when they

strategically assessed potential nuclear conflicts in the years following World War II

(Bartholomew, 2005, p. 5). Scenario planning has also been extensively used in

business applications, with the Royal Dutch Shell Company being among the first to use

it in the late 1960s and early 1970s to develop strategic plans that were to enable them

to anticipate and prepare for the global oil crisis of 1973 (Zegras et al., 2004, pp. 2-3).

Scenario techniques have been used in land use planning since the late 1940s

(Kahyaoolu-KoraCin et al., 2009, p. 1022). Since then the passage of NEPA in 1969

saw an increase in scenario techniques being used in transportation planning due to the

requirement that environmental impact statements include alternatives to project plans

(Bartholomew, 2005, p. 7).

In military and business applications of scenario planning the focus is typically

placed on assessing the interrelated causal relationships between external factors, such

as economic conditions and environmental resources and how they limit or enhance









possible strategies under consideration. The approach to transportation planning that

NEPA helped to create however focuses not on how external factors impact a scenario

but largely the reverse. Analysis of internally specified actions are used to create a

scenario that is then assessed based on its impact on external resources and

conditions, for example, how particular changes to the transportation network may

impact congestion. Interaction between internal and external factors, like how changes

to the network might impact land uses, however is largely ignored (Bartholomew, 2005,

p. 7).

Transportation planning has been highly dependent on computer modeling

systems that treat future land use data as an external factor that serves as an input to

the modeling process but remains constant across all scenarios being modeled. By

only considering one possible future land use pattern opportunities to assess the effects

other alternatives might have on the transportation system are lost (Bartholomew, 2005,

p. 7) and a plan is produced that is designed to address the problems forecasted for

only one possible vision of the future (Zegras et al., 2004, p. 4). Recent trends in land

use modeling described in a previous section of this review are largely a response to

this acknowledged shortcoming of the traditional transportation planning approach.

Rather than using scenario planning to produce several future land use patterns for

analysis with transportation models, research and practical emphasis continues to be

given to more complex land use models for the production of a single forecast that

serves as the only land use input into the transportation modeling process

(Bartholomew, 2005, p. 7).









Strengths and Limitations

Apart from modeling criticisms that scenario planning lacks sophisticated

techniques and a strong theoretical basis (EPA, 2000), and poor capabilities to model

complex economic and market processes (Zhao and Chung, 2006), other more

process-oriented criticisms have been identified. Bartholomew (2005) notes a certain

amount of bias may exist when too few scenarios are analyzed. For example, analyzing

only two scenarios often results in them being categorized in stark terms of good and

bad, and similarly three scenarios as low, medium and high risk (p. 14). While each

scenario should be objectively analyzed on its strengths and weaknesses when too few

are compared it is easier for simple categorizations like good and bad to be made.

The consideration of too many scenarios can also be problematic because of the

excessive amounts of information often needed for analysis. This can cause particular

difficulties when scenario analysis is linked to other modeling frameworks such as

transportation demand models which consume large amounts of data and time for

analysis (Zegras et al., 2004, p. 10). Difficulties in linking scenario analysis to other

quantitative tools also make it difficult to empirically evaluate impacts and compare

them across scenarios (Zegras et al., 2004, p. 11).

Scenario planning often takes place in a collaborative environment with many

stakeholders involved in the process, leading to claims that it is too subjective and

idealistic to reflect the real world. Similarly the idealistic nature of the process may lead

to the selection of scenarios that may not be possible to implement in the real world

(Myers and Kitsuse, 2000, p. 222). Some argue however that it is this idealistic nature

that gives scenario planning an advantage over technically generated forecasts

because it lends itself better to capturing macro-trends that can be overlooked when









analysis is too narrowly focused to meet more precise modeling parameters (Zegras et

al., 2004, p. 3).

Many of the strengths that scenario planning has to offer to the coordination of

land use and transportation are drawn from the weaknesses of the traditional

transportation planning approach that has favored the use of complex and technical

models to forecast future land use change. All forecasts of future land use change are

based on assumptions that are inherently subjective because the future cannot be

known. While simpler approaches to modeling change, such as scenario planning are

often criticized for being too subjective, excessive weight is often given to forecasts

derived from complex and technical computer models because they have the illusion of

being scientifically objective. With regard to transportation planning and demand

models in particular, Handy (2008) points out that not only do models produce forecasts

based on assumptions they also use forecasts that were also based on assumption all

of which are subjective in nature (p. 122). She cites as examples the use of K-factors in

gravity models to achieve a best fit with observed trip distribution data and which modes

are included in the mode choice step in travel demand models. Both examples embody

a subjectivity that is often less than apparent to non-technicians involved in the planning

process, often the public and their representatives (p. 122).

The creation of forecasts of land use change using models based on historical

economic trends and short-range planning policies, such as future land use elements of

comprehensive plans with a horizon of ten-years, can lead to narrow views of possible

futures. Myers and Kitsuse (2000) note that political and economic pressures on

planners are often translated into plans that emphasize short-range planning goals at









the expense of more visionary long-range goals (p. 222). The use of forecasts such as

these is particularly problematic in transportation planning. Land use change occurs

slowly and incrementally so that policies targeting major changes to development

patterns may take many years to become apparent as trends. When forecasts are

based on inputs such as short-range comprehensive plans, long-range future trends

may not be adequately captured. The result of using forecasts derived in this way can

be the implementation of transportation projects that are designed for short rather than

long-range future land use patterns. This can be problematic because history has

shown that transportation projects have a strong tendency to influence urban form once

they are built. However, because land use change typically happens at a slower pace

there is the danger that undesirable short-range development patterns will be

unintentionally continued by using forecasts based on short-range future land use

trends represented in comprehensive plans which are most typically used in traditional

approaches to transportation planning. Scenario planning may be better placed to

capture wide macro-trends missed by statistically derived forecasts due to its more

transparent and collaborative process of incorporating many points of view and

encouraging visions of the future that differ greatly from the present.

Suggestions have been made that transportation planning would benefit from the

use of forecasts derived using both scenario planning and more sophisticated land use

modeling techniques (Lemp et al., 2008; Myers and Kitsuse, 2000; Wachs, 2001;

Zegras, et al. 2004). This study uses a forecasting method that employs both scenario

analysis and land use suitability analysis to assess the impact different methods of

forecasting land use change have on the attainment of long-range planning goals. The









following section discusses the use of suitability analysis in understanding how future

land use change may occur.

Land Use Suitability Analysis

Overview

As the name implies land use suitability analysis is a process of identifying the

most appropriate location and distribution of future land uses (Collins, 2001, p. 611;

Malczewski, 2003, p. 4). Various layers of spatially referenced information about

selected characteristics of a study area can be analyzed collectively when overlain

together, providing a cumulative assessment of a location's attributes that is useful in

making decisions about its future use. A simple example would be combining

information regarding soils, slope, and hydrology to appropriately locate a building site

and decide if that is the most appropriate location among other potential sites. Land

use suitability analysis is used in a wide variety of settings, from fields such as ecology

for defining habitat suitability, agriculture for crop suitability, landscape architecture for

site analysis, and environmental sciences where impact assessments may be required.

(Malczewski, 2003, p. 4). Its applicability to planning related fields stems from its

strength as a tool to aid decision makers when formulating policies about site-specific

development proposals, their environmental impacts and their potential cumulative

effects over time (Collins et al., 2001, p. 611).

The use of computer models to perform land use suitability analysis has become

more common in the past two decades as off-the-shelf geographic information systems

(GIS) software programs have become more readily available, affordable and user-

friendly. The first use of computers can be traced back to the early 1960s although the

history of land use suitability analysis dates back to a much earlier time. The use of









sun-prints, hand drawn overlays or maps held up to a window, by early twentieth

century landscape architects Olmstead and Eliot are credited with being some of the

first known uses of the technique (Collins et al., 2001, p. 612). Perhaps the most well-

known early advocate of land use suitability analysis however was the landscape

architect lan McHarg who in the 1960s was responsible for a significant methodological

advancement. McHarg introduced a weighting system to his suitability analysis

whereby he would shade individual transparent overlays of a study area, each

representing a unique site characteristic, using lighter and darker colors to denote

increasing or decreasing suitability. When viewed collectively the degree to which a

location was more or less suitable could be determined by the cumulative shade of color

at that location (McHarg, 1992, pp. 31-42).

The introduction of GIS to suitability analysis was to have dramatic effects on the

speed and accuracy of analyses. While McHarg's earlier analysis was performed using

vector data, that is represented as unique points, lines or polygons, the development of

raster-based GIS programs enabled not only a greater number of data layers to be

analyzed but increased computational speed and spatial accuracy (Collins et al., 2001,

pp. 613-614). Raster data is represented using a grid or raster with each individual grid-

cell containing a unique value representing some attribute in the real world. Figure 2-4

shows the concept of raster analysis and how raster data can be combined to create

new aggregate datasets.

Similar to McHarg's use of lighter or darker shading to determine greater or lesser

suitability, raster analysis combines grid-cells to determine values of varying levels of

suitability. Using the example shown in Figure 2-4, the following analysis could be









undertaken for a study area represented by the raster extent shown to determine the

most suitable location for a new school based on two site characteristics, distance to

major roads where closer distances are less suitable, and distance to existing

residential areas where closer distances are more suitable. When the two input rasters

are combined the analysis shows the most suitable location for a school would be

located in the lower left grid-cell of the output raster with a value of 6.


Input Rastel



1 2 3

/231
Input Raster
1 1 /1

2 2 2

3 3 3 3
Output Ras

3 4 2

3 4 5
i ^


6 /


r 1







2







ster
7


Distance to Major Roads
< 0.5 miles = value 1 = low
0.5 to 1 miles = value 2 = medium
> 1 miles = value 3 = high



Distance to Residential Areas
> 1 miles value1 =low
0.5 to 1 miles = value 2 = medium
< 0.5 miles = value 3 = high


/ 4


Figure 2-4. Example of simple raster analysis. Note. Diagram courtesy of author.


Decisions on how to assign suitability values must be made at two different levels

during the analysis. Firstly each dataset in the analysis must have a range of suitability

values assigned to it, and secondly each dataset must be weighted according to its


/









relative importance in the overall analysis. In the example shown in Figure 2-4 an equal

importance or weight is given to each dataset in the analysis, however this may not be

so in a process involving several decision makers where each has a differing viewpoint

on the relative importance of the datasets. A typical procedure to overcome this is to

assign an agreed upon weight to each dataset so their relative importance is captured

during the analysis. Several methods exist for developing a weighting system from

using a Delphi method where group consensus among decision makers is tallied using

a voting system (Collins et al., 2001, p. 614), or simply by the use of expert advice (Carr

and Zwick, 2007, p. 59-60). Another alternative is pair-wise comparison where each

dataset is paired with another and the relative preference of one over the other is

assigned by each decision maker until all possible combinations have been compared.

Values are then normalized and averaged to determine a weight for each dataset (Carr

and Zwick, 2007, pp. 62-66). These techniques can also be used in decisions regarding

the assignment of value ranges to individual dataset.

One of the methods of forecasting future land use employed in this study uses a

form of land use suitability analysis that was designed as a suite of analytical models to

determine where potential conflicts may occur between competing land uses.

Developed by Margaret Carr and Paul Zwick from the University of Florida during the

2000s, the process known as LUCIS, or Land Use Conflict Identification Strategy will be

described in more detail in the remainder of this section.

Land Use Conflict Identification Strategy (LUCIS)

GIS is a powerful tool planners use to better understand the incremental land use

decisions that cumulatively affect humans and their natural environment. LUCIS was

developed with the main purpose of highlighting where potential conflicts may occur in









the future between competing uses of the land (Carr and Zwick, 2007, p. 7). LUCIS

uses the Environmental Systems Research Institute's software ArcGIS to perform

raster-based land use suitability analysis (Carr and Zwick, 2005, p. 60) generally

following the method described in the preceding paragraphs. It also uses role playing

techniques where major stakeholders act as teams of experts who are charged with

rating all available land according to relative suitabilities that best support a specific land

use they are advocates for (Carr and Zwick, 2005, p. 60). LUCIS uses three

generalized land use categories that groups of stakeholders represent; urban,

agriculture and conservation. Once each group of stakeholders rates the available

lands according to their particular preference, the three results are compared to see

where potential conflicts may occur. More specifically LUCIS follows a six step process

which is represented in Table 2-5.

Table 2-5. The six step LUCIS process.
Step Description
1 Define goals and objectives that become the criteria for determining
suitability
2 Inventory data resources relevant to each goal and objective
3 Analyze data to determine a relative suitability for each goal
4 Combine the relative suitabilities of each goal to determine preference
5 Normalize and collapse the preferences for each land use into three
categories high, medium and low
6 Compare the ranges of land use preference to determine likely areas of
conflict
Source. "Using GIS suitability analysis to identify potential future land use conflicts in North
Central Florida" by Carr and Zwick, 2005, p. 60.

The development of goals and objectives is an important part of the LUCIS

process, as it is with many other goal-driven activities such as the development of

comprehensive plans by local governments. Not only do goals and objectives give









direction to plans and processes they are also useful for the creation of alternative plans

or scenarios that can be measured and compared (Carr and Zwick, 2007, p. 84).

LUCIS uses a hierarchical set of goals, objectives and sub-objectives to drive the

analytical process, with a unique set of goals created for each of the three land use

categories. Table 2-6 shows an example of a subset of the goals, objectives and sub-

objectives that could be used for the agriculture category.

Table 2-6. Example of a subset of goals, objectives and sub-objectives for the
agriculture category
Agriculture Goals
Goal 1 Row Crops Identify the suitability for row crops
2 Livestock Identify the suitability for livestock
3 Specialty Farms Identify the suitability for specialty farms
4 Nurseries Identify the suitability for nurseries
5 Timber Production Identify the suitability for timber production
Goal 1 Row Crops Identify the suitability for row crops
Objective 1.1 Physical suitability
Sub-objective 1.1.1 Identify soils suitable for row crops
1.1.2 Identify land uses suitable for row crops

Objective 1.2 Economic suitability
Sub-objective 1.2.1 Identify land values suitable for row crops
1.2.2 Identify lands proximal to markets
1.2.3 Identify lands proximal to major roads
Source. Adapted from "Smart Land Use Analysis: The LUCIS Model" by Carr and Zwick, 2007,
pp. 86-87.

Once goals and objectives have been identified and an inventory of data to be

used is collected, suitability analysis is performed for all of the goals (Carr and Zwick,

2005, p. 62). The next step involves assigning preference to the suitability rasters,

meaning each group of stakeholders must rate the individual goals according to their

relative importance. LUCIS uses the technique of pair-wise comparison, described in

preceding paragraphs to assign weights to the different goals within each land use

category (Carr and Zwick, 2007, p. 63). For example, using the agriculture goals shown









in Table 2-6, row crops may be the most valued type of agriculture so it would be

assigned a higher weight than livestock which may be less valued by the agriculture

stakeholders. The end result of this step is the creation of three separate rasters, one

each for agriculture, urban and conservation, with grid-cell values that represent the

individual preference stakeholders assigned to them. So that the three rasters may be

compared each one is normalized and their values reclassified into three intervals

reflecting, high, medium and low suitability. The final step of LUCIS, represented in

Figure 2-5, involves the combination of the three preference grids so that each grid-cell

in the output raster has a unique value that represents the level of conflict between land

uses in that location

This information can be useful in a wide variety of ways to both planning

professionals and advocates for special interest groups, such as those supporting

environmental conservation or agricultural preservation. By knowing the location and

type of conflict that is likely to occur groups can more specifically target their efforts to

preserve, conserve or change land uses in the future (Carr and Zwick, 2007, p. 164).

For urban planners, LUCIS provides information that can be used at varying levels of

complexity. At a broad level LUCIS can help to calculate whether appropriate levels of

undisputed land uses exist to accommodate projected future growth. For example, if

population growth is projected to increase by 200,000 people in the next 10 years, then

using the existing gross urban population density the additional acres needed to

accommodate growth at the existing rate can be calculated. If that acreage is more

than what LUCIS indicates is available in the undisputed urban category (where urban

wins over agriculture and conservation) then urban land uses will likely encroach others











and result in conflict. This knowledge is invaluable to planners as the formulate policies


aiming to manage growth to minimize the loss of agricultural land and preserve future


areas for conservation.


High Preference = 3 132 233/ 121/
Medium Preference = 2
Low Preference = 1 332 122 213
Raster Value Type of Conflict Level
132 Con Wins None 333/ 222 111 Conflict Raster
233 Con-Urb High
121 Con Wins None 3 3 3
332 Ag-Con High
122 Con-Urb Medium A on Urb
Ag Con Urb
213 ....UrbanWins None Preference
333 Major High
222 -Major Medium
111 Major Low
Major Conflict




Figure 2-5. Combining stakeholder preferences to show where conflict occurs. Note.
Diagram courtesy of author.



A more complex use of LUCIS is as a tool for generating future land use scenarios


that form the basis of a scenario planning process for identifying the impacts of different


urban growth patterns. This study uses LUCIS as the basis of a method for forecasting


future land use scenarios that will subsequently be compared to a forecast generated


using a different method. Projected population and employment growth for the period


2007 to 2025 will be allocated based on levels of land use conflict identified using









LUCIS for Lake County, Florida. A detailed description of the allocation method follows

in Chapter Three.

Summary

The relationship between land use and transportation in the United States is

longstanding and complex. While history has shown that innovations in transportation

technology have enabled increasing levels of mobility for most Americans, their desire

to live in low density suburban locations has simultaneously driven their need for greater

mobility. At the centre of the land use and transportation relationship is the automobile

and the high level of dependency most Americans have developed for it.

Public concerns about the negative social and environmental impacts of

transportation and the urban growth it has fostered have led to government policies

aimed at controlling and mitigating these effects. Research into the relationship

between land use and transportation has been largely driven by these concerns.

Despite a trend of increasingly focused research into understanding this relationship,

there still remains a considerable gap in the understanding of the relationship's

dynamics. A coordinated approach to planning land use and transportation may offer

the greatest opportunity for positive change yet a best approach to achieve coordination

eludes both practitioners and researchers.

Research into the relationship between land use and transportation helps guide

decision makers in formulating policies aimed at minimizing negative effects of urban

growth. The use of transportation and land use models plays an important role in

understanding potential impacts of future changes to both the transportation system and

urban growth patterns. The quantitative demands required of these models are

predominantly a consequence of federal policies that more closely monitor the









environmental impacts of transportation investments and their associated increases in

automobile use. As more emphasis has been placed on the quantification and

forecasting of potential future impacts, land use and transportation models have

become increasingly more sophisticated and complex, relying heavily on computer

simulations based on a large number of assumed variables. Based in mathematical

and economic theories, complex models often give the impression of being scientifically

superior to less complex or rule-based models. As a result, over time forecasts

generated using more complex methods have become the basis from which future long-

range plans are developed, rather than being used as an assessment of outcomes of

proposed plans. Instead of planning for a desired vision of the future, emphasis is

centered on planning for a future forecasted on assumptions about variables that cannot

be known.

Emphasis given in the urban planning process to complex models and their

resulting forecasts has come at the expense of simpler models that offer equally

possible forecasts of the future. The use of other methods of forecasting such as

scenario planning techniques offer an opportunity to capture broader macro-trends that

more statistically derived forecasts are unable to capture due to their narrower focus on

micro-level simulations. When supplemented with more sophisticated approaches to

forecasting future land use change, such as land use suitability analysis, simpler

methods such as scenario planning may offer better opportunities to successfully

coordinate long-range land use and transportation plans than the more complex

mathematically driven models in common practice today. That hypothesis is tested in

this study which compares two methods of forecasting future land use change. One









method incorporates scenario planning techniques and suitability analysis to create a

vision based on a set of broadly defined mutually supportive policy goals. The other

method is mathematically derived used a gravity model and is based on historical

development trends and prescriptive land use guidelines from existing policy. The

following chapter details the methodology used to carry out this comparison.









CHAPTER 3
STUDY AREA AND TIME FRAME

Study Area Selection

Florida has a history of rapid population growth. While the national population

growth rate for the USA in the twentieth century was between 10 and 20 percent, the

state of Florida experienced rates of between 20 and 80 percent (Smith, 2005, p. 2).

The distribution of Florida's population has shifted geographically over time, with rapid

increases in the state's central and southern regions. Although the southeastern and

southwestern regions of Florida have experienced the most rapid rate of growth in the

twentieth century, with decade averages of 77% and 61% respectively (Table 3-1), the

central region has contained the largest proportion of the state's population since the

1960s (Figure 3-1). This trend is projected to continue. The central region is expected

to grow the most in terms of absolute population between 2010 and 2035, with 45% of

new Floridians predicted to reside in the central region (Table 3-2).

Table 3-1. Florida's population growth rates by region percentage change per
decade, 1900 to 2000
Region Northern Central Southeastern Southwestern
Years Increase (%) Increase (%) Increase (%) Increase (%)
1900 to 1910 31.1 60.6 70.0 89.9
1910 to 1920 13.6 36.9 120.2 79.0
1920 to 1930 16.5 74.6 171.3 68.3
1930 to 1940 19.3 23.1 74.6 14.6
1940 to 1950 31.5 43.4 79.3 34.1
1950 to 1960 36.9 89.7 113.5 105.5
1960 to 1970 18.4 37.0 48.4 61.6
1970 to 1980 24.9 48.3 44.5 75.8
1980 to 1990 23.8 38.9 26.5 52.0
1990 to 2000 21.6 22.9 23.1 31.2
Average Increase 23.8 47.5 77.1 61.2
Note. Adapted from "Florida population growth: Past, present and future" by S.K. Smith, 2005,
p. 23.


















1900




Central ~


Southwestern

Southeastern -


1920


Central


Southwestern


Southeastern
40


- .ot 26%h
NIorthern \'


Central


Southwestern 1

Southeastern -i
low*''


- or 20%
Northern r


Central


Southwestern 1W

Southeastern '-
^"'


2000
%----- 18%
Northern ,4

Central


Southwestern

Southeastern k<4
.-'


Percentage of Total Population


Highest


Lowest


0 100 200 Miles


Figure 3-1. Florida's population distribution by region, 1900 to 2000. Adapted from
"Florida population growth: Past, present and future" by S.K. Smith, 2005, p.
22.


1940









Table 3-2. Florida's projected population increase (per thousand people) by region,
2010 to 2035
Increase Total Growth
Region 2010 2035
2010 to 2035 2010 to 2035
Central 7,620 11,735 4,115 45%
Northern 3,833 5,820 1,988 22%
Southeastern 5,823 7,398 1,574 17%
Southwestern 1,987 3,436 1,449 16%
Note. Based on an average of medium and high projections from "Florida population studies
bulletin 153 Florida county population projections, 2008 2035" by University of Florida
Bureau of Economic and Business Research, & University of Florida Population Program, 2009.


Within the central region Orange County is expected to grow by the largest

amount for the period 2010 to 2035, increasing by approximately 736,000 people and

accounting for almost 18% of total growth in the region (Figure 3-2). Figure 3-2 shows

adjacent counties, such as Osceola, Polk, Brevard and Lake, will also experience

substantial population increases.


















































Population Growth per County & Percentage of Regional Growth


Lowest Highest


- -0 10 20 M
0 10 20 Mies t


Figure 3-2. Florida's central region, projected population growth from 2010 to 2035.
Note. Based on an average of medium and high projections from "Florida
population studies bulletin 153 Florida county population projections, 2008 -
2035" by University of Florida Bureau of Economic and Business Research, &
University of Florida Population Program, 2009.


A comparison of projected gross population densities for counties in the central


region provides insight into the effect population growth will have in specific locations in


this region. For example, although Figure 3-2 shows Orange County will have the


largest increase in absolute numbers of new residents between 2010 and 2035, its


gross population density will only increase by 64% (Table 3-3). Conversely, Osceola


County will have only approximately half the amount of growth as Orange County, and


LAKE
252200
615


IHERNANDO
130100
3 2%
PASCO
1 400
7%
F iNELI !
'. 11
2 2.



5


e3EVARD
OSCEOLA 24400
336850 6%
POLK : 82%
322400
7.8%
INDIANRIVET
23%

STLIJCIE
80ji


.'OLIJUCI t
204 .









Lake County approximately one-third, yet their respective increases in gross population

densities are 117% and 84%. The differences in densities are even more pronounced

when viewed for the period 2000 to 2035, with Lake County increasing by 163% and

Osceola County by 263% (Table 3-3). This is a result of both Lake and Osceola

Counties having relatively lower existing populations than neighboring Orange County

and explains how a relatively low amount of population growth can still have dramatic

effects in terms of changing population densities.

Table 3-3. Florida's central region, gross population densities and increases from 2010
to 2035.
Gross Density Increase Increase Increase
(people per acre) (%) (%) (%)
2000 to 2010 to 2000 to
County 2000 2010 2035
2010 2035 2035
Sumter 0.15 0.25 0.68 90 139 353
Osceola 0.21 0.35 0.75 67 117 263
St. Lucie 0.54 1.21 1.71 125 42 218
Lake 0.33 0.47 0.86 43 84 163
Hernando 0.42 0.57 0.99 33 75 133
Marion 0.25 0.33 0.56 31 69 122
Indian River 0.36 0.47 0.78 29 66 115
Pasco 0.73 0.95 1.56 30 65 114
Orange 1.55 1.99 3.26 28 64 110
Citrus 0.30 0.38 0.59 25 56 95
Levy 0.05 0.06 0.09 24 57 94
Polk 0.41 0.50 0.78 24 54 90
Hillsborough 1.49 1.83 2.79 23 53 88
Brevard 0.73 0.88 1.26 19 43 71
Volusia 0.62 0.73 1.01 17 39 63
Okeechobee 0.07 0.08 0.11 14 33 52
Pinellas 5.30 5.43 5.95 2 10 12
Seminole 1.84 0.76 1.19 -59 58 -35
Note. Based on an average of medium and high projections from "Florida population studies
bulletin 153 Florida county population projections, 2008 2035" by University of Florida
Bureau of Economic and Business Research, & University of Florida Population Program, 2009.
Gross density was calculated as gross area (land area less open water) divided by projected
population.

The significant increases in gross densities for Lake and Osceola Counties may be

a "spill-over" effect from the higher density Orange County where employment

opportunities may be greater yet housing opportunities less affordable to specific groups









of workers. Census Bureau data regarding travel patterns for journeys to work between

these counties support this theory. Table 3-4 shows the residence and work counties

for workers in Lake, Osceola and Orange Counties for 1990 and 2000. In particular it

shows the number of workers travelling to Orange County from its adjacent counties,

and from Lake and Osceola Counties to their adjacent counties. Although most workers

live and work in the same county, a large number of Lake and Osceola residents travel

to Orange County for work, with Lake County experiencing the greatest relative

increase of 135% between 1990 and 2000. In 2000 25% of workers in Lake County and

45% of workers in Osceola County traveled to Orange County to work.

Table 3-4. Residence and work counties for workers in Lake, Osceola and Orange
Counties for 1990 and 2000


Residence
County
Orange
Osceola
Lake
Seminole
Polk
Brevard
Lake
Lake
Lake
Lake
Lake
Lake
Lake


Work


County
Orange
Orange
Orange
Orange
Orange
Orange
Lake
Orange
Seminole
Volusia
Sumter
Osceola
Marion


Change
1990 2000 1990 to


317,493
19,495
8,516
70,609
5,496
2,953
42,777
8,516
1,261
1,406
510
457
658


376,709
34,085
20,009
80,875
11,823
6,122
51,842
20,009
2,979
1,536
1,214
1,110
629


2000
59,216
14,590
11,493
10,266
6,327
3,169
9,065
11,493
1,718
130
704
653
-29


% Change
1990 to 2000
19
75
135
15
115
107
21
135
136
9
138
143
-4


Osceola Osceola 29,323 38,416 9,093 31
Osceola Orange 19,495 34,085 14,590 75
Osceola Lake 66 1,628 1,562 2367
Osceola Polk 362 560 198 55
Osceola Brevard 364 330 -34 -9
Osceola Indian River 49 35 -14 -29
Osceola Okeechobee 5 24 19 380


data 2000 County to


Note. Adapted from "Journey to work and place of work Census 2000
county worker flow files" by United States Census Bureau, 2008.









Further support for the theory that Orange County's growth is spilling over into

surrounding lower density counties can be found in housing data for 2000 and 2008

from the United States Census Bureau. Table 3-5 shows the median value of owner-

occupied houses for Lake, Osceola and Orange County. Median values for both 2000

and 2008 were highest in Orange County and lowest in Lake County, with rates of

increase similarly highest in Orange and lowest in Lake County. The median value of

homes in Osceola was 8% less than in Orange County in 2000 and by 2008 homes

were more affordable in Lake County where their median value was 23% less than

those in Orange County. Although housing affordability is only one potential factor

influencing people's decisions of where to reside in relation to their employment, should

this trend continue it may result in an increase in the number of Lake and Osceola

County residents traveling to Orange County to work in the future.

Table 3-5. Median values of owner-occupied houses in 2000 and 2008
Median Value ($) Median Value ($)
County 2000 2008
Orange 107,500 243,400
Osceola 99,300 207,500
Lake 100,600 187,800
Note. Adapted from "2008 American Community Surveyl-Year Estimates" by United States
Census Bureau, 2008, and from "Profile of Selected Housing Characteristics: 2000" by United
States Census Bureau, 2000.


An increasing number of residents traveling into Orange County to work will have

important implications for long-range transportation planning in adjacent Lake and

Osceola Counties. In Florida long-range transportation plans (LRTPs) are prepared at a

regional scale by agencies known either as a metropolitan planning organization (MPO)

or transportation planning organization (TPO). These organizations are responsible for

the creation of 20-year transportation plans which are updated every five years that









include strategies consistent with local and state planning objectives for the

municipalities within their region (Florida Department of Transportation, 2009, p. 4-2).

Osceola and Orange Counties are within the same MPO boundary, Metroplan

Orlando, meaning the transportation needs for these counties are planned jointly under

the same LRTP. Transportation planning for the projected needs of Lake County

however is addressed by the Lake-Sumter MPO (LSMPO) under their own LRTP. How

well these two MPOs coordinate their long-range plans is particularly relevant to the

enhancement of public transportation in Lake County which seeks to increase the

proportion of choice riders, or those traveling to work. With one in four residents of

Lake County traveling to Orange County to work, the need to coordinate transit plans is

especially important if increased ridership is to be achieved.

A MPOs long-range transportation plan is statutorily required to be consistent with

the future land use element and policies of the comprehensive plans of the local

governments within its jurisdiction (FDOT, 2009a, p. 4-14). In the case of Lake County

and the LSMPO, its LRTP must be consistent with one county and fourteen municipal

comprehensive plans. This represents a large number of plans to be consistent with

considering the relatively low number of people the LSMPO serves, around 211,000 in

2000 (UFBEBR, 2009). Orange, Seminole and Osceola Counties which fall under

Metroplan Orlando's LRTP, on the other hand, had a population of around 1.4 million in

2000 but only 26 comprehensive plans with which to coordinate. Figure 3-3 shows

municipality, MPO and TPO boundaries for the central Florida region. Planning

consistently across many jurisdictional boundaries may be particularly challenging for

the LSMPO if the number of residents traveling to Orange County is isolated to just a










few local governments rather than being spread more evenly across the region. Local

governments with few potential choice riders may have less interest in creating future

land use plans that support a future transit plan aimed specifically at increasing this type

of ridership.


Municipalities Brevard MPO Ocala-Marion County TPO
Osceola County Municipalities Lake-Sumter MPO Polk County TPO
I Lake County Municipalities Metroplan Orlando Volusia County MPO

0 6 12 Mles


Figure 3-3. Municipality, MPO and TPO boundaries for Florida's central region. Note.
Map courtesy of author.




Lake County was chosen as the area for this study for several reasons. Firstly the

county is situated in the most populous region of the state and is projected to receive


POLK
Q1B


A1 M









the largest number of new residents over the next 25 years. Secondly, Lake County's

adjacency to rapidly growing Orange County creates opportunities for population growth

to "spill over" into Lake County, creating challenges for the coordination of long-range

transportation plans both within Lake County and with Orange County. And thirdly,

Lake County has a relatively high number of local governments within its MPO region,

which may increase the potential for uncoordinated urban growth and reduce the ability

of the LSMPO to reach some of its long-range planning goals, particularly those

concerning public transportation. These factors highlight a long-range planning

situation in Lake County that could benefit from the use of scenario analysis in the

formulation and coordination of its land use and transportation plans.

Study Time Frame

The time frame for this study was the period 2007 to 2025. This was determined

according the time period for which the data used was relevant and for the planning

horizons covered by the policy documents used in the study. It was necessary to select

a point in time from which population and employment projections would be calculated

for the scenarios to be compared in the study. The year 2007 was chosen as the

beginning of the time period because it was the year for which the most current property

parcel data was available for Lake County.

The end-year of 2025 was chosen based on several factors. Firstly, of the two

methods of forecasting land use change that are compared in this study, one is that

which was adopted in the LSMPO's current LRTP which covers the period 2005 to

2025. Secondly, the transportation demand model (TDM) used in the study is that used

by FDOT in their state-wide modeling of transportation demand. The model can be run

for selected future years of which 2025 is one. Thirdly, at the commencement of this









study the most current comprehensive plan and therefore future land use map for Lake

County was also for the period 2005 to 2025. And lastly, projected population and

employment data for all of Florida's counties was available from the University of

Florida's Bureau of Economic and Business Research (BEBR) for several years

including 2025. Since two of the main policy documents guiding this study describe

plans up to the year 2025 and that a TDM specific to that year which uses data created

by one of the methods under investigation, 2025 was deemed most appropriate year to

end the study period.









CHAPTER 4
METHODOLOGY

Methodology Overview

This study uses land use and transportation modeling techniques to compare

three future urban growth scenarios for Lake County, FL in the year 2025. Two different

methods are used to allocate projected population and employment growth for the

period 2007 to 2025 to specific locations within Lake County. One growth scenario

uses the method adopted by the Lake-Sumter Metropolitan Planning Organization

(LSMPO) in the preparation of its 2025 Long Range Transportation Plan (LRTP). Two

alternative growth scenarios use a method that combines the use of the Land Use

Conflict Identification Strategy (LUCIS) and land use suitability analysis, and

concentrate urban growth at varying densities around future transit routes. This study

determines which of the three scenarios best reflects two long-range planning goals

Lake County seeks to achieve; specifically a more compact urban form, and increased

energy conservation by reducing single-occupancy vehicle (SOV) dependency and

increasing transit ridership. The findings of this study demonstrate which of the two

methods results in a future land use pattern that was most supportive of the specified

long range planning goals. By highlighting how different methods of forecasting future

land use effect the achievement of long-range planning goals, this study will alert state,

regional, and local government planning agencies of the benefits alternative methods

can have in improving the coordination of land use and transportation plans.

Figure 4-1 is a diagram explaining the methodology used in this study that will be

discussed in detail in the following sections of this chapter.












FLUAM Scenario

Projected Population Land Use Data
STEP & Employment Data Future Land Use Element of
2025 +. Lake County's
S People and Jobs Adopted Comprehensive
(Census, BEBR) Plan
- ------ ^f -- -- -- --
Allocation Method
Future Land Use
Allocation Method
(FLUAM)
STEP 4
2 FLUAM Scenario
Pop & employ allocated
to parcels of similar type
(land use & land value)
at historical densities

Transit System
A Tr
STEP As exits in FDOT's Using
3Central Florida Model
Regional Planning miles
Model


LUCIS-plus Scenarios

Projected Population
Land Use Data & Employment Data
LUCIS suitability + 2025 People and Jobs
and conflict rasters (Census. BEBR, TAZ,
ECFRPC-REMI)
----- -------- -- -
Allocation Method
LUCIS-planning
land use scenarios
(LUCIS-plus)


LUCIS Low Scenario LUCIS High Scenario
Pop & employ allocated using Pop & employ allocated using
LUCIS & suitability analysis w/ LUCIS & suitability analysis
S 50% increase in densities in w/100% increase in densities in
S mile of future transit stations 'I mile of future transit stations


ansportation Demand Modeling
SFDOT's Central Florida Regional Planning
versionn 4.5 and CUBE software model vehicle-
traveled and mode choice for each scenario


FLUAM Scenario
Transportation Outputs
Vehicle-miles traveled?
Highway and transit trips?


FLUAM Scenario
Land Use Outputs
Percentage of growth contained within
3-mile buffer of future transit?


LUCIS Low Scenario
Transportation Outputs
Vehicle-miles traveled?
Highway and Iransil trips?

+
LUCIS Low Scenario
Land Use Outputs
Percentage of gro ltn contained within
3-mile buffer of future transit?


LUCIS High Scenario
Transportation Outputs
Vehicle-miles traveled?
Highway and transit trips?


LUCIS High Scenario
Land Use Outputs
Percentage of growth contained within
3.mile buffer of future transit"


Figure 4-1. Diagram of study methodology.


STEP
1






STEP
2






STEP
3


COMP PLAN Objective 1-10 Discourage urban sprawl through a future land use pattern which promotes orderly, compact development
LRTP Goal 4 Reduce dependency on single occupant vehicles, increase transit ridership, and increase energy conservation

Which Scenario best reflects these long-range goals and objectives?


4l











Step One

The diagram of the study methodology shown in Figure 4-1 is a generalized

representation of the overall process and is divided into two halves. One half refers to

the creation of a scenario using the Florida land use allocation method (FLUAM) which

is the forecasting method adopted by the LSMPO in the creation of their 2025 LRTP.

The other half of the diagram refers to the creation of two scenarios using the LUCIS-

plus allocation method. LUCIS-plus (LUCIS-planning land use scenarios) is the name

tentatively given to a population and employment allocation method currently under

development in the Department of Urban and Regional Planning at the University of

Florida by Dr. Paul Zwick (P. Zwick, personal communication, March 19, 2010). The

methodology is further divided into three steps, of which this section discusses the first.

Florida Land Use Allocation Method Scenario Data

It should be noted that the FLUAM scenario was not created in this study. It is an

accompanying dataset to the CFRPM version 4.5 provided by the FDOT District 5 for

use in this study. The 2025 land use and socio-economic data for Lake County used

in the CFRPM was supplied to FDOT by the LSMPO (J. Banet, FDOT, personal

communication, February 9, 2010). Chapter Five of the Lake-Sumter 2025 LRTP

describes the methodology used to forecast future land use and socio-economic data

for their 2025 plan (Lake-Sumter Metropolitan Planning Organization, 2006, p. 5-1).

Datasets used in the forecast are included in Chapter 5 of the LRTP (pp. 5-2 5-3) and

are listed as follows:

University of Florida BEBR population projections average of medium

and high projections for 2025









Vacant parcels from the year 2000;

The Future Land Use Element from the Lake County Adopted

Comprehensive Plan;

Traffic analysis zones used in the CFRPM;

Locations of master planned unit developments (PUDs) and developments

of regional impact (DRIs);

Environmentally sensitive land, such as the Green Swamp, Wekiva Basin,

and the Ocala National Forest;

Lakes and rivers; and

Other lands that are not able to be developed.

Land Use Conflict Identification Strategy Planning Land Use Scenarios Data

The LUCIS-plus scenarios were created using similar types of data as the FLUAM

scenario such as population and employment projections, parcel and traffic analysis

data, and environmental data. A comprehensive list of data used in the creation of the

LUCIS-plus scenarios is shown in Appendix C. A general list of data used is as follows:

Population and household estimates for 2007;

TAZ data for the years 2000 and 2012;

Average of medium and high population projections for 2025;

Employment projections for 2005, 2010 and 2025;

Parcel data for 2007;

Lakes and ponds;

Conservation areas;

LUCIS suitability and conflict rasters;









Visionary transit lines and stops for 2035; and

Projected school enrolment data for 2025.

The list shown in Appendix C shows the tabular and GIS data used, however other

types of information were also used in the construction of the LUCIS-plus scenarios,

such as policy guidelines. As these additional sources of information are used in the

methodology they will be discussed at that point in more detail.

Step Two

The second step of the methodology involves the allocation of projected

population and employment to specific locations across the study area. This is

achieved using different methods for the FLUAM scenario and the LUCIS-plus

scenarios. The datasets created for each of these scenarios are inputs to the third step

of the methodology which involves the modeling of transportation demand for each

scenario using the CFRPM. The CFRPM requires the datasets to contain a particular

set of attributes necessary to calculate travel demand in the model. An overview of

those attributes begins this section on Step Two and is followed by a more detailed

explanation of both FLUAM and LUCIS-plus and how they are used to create the three

scenarios compared in this study.

Transportation Analysis Zone Data Overview

The CFRPM requires socio-economic data for each TAZ in the study area, often

referred to generally as TAZ data. The TAZ data are both a reflection of forecasted land

uses and the distribution of population and employment across those land uses. The

TAZ data for the CFRPM are further divided into two sub-sets. One dataset called

ZDATA1 is used to calculate trip productions and accounts for residential land uses,

their population and number of automobiles. The other dataset called ZDATA2 is used









to calculate trip attractions and accounts for employment land uses and the number of

employees in each location. The CFRPM requires each dataset, ZDATA1 and

ZDATA2, to be in a database-file, or .dbf format. Appendix D lists the attributes and their

descriptions for each dataset.

In this study, after population and employment have been allocated for each

scenario the data is tabulated so that a ZDATA1.dbf and a ZDATA2.dbf exists for each

of the three scenarios to be modeled with the CFRPM.

FLUAM the Florida Land Use Allocation Method

FLUAM is the method used by the LSMPO for the creation of socio-economic data

used in the preparation of their 2025 LRTP, and thus is the method that was used for

the FLUAM scenario of this study, which uses the same dataset. Data processing

occurs using both GIS vector and tabular data, with final data compiled in a tabular

format. FLUAM is comprised of four general steps (LSMPO, 2006, p. 5-1):

1. Development of control totals;

2. Determination of developable lands;

3. Input of approved developments, manual adjustments, and overrides; and

4. Allocation of growth to traffic analysis zones.

Development of control totals

Tabular data are used to develop control totals for population, dwelling units and

employment totals for the county which are based on historical growth rates and an

average of medium and high projections from BEBR for 2025 (LSMPO, 2006, p. 5-2).

School enrollment forecasts also are included in the control totals as well as hotel/motel

dwelling units from approved developments (LSMPO, 2006, p. 5B-6).









Determination of developable lands

Developable lands are determined using parcel data in a GIS vector format for the

year 2000 as this is date of the last Census and therefore the most accurate population

data available. Using GIS processes the parcel data, which contains land use codes

used by the Florida Department of Revenue (FDOR), is queried to locate vacant parcels

and a GIS layer is created containing these parcels. This layer is combined with GIS

data for environmentally sensitive lands, lakes and rivers, and other undevelopable

land, which are subsequently selected and removed from the layer. PUDs and DRIs

are also removed as they are added in a later step of the process. TAZ and future land

use data are combined with the layer so that all vacant parcels can be summarized

according their future land use by TAZ number. A table is then created using the

attributes of this GIS layer and the total acreage of vacant developable land uses

available in each TAZ is summarized (LSMPO, 2006, p. 5-3). Figure 4-2 shows a

graphic representation of this part of the FLUAM process.

Input of approved developments, manual adjustments and overrides

For the development of factors used in the allocation process the study area is

delineated into ten planning areas that have similar development characteristics based

on the following (LSMPO, 2006, pp.. 5B-12-5B-13):

Existing land use;

Future land use;

Existing population and employment;

Location of cities;

Major roadway corridors;










* Character of areas; and

* Functional relationship of land uses.


.^ ......>- f 'u. .-..
Cvp~CZaK "w



uwy -- -...r -
-- '-..,.,'- -- ~'--





-.- .
.. .-
.-" '- '^. .-f^ s<;" .~
-, -
N' CrL

Sr


Pare-ii T.ggo .iria Trfgc Arnty" ZWMs

- Traflc Aniiybe Zor, -

--Prai(*l T reol .,ilh F 4il0r 1 I .

F LAur LanC L4*

.- UIOe'-ipabi. Lands

S Vacard Parcel


Figure 4-2. Determination of developable lands in FLUAM

Source. "Lake-Sumter 2025 Long-Range Transportation Plan" by LSMPO, 2006, p. 5-4


Activity centers are then located for each planning area by calculating the amount

of employment and dwelling units per TAZ. Depending on their amount of employment

and dwelling units each TAZ is assigned a weight and these weights are then combined

for each planning area to determine where the centers of most activity are located

(LSMPO, 2006, p. 5B-13).

Densities and intensities of land uses are required by FLUAM to distribute

population and employment into the developable lands layer. These are calculated for

each planning area by comparing the existing built-out densities for each land use and

the maximum allowable densities contained in the Lake County Comprehensive Plan









(LSMPO, 2006, p. 5B-10). For example, an existing urban area that is fully developed,

or built-out, may have a residential density of four dwelling units per acre, whereas the

Comprehensive Plan indicates the maximum density allowed was five dwelling units per

acre. By dividing the built-out density by the maximum allowed density, a multiplier

factor can be calculated for guiding density in future locations of that land use. In this

example the density multiplier would be 0.8.

Densities are estimated for each TAZ based on their Comprehensive Plan

designation and multiplier factors applied to the unoccupied developable land. If for

example a TAZ has five acres of developable land for residential uses and an approved

density of four dwelling units per acre, then using a density multiplier of 0.8 the

maximum number of dwellings for the vacant developable land in this TAZ is eight (i.e. 5

x 4 x 0.8 = 8). Employment density factors are calculated in the same way (LSMPO,

2006, p. 5B-11). By calculating the maximum allowable dwelling units and employment

that vacant developable land can accommodate, the allowable growth for each TAZ is

calculated. The maximum development each TAZ can accommodate is then calculated

by adding the allowable growth to the existing development which is then used to

determine if the allocated growth is physically possible within each TAZ. If the growth is

not possible the model re-allocates it to other TAZs (LSMPO, 2006, p. 5B-12).

Allocation of growth to TAZs

The allocation of dwelling units and employment is based on the control totals and

the maximum allowable development for each TAZ, and is performed using a

spreadsheet. Allocations for PUDs and DRIs are input manually into the spreadsheet

as the population and employment numbers are known for these areas, based on the

information included in their development approvals. The remainder of the population









and employment totals is allocated using a gravity model which assigns each TAZ an

attractiveness factor (LSMPO, 2006, p. 5B-14). A gravity model is based on the

concept that activity centers within a study area exert an influence on surrounding

locations based on the activity centre size and number of available activities. An

attractiveness factor for each TAZ is calculated using the size of the nearest activity

centre divided by the square of the distance of the TAZ to the activity centre. TAZs with

higher attractiveness factors are considered to develop faster than those with lower

factors (LSMPO, 2006, p. 5-4). The attractiveness factor is used in combination with

how close each TAZ is to other TAZs and the planning area's centre. Growth is then

allocated first to, and filling, TAZs closest to the planning area centre in a way that

simulates compact urban development (LSMPO, 2006, p. 5B-14).

School enrollment is entered manually into the TAZs where each school is located.

The school enrollment from the prior LRTP is reviewed and where determined, due to

forecasted increased enrollments in specific locations, additional school locations are

assigned. Approved developments that include new schools are added to the

appropriate TAZ and the approved number of students is allocated (LSMPO, 2006, p.

5B-15). Hotel/motel units are also allocated manually and are based on the number

and location of units in 2000. Units created as part of approved developments are

added, and likely locations of new units is determined through consultation with County

and municipal staff (LSMPO, 2006, p. 5B-15). Finally, all of the socio-economic data is

tabulated into two .dbf files, ZDATA1 and ZDATA2 for inputting into the CFRPM.

LUCIS-plus The LUCIS Planning Land Use Scenarios Method

LUCIS-plus is a GIS raster-based method for allocating regional population and

employment growth. It is currently being developed by Dr. Paul Zwick at the University









of Florida as a scenario planning tool for use in a wide range of planning fields,

including environmental and conservation planning, transportation planning and hazard

mitigation planning (P. Zwick, personal communications, 2010). The method employs

the use of the ESRI software ArcMap and its associated ModelBuilder application. A

suite of models are created that produce rasters capable of representing different future

land use scenarios. Research and development of LUCIS-plus has also included the

creation of several complimentary tools that work in conjunction with the suite of models

that enable planners to input scenario parameters via simple GUIs (graphical user

interfaces) containing drop-down menus for the selection of desired variables (A. Arafat,

personal communication, March 15, 2010). Additionally, programs using Python and

VB (Visual Basic) scripts have been developed as part of the LUCIS-plus research so

that the raster data can be summarized into a manageable tabular format for easier

analysis and integration into other modeling frameworks, such as transportation

demand and hazard simulation models.

This study uses LUCIS-plus as a method for creating two scenarios however it

does not use the complimentary tools or additional programs noted above. Instead data

inputs and variable parameters are entered manually into models, allocation fields

within the rasters are calculated manually, and output rasters are also manually

summarized for the creation of ZDATA tables.

There are differences between the FLUAM and the LUCIS-plus methods however

it is important to note one major difference between them before describing the study

methodology further. The scale at which growth is allocated in both methods is vastly

different. FLUAM allocates growth at the TAZ level, meaning future population and









employment are distributed across zones that vary in size from approximately 20 to

40,000 acres. The average size of a TAZ is approximately 2,800 acres. LUCIS-plus

allocates growth at a much smaller scale as it uses rasters containing cells that are 31

meters square, approximately one-quarter acre in size, to reflect the average-sized

residential building lot. In the LUCIS-plus scenarios growth is allocated to raster cells

and then aggregated to a TAZ level at the end of the allocation process.

It is possible to aggregate data from the smaller scale LUCIS-plus scenarios up to

the larger scale TAZ level so that it can be compared with data from the FLUAM

scenario. However, it is not possible to disaggregate the larger scale FLUAM data

down to a smaller scale for comparison with the LUCIS-plus data. Smaller scale data

used in the creation of the FLUAM scenario, such as the specific parcel and land use

data, was not available to this study. In the absence of such information any attempt to

disaggregate the larger scale data would require a large measure of estimation that

would nullify any meaningful comparison between the FLUAM and LUCIS-plus

scenarios. The comparison of scenarios in this study is therefore undertaken at the

larger TAZ scale rather than the smaller quarter-acre scale.

Following is a general list of steps used to create the two LUCIS-plus scenarios for

this study; each will subsequently be described in more detail. Not all of the steps listed

are strictly speaking a part of the LUCIS-plus method. For example, the second, third

and fifth steps involve the calculation of gross land use densities, control totals for

population and employment growth, and specific growth patterns such as land use

intensities. These are variables needed for input into models used in LUCIS-plus, but









the steps themselves are not a part of the allocation method and could be developed

using a variety of different methods.

The steps used for the LUCIS-plus scenarios are:

1. Creation of a combine raster that includes all the variables needed to

identify specific areas suitable for growth;

2. Establishment of existing gross densities;

3. Establishment of control totals for population and employment growth;

4. Allocation of existing population and employment at gross densities;

5. Establishment of specific guidelines for how population and employment

growth will be spatially distributed;

6. Allocation of population, employment and school enrollment to specific

locations identified as suitable for growth.

Creation of combine raster to identify areas suitable for growth

The LUCIS-plus method centers on the use of a final GIS raster that represents

the study area and into which future population and employment is allocated. The final

raster, referred to as a combine raster, is a combination of several other independently

generated rasters each containing information about the study area that are useful to

planners when forecasting where future land use change might occur.

'Combine' geo-processing tool. A description of the ArcMap geo-processing

tool 'combine' is necessary for the explanation of the model used in the LUCIS-plus

method. Using the 'combine' tool multiple rasters can be combined to create a single

output raster. All GIS vector and raster files contain an associated table in which

information or attributes about the geographic space the files represent is stored. The

attribute table of any GIS raster contains rows and columns, with each row containing a


100









unique record relating to geographic locations or cells in the raster and each column a

unique attribute or item of information about the raster cells. The output raster of a

combine process contains columns that represent a unique item for each input raster

and rows that contain unique combinations of the values of the input rasters. Figure 4-3

illustrates how the values of multiple input rasters are attributed to an output raster

during a combine process. The value column in the output raster contains a record for

each unique combination of cells from the input rasters. For the example shown in

Figure 4-3, the combination of values of both the input rasters creates in the output

raster a unique value of 1, seen in both the top left and bottom right cells of the output

raster. In the attribute table of the output raster this creates a row containing

information about cells that contain the unique value of 1 (value = 1), of which there are

two cells (count =2) and the original value from each input raster is retained in the

columns bearing the names of the input rasters. Where a raster cell contains no data,

meaning the location the cell represents in the raster contains no information in the

geographic space it relates to, it has a value of NoData. When any input raster cell is

combined with a cell containing NoData the corresponding cell in the output raster will

also contain NoData.


101















n M Value = NoData
InRasl InRas2 OutRas
Value Count Code Value Count Type Value Count InRasl InRas2
0 5 002 PAX 1 2 1 0
1 004 1 4 HAR 2 2 1 1
2 3 006 2 3 WIN 3 1 0 1
4 2 000 3 3 SAN 4 3 0 0
5 1 1 3
6 1 2 1
7 2 2 2
8 1 4 3
9 1 0 2


Figure 4-3. The 'combine' process.

Source. "ArcGIS Desktop 9.3 Help Combine" by ESRI, 2010.


Allocation buffers. In this study five combine rasters are created for each

LUCIS-plus scenario. In total 10 combine rasters are created; five for a low density

LUCIS-plus scenario and five for a high density LUCIS-plus scenario. For each

scenario the five combine rasters represents an allocation of population and

employment into a specific zone or allocation buffer. The combine rasters contain 31

meter by 31 meter cells, creating a high spatial resolution and increasing file size as the

extent of each allocation buffer increases. To increase computational speed of the

models and the allocation process it was determined that a separate combine raster for

each allocation buffer would be necessary. Data from the five combine rasters are

collated once all allocations have been made for each scenario to create the ZDATA1

and ZDATA2 .dbf files necessary for step three of the methodology.

The five allocation buffers are illustrated in Appendices E through Hand include a

description and rationale of the buffer distances. The location of transit stops used in

the creation of the buffer zones are those contained in a GIS vector file created by the


102









ECFRPC of visionary future transit routes and stations used in modeling future growth

for their 2060 Strategic Regional Policy Plan (P. Laurien, personal communication,

January 20, 2010).

Combine raster model. Each combine raster is created using a similar LUCIS-

plus model. Model inputs and outputs are the same for each of the combine rasters,

with the exception of the buffer raster which determines the spatial extent of the

allocation and the inclusion of DRI (developments of regional impact) rasters in buffers

4 and 5. Figure 4-4 shows the model of the combine raster for the allocation of buffer 1

and will serve as an example for explaining the functioning of all the combine raster

models. For ease of explanation the illustration of the model highlights three phases

that will be described separately. When the model is run the processing occurs in the

order the phases follow.

Combine model phase one. Data inputs to the first phase of the model include

a raster of the extent of buffer 1 and a raster of the opportunities for mixed use

development. The buffer 1 raster was created by converting a vector shapefile of the

area to a raster and then reclassifying the cells so that only cells representing the

location of buffer 1 contain a value and all other cells are assigned a NoData value.

When this raster is combined with all other input rasters in the combine model all cells

that overlap the NoData cells will also be assigned a NoData value, thereby masking

those cells out of the output raster. This process is referred to as creating a mask. In

the combine models of this study the buffer rasters act as masks for delineating where

specific configurations of urban growth will occur.


103


























Phase 2


Figure 4-4. Model of combine raster for allocation buffer 1. Note. Diagram courtesy of author.


Diagram Key


Diagram Key









The raster representing the opportunities for mixed use development was created

in a separate model that combines individual preference rasters for each of the following

land uses; commercial, institutional, multi-family, retail and service which represent the

mix of uses that are desirable in TOD areas. Each cell in the mixed use opportunities

raster contains a 5-digit value that represents the level of conflict or opportunity that

exists between the different land uses in that cells location. For example a cell with a

value of 33333 represents a good opportunity for mixed use as all of the land uses are

highly preferred in that location, and a cell with a 11111 value has few mixed use

opportunities.

Using an extraction tool the values from the mixed use opportunities raster are

extracted based on their location within buffer 1. Cells with a high opportunity for mixed

use were extracted next. Only cells that had a high preference (3) for multi-family

combined with a medium (2) or high (3) preference for any of the other land uses were

extracted. The suitability of areas for mixed use development is also influenced by the

size or acreage of land under consideration. Areas of less than an acre are likely to be

less desirable for this type of development. Therefore it is necessary to locate cells of

high mixed use opportunity that are clustered into groups of raster cells of four or more;

as each raster cell is approximately one-quarter acre. This is performed in the model

using the 'region group' process which is illustrated in Figure 4-5.


105











1 1 1 1

1 1 3 3

5 5 3
---I^^
1 1 ';5-
U Value = NoData
InRasI OutRas


Figure 4-5. The region group process

Source. "ArcGIS Desktop 9.3 Help Region Group" by ESRI, 2010.


During the region group process the value of the output raster cells represents a

unique value that corresponds to its relationship with its neighboring cells. The process

moves through the raster starting in the cell in the top left corner moving to the right and

down through subsequent rows. The attribute table of the output raster contains a

column listing the count of cells with the same value. Using Figure 4-5 as an example,

if the cell sizes had an area of one-quarter acre then none of the cells are grouped in a

one acre configuration, meaning in the attribute table none had a count of 4 or greater.

In this example the highest count is 3, representing an area of only three-quarters of an

acre.

At this point in phase one the output raster (Buffer 1 Reg Group) contains all the

cells in buffer 1 that have a high opportunity for mixed use development. The cells also

have a value that represents their relationship to neighboring cells so that groups of a

certain size can be located. The final stage of phase one involves reclassifying the cells

with a value of NoData to zero so that all data is retained during the subsequent

combine process of the model.


106









Combine model phase two. The second phase of the model uses the combine

process described earlier. Fourteen rasters form the data inputs to the process which

represent the different attributes necessary for defining suitable locations for future

urban growth. Seven of the fourteen rasters represent different land use suitabilities

determined using the LUCIS process described in the literature review. The original

suitability rasters were created by Dr. Paul Zwick of the University of Florida as part of a

study for the ECFRPC in 2008-2009 concerning future land use change in the central

Florida region. The seven rasters represent the suitability for the following land uses:

single family residential, multi-family residential, commercial, service, retail, institutional,

industrial.

Each suitability raster contains values ranging from 1 to 9, with 9 representing the

highest suitability and 1 the lowest suitability. In the attribute table of each suitability

raster therefore only 9 records or values exists, and each value has a count

representing the number of cells with that unique value. In this study cells are selected

based on their suitability using a tool that queries the attribute table to select cells with

particular values, for example select all cells with a value equal to 9, or select all cells

with a value of less than 7 but greater than 4. It is not possible therefore to select a

fewer number of cells than there is in total for that value. For example, if only 90 cells

were needed with a value of 9 but that row in the attribute table (with a value of 9)

contained a total of 200 cells it would not be possible to select fewer than 200 using this

selection method. Therefore it is necessary to reclassify the data so that there is a

greater range of values on which to base selections. A slice tool is used to reclassify


107









the data into smaller slices, or more values in the attribute table. The seven suitability

rasters were reclassified into 1,000 slices of equal area.

A raster representing the level of conflict between the three major land use groups

of agriculture, conservation and urban is included as an input. This raster was also

created by Dr. Zwick for the ECFRPC study mentioned earlier. The cell values of this

raster represent the preference for each of the three land use groups, with a value of

333 corresponding to an area that is equally preferred by all uses, and a value of 111

representing an area that has a low preference for all three.

A raster of land uses is included as an input so that existing population and

employment can allocated according to their land use. The land use values correspond

to the generalized TAZ land use categories listed in Appendix I. A raster of the TAZs is

also included so that each cell can be summarized in the final stage of the allocation

process according to its TAZ. A buffer raster is included in the combine process that

acts like a mask so that only cells located in that buffer are included in the output

combine raster. The mixed use opportunities raster described in phase one is also

included as is the region group raster that was a result of phase one.

The remaining input raster is one that represents areas that have a potential for

redevelopment. This was created by querying the vector parcel data to locate areas

that met certain criteria and then exporting the file to a raster format. The selection

criteria for redevelopment parcels are as follows:

Cells within half-mile buffer of future transit stations;


108









Parcels with a 'just value' that is below the average 'just value' for

properties of the same land use 'just value' is the price a property may

sell for in an open market it is synonymous with 'market value';

Parcels with buildings that will be more than 50 years old in 2025, i.e. were

built prior to 1975; and

Exclude parcels with the following uses churches, orphanages, hospitals,

schools, sanitariums, homes for the aged, and cemeteries.

Combine model phase three. The third phase of the LUCIS-plus model

involves the adding of necessary fields for the allocation process. These fields are as

follows:

MFPOP -for multi-family population

SFPOP for single family population

COMEMP for commercial employment

SEREMP for service employment

INDEMP for industrial employment

EMPDEN for employment density

POPDEN for population density

NEWLU for new land use code

ACRES for total acres

The LUCIS-plus model is then run five times for each scenario, each time using one of

the five buffers as a mask.


109









Establishment of existing gross densities

To identify areas suitable for future growth the size and distribution of existing

population and employment must first be determined. GIS vector-format parcel data

from the year 2007 was the most current land use data available during the study and

forms the basis for establishing the extent and land use distributions of the urban form

in the study area. The parcel data contains a land use code used by the FDOR, which

identifies individual land uses across a range 99 different uses. This study compares

only five different land use categories; single family residential, multi-family residential,

commercial, service, and industrial uses. The FDOR codes are aggregated to reflect

the five generalized land use categories of this study. Appendix I lists the FDOR codes

aggregated to reflect TAZ land use codes. A table containing the TAZ land use codes is

joined to the GIS parcel layer and the TAZ land use codes are copied over to a new

field within the parcel layer called TAZ_LU.

Land that is not available for development is then removed from the parcel layer

using the GIS 'erase' geo-processing tools which removes areas from the parcel layer

that overlap with the 'erase' layer. Conservation areas and lakes and ponds are

removed from the parcel layer in this way. Roads, rights-of-way and other

undevelopable land uses such as undefined and centrally assessed parcels are

removed by selecting them from within the parcel layer and deleting their records from

the layer's associated attribute table. The total number of acres within each of the TAZ

land use categories is then calculated by summing the area of the individual parcels

within each category. Table 4-1 shows the total acreage in the study area by TAZ land

use categories. The total number of acres for each land use is the denominator in the


110









calculation of gross density: Total population or employment in land use X Total

acreage of land use X.

Table 4-1. Total acreage by TAZ land use categories.
TAZ Land Use Category Total Acres
Single family residential 67,734
Multi-family residential 1,482
Mobile homes residential 29,312
Commercial 3,256
Service 37,276
Industrial 161,044
Vacant residential 69,213
Vacant commercial 5,859
Vacant industrial 1,325
Vacant 29,969
Total acres 406 ,470


The numerator in the gross density calculation is the total population or

employment in a particular land use. Population values are calculated by averaging

data from the Census Bureau and available TAZ data. Census Bureau estimates of

population and dwelling units by type of residence for the year 2007 are summarized

according to their residential TAZ land use code; single family, multi-family or mobile

home. These population totals are then divided by the total number of acres in each

land use category to determine a gross density. Table 4-2 lists Census Bureau

population estimates and the gross density for each residential TAZ land use.

Table 4-2. Census Bureau estimates for 2007 for Lake County population by residential
TAZ land use categories
TAZ Land Use Category Population Total Acres Gross Density
Single family residential 171,335 67,734 2.53
Multi-family residential 40,360 1,482 27.24
Mobile homes residential 73,272 29,312 2.51
Total population 284,967 98,528 2.89
Source. Adapted from "Selected housing characteristics: 2005-2007, Lake County Florida" by
United States Census Bureau, 2007.


111









TAZ data is also used to supplement the Census Bureau population estimates.

TAZ data for the years 2000, 2012 and 2025 is provided with the CFRPM and the

datasets for 2000 and 2012 are compared to calculate an average annual increase in

population and employment totals over that period. The average annual increase is

then applied to the 2000 TAZ data to project totals forward to the year 2007. Population

and employment totals are summarized by TAZ land use category and gross densities

calculated. Table 4-3 lists the population estimates and gross densities; employment

estimates are listed in a subsequent table.

Table 4-3. TAZ estimates for 2007 for Lake County population by residential TAZ land
use categories
TAZ Land Use Category Population Total Acres Gross Density
Single family residential 163,994 67,734 2.42
Multi-family residential 38,747 1,482 26.15
Mobile homes residential 70,127 29,312 2.39
Total population 284,967 98,528 2.77
Gross density estimates from both the Census and TAZ data are averaged to

establish the gross residential densities that will be used to distribute existing residential

population across the study area. Table 4-4 lists the existing gross residential densities

for the study area.

Table 4-4. Existing gross residential densities for study area
Existing
TAZ Land Use Category Census TAZ Gros cities
Gross Densities
Single family residential 2.53 2.42 2.48
Multi-family residential 27.24 26.15 26.69
Mobile homes residential 2.51 2.39 2.45
Gross density 2.89 2.77 2.83


Employment gross densities are calculated in a similar way, by averaging two

sources of estimates. Employment projections for the years 2005, 2010 and 2025

created by the ECFRPC for a 2008-2009 study of land use modeling techniques are

used to establish employment estimates for the year 2007. The 2007 estimate was


112









based on the average annual increase in employment between the 2005 and 2010

estimates. The ECFRPC used the REMI Policy Insight model, an econometric input-

output model, to project employment across 23 categories. These employment

categories are listed in Appendix J according to their TAZ land use category;

commercial, service or industrial uses. Table 4-5 lists the gross employment densities

using the ECFRPC employment projections.

Table 4-5. ECFRPC estimates for 2007 for Lake County employment by TAZ land use
categories
TAZ Land Use Category Employment Total Acres Gross Density
Commercial 19,657 3,256 6.04
Service 75,681 37,276 2.03
Industrial 23,805 161,044 0.15
Total employment 119,143 201,576 0.59


Gross employment densities are calculated using the 2007 TAZ data and are

listed in Table 4-6.

Table 4-6. TAZ estimates for 2007 for Lake County employment by TAZ land use
categories
TAZ Land Use Category Employment Total Acres Gross Density
Commercial 28,286 3,256 8.69
Service 62,069 37,276 1.67
Industrial 22,211 161,044 0.14
Total employment 112,566 201,576 0.56


Gross density estimates from both the ECFRPC and TAZ data are averaged to

establish the gross employment densities that will be used to distribute existing

employment across the study area. Table 4-7 lists the existing gross employment

densities for the study area.


113









Table 4-7. Existing gross employment densities for study area
Existing
TAZ Land Use Category ECFRPC TAZ Gros cities
Gross Densities
Commercial 6.04 8.69 7.36
Service 2.03 1.67 1.85
Industrial 0.15 0.14 0.14
Gross density 2.89 2.77 0.78


Establishment of control totals for growth

To establish the amount of growth to be allocated into the future scenarios,

estimates of 2007 population and employment are deducted from forecasts for 2025.

Table 4-8 lists the estimates and forecasts used and their averages that result in the

control totals used for this study.

Table 4-8. Population and employment control totals
Population Projections / Estimates Total Average Control Total
2025 Population BEBR (med-high) 449,050
2025 Population TAZ 463,500
2025 Population Average 456,275

2007 Population Census Bureau 288,580
2007 Population TAZ 272,860
277,224
Total Population Growth 2007 to 2025 179,051

Employment Projections / Estimates Total Average Control Total
2025 Employment ECFRPC 173,138
2025 Employment TAZ 185,580
179,359
2007 Employment ECFRPC 119,143
2007 Employment TAZ 112,566
115,855
Total Employment Growth 2007 to 2025 63,504


Allocation of existing population and employment using gross densities

Using the attribute table of the combine raster for each of the buffers, cells are

selected based on their existing TAZ land use category. Once cells are selected the

field calculator tool is used to create a new value for the population or employment field


114









by multiplying the count of cells by the appropriate gross density. The following script

describes the calculation for the allocation of existing single family uses:

[COUNT] (2.48/4.211) where [COUNT] represents the number of cells with SF

use, 2.48 the SF gross density per acre and 4.211 the value by which the per acre

density must be divided to reflect the 31 by 31 meter cell size.

All cells with a TAZ land use of SF, MF, COM, SER and IND are selected and

population or employment is allocated to them using their existing gross density. The

NEW LU field is also calculated to show a new land use code for each cell that reflects

the TAZ land uses. This is done so that the data can be summarized at the end of the

allocation process and also to show where which cells may have changed land uses.

Table 4-9 shows the NEW LU codes.

Table 4-9. NEW LU codes
Land Use NEW Land Use NEW_L
Category _LU Category U
Single family SF 1 Vac. Commercial VACCOM 8
Mobile home MOB 2 Vac. Residential VACRES 10
Multi-family MF 3 Agriculture AG 11
Industrial IND 4 Conservation CON 12
Commercial COM 5 Ag Preservation AGPR 13
Service SER 6 Mixed Use MU 14
Vac. Industrial VACIND 7 Vacant VAC 15


The attribute table of each of the five combine rasters is summarized so that the

total number of population, employment and acres can be tabulated.

Establishment of guidelines for distribution of growth

TOD guidelines. The two LUCIS-plus scenarios are examples of an urban growth

pattern that is compacted within existing transportation corridors and clustered near

future transit stations and routes to enhance the achievement of long-range

transportation goals. The low scenario allocates at a lower density and the high at a


115









higher density. Draft guidelines created by the FDOT for the design of TODs are used

to determine the appropriate mix of land uses, densities and intensities of development

within a one-quarter mile distance of transit stations. A range of development standards

are presented in the guidelines that reflects a variety of different urban contexts that,

depending on the type of planned future transit such as light rail or rapid bus, would be

most appropriate considering the existing and future land uses of the area (FDOT,

2009b, p. 2). According to Lake County's Transit Development Plan (TDP) for 2009 to

2020, the future transit system planned is one that enhances existing fixed route bus

transit, and incorporates new bus rapid transit and commuter rail routes. Appendix K

shows the planned transit system and a description of the modes and routes it includes.

Based on the type of transit in the TDP and Lake County's existing low density urban

development, the TOD standard chosen to guide this study is one described in the

FDOT design guidelines as T3. Appendix L shows the T3 standards.

4-6 shows an Excel spreadsheet used to calculate the densities for mixed use

development within TOD areas for the low LUCIS-plus scenario. This scenario uses the

low value in the ranges suggested by FDOT for TOD population, employment and

dwelling unit densities.


116















Total acres in buffers 1 & 2 (TOD areas)
Total acres in redev in buffers 1 & 2 (for mixed use)

FDOT- TOD Guidelines
Population per acre = 15
Jobs per acre = 5
Dwelling units per acre =5

Existing 2007 Pop Buffers 1 & 2
RES population (SF, MF, MOB)
JOBS (COM, SER, IND)
DU (RES pop / 2.48 ppl per hhold

Additional pop needed in Buffers 1 & 2 to meet
minimum TOD guideline
RES population (MF only)
JOBS (COM, SER, IND)
DU (RES pop / 2.48 ppl per hhold)

MU Densities for Buffers 1 & 2
RES population (MF only)
JOBS (COM= 25%)
JOBS (SER= 75%)


2107 acres
526 acres


31606 (2107 acres @ 15 people/acre)
10535 (2107 acres @ 5 jobs/acre)
10535 (2107 acres @ 5 DU/acre)


2867
4013
1156



28740
6523
11589


54.59 people per acre
3.10 jobs per acre
9.29 jobs per acre


Figure 4-6. Calculation of mixed use densities for the low LUCIS-plus scenario.




Figure 4-7 shows an Excel spreadsheet used to calculate the mixed use densities


for the high LUCIS-plus scenario. This scenario uses the median of the medium and


high values in the ranges suggested by FDOT for densities in TODs.


117














Total acres in buffers 1 & 2 (TOD areas) 2107 acres
Total acres in redev in buffers 1 & 2 (for mixed use) 526 acres

FDOT- TOD Guidelines
MED pop per acre = 47 99034 (2107 acres @ 47 people/acre)
MED jobs per acre = 22 46356 (2107 acres @ 22 jobs/acre)
MED DU per acre = 17 35821 (2107 acres @ 17 DU/acre)

Existing 2007 Pop Buffers 1 & 2
RES population (SF, MF, MOB) 2867
JOBS (COM, SER, IND) 4013
DU (RES pop / 2.48 ppl per hhold 1156

Additional pop needed in Buffers 1 & 2to meet
minimum TOD guideline
RES population (MF only) 96167
JOBS (COM, SER, IND) 42343
DU (RES pop/ 2.48 ppl per hhold) 38777

MU Densities for Buffers 1 & 2
RES population (MF only) 182.66 people per acre
JOBS (COM = 25%) 20.11 jobs per acre
JOBS (SER = 75%) 60.32 jobs per acre





Figure 4-7. Calculation of mixed use densities for the high LUCIS-plus scenario.




Establishing the amount of growth to be allocated into each buffer was performed

using an Excel spreadsheet that calculated the amount of growth in population and

employment for the period 2007 to 2025. Growth is compacted in both the LUCIS-plus

scenarios to within a three mile distance of existing and future transit routes. Based on

the existing proportions of single, multi family and mobile home populations within that

distance, adjustments are made in both scenarios to increase proportions thereby

compacting growth. In 2007 forty percent of the total population lived within a three mile

distance to transit and sixty percent outside of that distance. Sixty percent of the total

2007 population was in single family residential, fourteen percent in multi-family, and


118













twenty-six percent in mobile homes. Figures 4-8 and 4-9 show Excel spreadsheets


used to determine how growth is to be allocated for both the low and high scenarios,


respectively.


Population


BEBR 2025 projection (med-high)
TAZ 2025 projection
Total existing 2007 pop allocated


Estimated number of people to allocate 2007 to 2025
Number to TOD (redev in buffers 1 & 2)
Number to Vacant Res and Greenfield

2007 40% of total pop inside Buffers 1 through 4
2007- 60% of total pop outside Buffer4

2007 to 2025 Growth 65% of pop inside Buffers Ito 4
2007 to 2025 Growth 35% of pop outside Buffer 4


50% of total 2025 pop to SF
(2007- 60% of total pop in SF)



30% of total 2025 pop to MF
(2007-14% of total pop in MF)




20% of total 2025 pop in MOBILE
(2007- 26% of total pop in MOB)


Growth 4
50% inside Buffer 1to 4
50% outside Buffer 4

Growth 4
Inside Buffers 1 & 2
Inside Buffers 3 & 4
Inside Buffer 5 (DRI)


Low LUCIS-plus Scenario
449050
463500 456275 Average
277224
179051


179051
* 28740 16% of total growth
+ 150311 84% of total growth
179051 people


* 116783
+ 62269
179051 people


61182
30591
30591

98004
28740
57452
11812


(all outside Buffer 4) + 19865
179051 people


2025 2007
228138 166955




136883 38879





91255 71390
456275 277224


Industrial employment
Commercial employment
Service employment


8167
13143
41394
62704 jobs


2025 2007
32,352 24185
37,119 23976
109,889 68495
179359 116655


Figure 4-8. Guidelines for allocation of growth for low LUCIS-plus scenario


119


Employment














High LUCIS-plus Scenario


BEBR 2025 projection (med-high)
TAZ 2025 projection
Total existing 2007 pop allocated


Estimated number of people to allocate 2007 to 2025

Number to TOD (redev in buffers 1 & 2)
Number to Vacant Res and Greenfield

2007 40% of total pop inside Buffers 1 through 4
2007- 60% of total pop outside Buffer 4


78% inside Buffers 1 through 4
22% outside Buffer 4


40% of total 2025 pop in SF




40% of total 2025 pop in MF





20% of total 2025 pop in MOBILE


Growth 4
100% inside Buffer 1to 4
0% outside Buffer 4

Growth 4
Inside Buffers 1 & 2
Inside Buffers 3 &4
Inside Buffer 5 (DRI)


449050
463500 456275 Average
277224
179051
117051


+ 96167 54% of total growth
+ 82884 46% of total growth
179051 people


4 157912
4 19597
177509 piopli


15555 people
15555
0

143631 people
96167
47464
0


h (all outside Buffer 4) 4 19865
179051 people


2025 2007
182510 166955




182510 38879





91255 71390
456275 277224


Industrial employment
Commercial employment
Service employment


8167
13143
41394
62704 Iobi


2025 2007
32,352 24185
37,119 23976
109,889 68495
179359 116655


Figure 4-9. Guidelines for allocation of growth for high LUCIS-plus scenario





Allocation of growth to suitable areas based on established guidelines


Order of allocations. In both LUCIS-plus scenarios growth is allocated in order


by buffer, beginning with buffers 1 and 2 which represent TOD areas, followed by buffer


3 which represents commercial corridors, then buffer 4 which includes remaining areas


within a three mile distance of transit, and finally buffer 5 which is the remainder of the


study area.


120


Population


Employment











Using a spreadsheet into which are entered the existing and control totals for

population and employment, a total amount of growth to be allocated for each buffer is

determined and can be further disaggregated by land use according to the established

guidelines. As growth is allocated the amounts are deducted from the control totals to

determine how much growth remains to be allocated. Figure 4-10 shows an Excel

spreadsheet that illustrates this process.


Total 34951 6410 80721 205498 308050 635630


RES
EMP
Total


26834
8116
34951


4146 53682
2264 27038
6410 80721


145624 226000 456287
59874 82051 179343
205498 308050 635630


4 635634



-12 456275
16 179359
4 635634


121


SUMMARY allocations into existing 2007 land uses

Left to
Buffer 1 Buffer 2 Buffer 3 Buffer 4 Buffer 5 Total Allocate 2025 Tota
SF 396 567 5601 58694 101697 166955 61182 228138
MF 872 1032 10291 34718 63356 110268 117869 228138
SER 545 393 4334 23529 39693 68495 41394 109888.5
COM 2276 794 10228 3882 6797 23976 13143 37118.5
IND 0 5 120 3286 20774 24185 8167 32352
Total 4089 2791 30574 124108 232317 393879 241755 635634



RES 1268 1599 15893 93411 165053 277224 179051 456275
EMP 2821 1192 14681 30697 67264 116655 62704 179359
Total 4089 2791 30574 124108 232317 393879 241755 635634
SUMMARY all land uses

DONE DONE DONE DONE DONE Left to
Buffer 1 Buffer 2 Buffer 3 Buffer 4 Buffer 5 Total Allocate 2025 Tota
SF 226 412 8168 87131 132235 228173 -36 228138
MF 26608 3734 45515 58493 93764 228114 24 228138
SER 4879 1419 14144 43672 45774 109888 0 109889
COM 3237 840 12769 10343 9930 37119 0 37119
IND 0 5 125 5859 26347 32336 16 32352


Figure 4-10. Calculation of remaining growth to be allocated









For example, if the total amount of population growth for TOD areas (buffers 1

and 2) is 28,740 then beginning in buffer 1, a selection of the most suitable locations

can be made, and based on that area the amount of population that can be allocated

can be determined by dividing that area by the population or employment density. Thus

if 1,000 suitable cells are located in buffer 1 for multi-family population (i.e. 237.43

acres) then using the TOD density for multi-family (54.59 people/acre) a total of 12,963

people can be allocated in buffer 1. The remaining 15,777 people (28,740 less 12,963)

need to be allocated in buffer 2. A separate spreadsheet is used for each buffer's

allocations so that a running total is kept to guide what remains to be allocated. Once

all growth has been allocated in the buffer, the attribute table of the combine raster for

that buffer is summarized and the new population and employment totals are inserted

into the spreadsheet of control totals (Figure 4-10) to determine what remains to be

allocated in subsequent buffers.

Selection of suitable areas and allocation of growth. The selection of suitable

areas for allocating growth is steered firstly by the established guidelines and secondly

by land use suitability analysis. The allocation of growth to buffer 1 will be described in

detail as an example of how this process was undertaken for all of the buffers, and then

an overview of the selection process for the remaining buffers is provided Appendix M.

Example allocation to buffer 1. Using the control totals the amount of

population and employment growth to be allocated is determined. As buffers 1 and 2

are TOD areas all population and employment growth is first to be allocated into

suitable locations for mixed use redevelopment and then into vacant land uses

(VACRES, VACCOM, VACINST or VAC). In this analysis all mixed use development


122









will incorporate both residential and commercial/service components. The residential

component is multi-family population and the employment component is divided

between commercial and service according to their relative proportion of total

employment growth. Twenty-five percent of all non-industrial employment growth is

forecasted to be commercial and seventy-five percent service, so in mixed use areas

the employment growth is distributed according to those percentages.

In the attribute table of the buffer 1 combine raster, a selection by attribute is

performed to firstly locate cells within redevelopment areas (RCLREDVBF1_2 > 0).

From this selection, cells that are grouped in areas of one acre or more (four or more

cells) are located (RREGIONBUFF1 >= 4). From this selection, cells that have a

medium to high mixed use opportunity are selected (RCINMFRSCONF = 33333 OR

RCINMFRSONF = 23333 OR ....RCINMFRSCONF = 22223). For the selected cells the

field calculator tool is used to allocate population and employment at the mixed use

densities into the appropriate fields (MFPOP, COMPOP and SERPOP). A NEW_LU

code of 14 (MU) is also calculated. The count of cells in the selection is entered into the

buffer 1 working spreadsheet and the remaining number of cells to be allocated is

calculated.

The remaining growth for buffer 1 is to be allocated into vacant land uses (VAC,

VACRES, VACCOM and VACINST). This selection starts by selecting all cells that

have not already been allocated in the previous step (NEW_LU = 0), then selecting all

cells that have vacant uses (TAZ_LU = VAC OR TAZ_LU = VACRES...). From this

selection all cells that have a medium to high mixed use opportunity are selected,

followed by those grouped in areas one acre or larger. Using the field calculator,


123









population and employment is allocated to the appropriate fields and an NEW_LU code

of 14 is given. The count of cells in the selection is entered into the spreadsheet and if

any growth remains unallocated it is carried over to buffer 2. Finally the attribute table

of the combine raster is summarized and the new population and employment totals are

added to the control totals spreadsheet (Figure 4-10) so that an accurate count of

remaining growth can be determined.

Allocation to remaining buffers. Allocation of growth to the remaining buffers

continues in a similar way. Opportunities for mixed use redevelopment exist in buffers

1, 2 and 3 and are selected first in those buffers, followed by vacant land uses based on

their level of land use conflict, followed by the suitability for individual land uses. Mixed

use opportunities do not exist in buffers 4 and 5 so selection in those buffers begins with

allocation into DRIs and the remainder according to land use conflict and land use

suitability. The method of allocation is the same for both the low and high LUCIS-plus

scenarios with the only difference between them being their level of compactness (as

determined by their allocation guidelines) and their mixed use densities. Appendix M

lists the general steps and criteria used for allocation in all buffers.

Creation of ZDATA tables. Once all growth has been allocated, summaries of all

the combine raster attribute tables are collated into one spreadsheet. The data is then

summarized according to TAZ number and the data separated according to the required

attributes of the ZDATA1 and ZDATA2 .dbf files. Appendix D lists those attributes and

highlights in bold type the attributes that are derived from the LUCIS-plus method. The

dwelling units totals (SFDU and MFDU) are calculated by dividing the SFPOP and

MFPOP by the average household size of 2.48 people. With the exception of school


124









enrolment, the remaining fields are carried over from the 2025 TAZ data that

accompanies the CFRPM. School enrolment is calculated outside of the LUCIS-plus

method and is done using a combination of GIS and spreadsheet functions.

Allocation of school enrolment. School enrolment is calculated and allocated

following several steps. Using a GIS vector file of the school area zones, the 2025

population (as allocated in each LUCIS-plus scenario) is summarized. Then using data

from the Census Bureau 2007 household estimates, the percentage of school aged

children is determined according to school type (elementary, middle and high school).

These percentages are used to determine the number of school aged children in each

school area zone according to school type. Figure 4-11 shows a spreadsheet used to

calculate school enrolment for 2025 for the low LUCIS-plus scenario. A similar

calculation is done for the high LUCIS-plus scenario. TAZs where the individual schools

are located are selected in the ZDATA2 .dbf file and the enrolment allocated.


125











% of total pop (from Census 2007 estimates for Lake County)
Elementary age 6.6%
Middle School a 3.5%
High School age 4.8%
Total School age 15.0%


2025 Pop
Umatilla High 34882
Eustis High 57254
Mt Dora High 31238
Tavares High 59253
South Lake Higt 99666
Leesburg High 130418
East Ridge High 43801
456511


Elementary Middle
2317 1228
3803 2016
2075 1100
3936 2086
6620 3509
8663 4591
2909 1542
30324 16071


High School Enrol f total enrol
1687 5232 8%
2769 8588 13%
1511 4686 7%
2866 8888 13%
4821 14950 22%
6308 19563 29%
2119 6570 10%


22082


68477


100%


College enrolment


TOTAL


4328
72805


Assumption
Total projected enrolment to be divided equally between the number of facilities of that type in
each SAZ. The assumption is that an adequate number of facilities will exist to accommodate
enrolment and that each facility would have an equal carrying capacity



Figure 4-11. Calculation of school enrolment for 2025 for the low LUCIS-plus scenario

Step Three

The third and final step of the methodology is the modeling of transportation

demand for the three future land use scenarios created in the previous step. The study

uses the CFRPM version 4.5 which is run using the Citilabs software CUBE/VOYAGER.

Inputs to the modeling process that are unique to each scenario include TAZ data

(ZDATA1 and ZDATA2 .dbf files) and a transit route system (TROUTE.lin). Other inputs

to the CFRPM, such as the highway network and its associated characteristics are not

altered from those contained in the 2025 scenario of the CFRPM and remain the same

across all scenarios.


126









FLUAM Scenario TAZ Data and Transit System

The FLUAM scenario uses the 2025 TAZ data files that accompany the CFRPM.

This scenario is modeled using the 2025 transit system that accompanies the CFRPM,

referred to in this study as transit system A. Figure 4-12 is a map of transit system A

routes and stops. This system is comprised of six local bus routes, each operating on

60 minute headways in both peak and non-peak hours. Headway indicates the

frequency of service. In the CFRPM headway refers to peak times and headway2 to

non-peak times. Table 4-10 lists the modes and headways for each route in transit

systems A and B.

Table 4-10. Transit systems A and B routes, modes, and headways
Transit System A Transit System B
Route Mode Headwayl Headway2 Mode Headwayl Headway2
Lake 1 Local bus 60 min 60 min Local bus 30 min 60 min
Lake 2 Local bus 60 60 Local bus 30 60
Lake 3 Local bus 60 60 Local bus 30 60
Lake 4 Local bus 60 60 Local bus 30 60
Lake 5 Local bus 60 60 Local bus 30 60
Lake 6 Local bus 60 60 Local bus 30 60
Lake 7 -- Local bus 60 60
Lake 8 --- Local bus 30 60


127






































Figure 4-12. Transit system A routes and stops. Note. Map courtesy of author.


LUCIS-Plus Scenarios TAZ Data and Transit System

The LUCIS-plus scenarios use their respective TAZ datasets created in the

previous step of this methodology. Both LUCIS-plus scenarios use the same transit

system which was created to resemble the system outlined in the Lake County TDP and

shown in Appendix K. It is referred to in this study as transit system B. Figure 4-13

shows a map of transit system B routes and stops. This system is comprised of eight

local bus routes, two more than transit system B. Routes 1 through 5 remain

unchanged from transit system A, keeping the same route configuration and number of


128









stops. Route 6 differs from transit system A in that it is extended at its southern end to

join Route 1 and contains four additional stops. This was done to resemble the bus-

rapid transit cross-county connector route planned in the TDP. Two additional routes

were added to this system; Route 7 which resembles the configuration of the local bus

route proposed to connect Altoona and Zellwood, and Route 8 which resembles the

route of the commuter rail line planned from Zellwood to Eustis. The transit system

contained in the TDP contains two additional modes than those that were created in

transit system B; bus-rapid transit for Routes 1, 2 and 6, and commuter rail for Route 8.

In transit system B these modes are replaced with local bus service. Due to the time

and technical constraints of this study it was not possible to process the model using

those modes for the routes indicated. To compensate for some difference in the level of

service, the local bus service for these routes were enhanced with additional stops and

headway times were decreased.

Central Florida Regional Planning Model Model Run for Each Scenario

It should be noted that the CFRPM is a regional TDM that encompasses a ten

county area comprised of Brevard, Flagler, Lake, Marion, Orange, Osceola, Polk,

Seminole, Sumter and Volusia counties. When the CFRPM is run in this study it models

the whole region and it is necessary to extract the Lake County data from the model

results.

The CFRPM contains three transportation modeling scenarios, a base scenario

that models known data from 2000, and two future scenarios for 2012 and 2025 which

rely on forecast data. This study uses the year 2025 transportation scenario and

models all three future land use scenarios by altering the TAZ data (ZDATA1 and

ZDATA2 .dbf files) and transit route data (TROUTE.Iin) used as inputs into the model.


129







































Figure 4-13. Transit system B routes and stops. Note. Courtesy of author.


The model is first run for the FLUAM scenario as no changes to the model inputs

are required since the FLUAM scenario uses the existing TAZ and transit system data

of the year 2025 transportation scenario. The CFRPM produces a number of reports

that quantify the results generated by the different steps of the model. This study uses

results from the highway assignment report, specifically the total VMT, and the mode

split report which reports the number of highway and transit trips according to different

modes and trip purposes. The CFRPM creates reports in an HTML format. After the


130









model run is complete these reports are opened using a web browser and then copied

into a spreadsheet for further analysis.

The model is run a second time for the FLUAM scenario using the same TAZ data

(ZDATA1 and ZDATA2) but replacing the transit system A data (TROUTE) file with the

transit system B data. This is done so that any effect the different transit systems has

on VMT and mode choice can be ascertained. When the model run is complete the

highway assignment and mode split reports are copied into a spreadsheet for further

analysis.

Each LUCIS-plus scenario is run two times, both with their respective TAZ data

files created in the previous step of the methodology. One model run however uses the

transit system A data and the other uses the transit system B data. After each model

run the highway assignment and mode split reports are copied into a spreadsheet for

further analysis. The data from each of the six model runs is then summarized into a

single spreadsheet for the analysis of this study.


131









CHAPTER 5
FINDINGS

This study compares methods of forecasting future land use change to determine

what affect different methods may have on the achievement of long-range planning

goals. Four interrelated land use and transportation goals were selected for this study:

1. Discourage urban sprawl through a future land use pattern that promotes

orderly, compact development;

2. Increase energy conservation as measured in vehicle-miles of travel;

3. Reduce dependency on single-occupant vehicles; and

4. Increase transit ridership.

By comparing several data outputs from the different forecasting methods and the

transportation demand model that uses them, the extent to which each planning goal

was achieved can be measured.

Compact Development

The extent to which a land use pattern is compact can be measured according to

the number and density of people and employment located within specific zones within

a region. In this study the compactness of urban form is measured across two zones;

one that includes the area of land within a 3-mile radius of future transit routes, and the

other the area outside that distance. Figure 5-1 shows the two zones.

Table 5-1 shows the distribution of population and employment across the two

zones according to the scenarios created using the different forecasting methods. The

distribution of population and employment is also shown for TAZ data for the year 2007

for comparative purposes.


132













PUTNAM


FLACLER







MARION























SEMINOLE




SUMMER





ORANGE

HERNANDO

EFuture Transit Corridors

Future Transit 3 mile buffer
II






































TAZ's intersected by 3 mile buffer

TAZ's outside 3 mile buffer



0 4.5 9 Miles
POLK OSCEOLA







Figure 5-1. TAZs within and outside a 3-mile distance of future transit. Note. Map
courtesy of author.


133











Table 5.1 shows that the LUCIS-plus high scenario has the most compact land

use pattern compared to the other two scenarios. The LUCIS-plus high scenario has

60% of the study area population and 59% of employment located within a 3-mile

distance of future transit routes. The LUCIS-plus high scenario has 10% more

population living in the 3-mile zone than the LUCIS-plus low scenario and 11% more

than the FLUAM scenario. The LUCIS-plus high scenario also has 2% more

employment in the 3-mile zone than the FLUAM scenario and 5% more than the LUCIS-

plus low scenario.


Table 5-1. Population and employment distribution -
compared to 2007 TAZ data


Within 3-miles of transit
Single Family
Multi-Family
Buffers 1 to 4 Population
% of Total Population


Commercial


Service
Industrial
Buffers 1 to 4 Employment
% of Total Employment


Outside 3-miles of transit
Single Family
Multi-Family
Buffer 5 Population
% of Total Population
Commercial
Service
Industrial
Buffer 5 Employment
% of Total Employment


2007 TAZ
116,559
39,779
156,338
57%
20,478
45,688
12,669
78,835
70%


2007 TAZ
88,797
31,048
119,845
43%
7,808
16,381
9,542
33,731
30%


FLUAM
182,239
46,023
228,262
49%
31,865
49,353
17,488
98,706
57%


FLUAM
193,196
42,042
235,238
51%
17,128
39,615
16,288
73,031
43%


FLUAM and LUCIS-plus scenarios


LUCIS-
plus Low
95,881
134,287
230,168
50%
26,897
63,744
5,986
96,627
54%


LUCIS-
plus Low
132,507
93,840
226,347
50%
10,213
46,132
26,352
82,697
46%


LUCIS-
plus High
80,647
192,177
272,824
60%
30,224
70,502
5,990
106,715
59%

LUCIS-
plus High
101,870
83,149
185,019
40%
7,070
39,831
26,362
73,263
41%


134


~


--










The compactness of the land use pattern can also be measured in terms of the

density of population and employment in the two zones. Table 5-2 shows the average

TAZ population and employment densities within the different buffer zones shown in

Appendices E through H. The average TAZ densities were determined by calculating

the population and employment densities for each TAZ within a buffer zone and then

calculating the average of those densities. Table 5-2 shows that the highest average

population and employment densities within a 3-mile zone of future transit routes, that is

within buffers 1, 2, 3 and 4, exist in the LUCIS-plus high scenario. It also has the

lowest average population and employment densities outside of that 3-mile zone.

Table 5-2. Average TAZ population and employment densities by buffers
LUCIS-plus Low LUCIS-plus High
Population Densities FLUAM Scenario Scenario Scenario
people per ac people per ac people per ac
Buffer 1 & 2 TAZs 4.0 7.6 13.3
Buffer 1,2 & 3 TAZs 4.0 5.2 7.7
Buffer 1,2,3 & 4 TAZs 3.5 4.2 6.0
Buffer 5 TAZs 1.2 0.9 0.8
LUCIS-plus Low LUCIS-plus High
Employment Densities FLUAM Scenario Scenario Scenario
jobs per ac jobs per ac jobs per ac
Buffer 1 & 2 TAZs 4.0 1.7 5.1
Buffer 1,2 & 3 TAZs 3.0 1.5 2.8
Buffer 1,2,3 & 4 TAZs 2.4 1.4 2.2
Buffer 5 TAZs 0.4 0.3 0.3

Energy Conservation

Energy conservation can be measured by determining how many vehicle-miles of

travel (VMT) have been reduced by a particular land use pattern. VMT is a measure of

total traffic on a road and is calculated as the product of the average daily traffic count

and the length of the road (FDOT, unknown). In this study the three land use scenarios

were each modeled using two different transit routes. To ascertain the extent to which


135









the different transit systems may have reduced VMT the systems are compared across

the different land use scenarios. Table 5-3 shows that in terms of reducing daily

volumes of VMT there was little difference between the transit systems, with all

differences being less than two-tenths of a percent.

Table 5-3. Daily volume of vehicle-miles traveled in Lake County comparison of
transit systems by land use scenario
FLUAM LUCIS-plus LUCIS-plus
FLUAMI
Low High
Transit System VMTT VMTT VVMT
Transit System B 13,221,000 12,876,000 12,489,000
Transit System A 13,215,000 12,892,000 12,486,000
Difference between systems 6,000 -16,000 3,000
% change -0.05% 0.12% -0.02%


As the difference that the two transit systems made to VMT is only very slight the

remainder of the results will only display those where transit system B was modeled.

Table 5-4 displays a comparison of VMT across all of the scenarios and shows that the

LUCIS-plus high scenario generated the least amount of VMT. It resulted in 3% fewer

VMT than the LUCIS-plus low scenario and 5.5% fewer than the FLUAM scenario.

Table 5-4. Daily volume of vehicle-miles traveled in Lake County comparison of all
scenarios for transit system B
Transit
System B
Land Use Scenario VVMT
FLUAM scenario 13,221,000
LUCIS-plus low scenario 12,876,000
LUCIS-plus high scenario 12,489,000
% difference between FLUAM & LUCIS-plus low -2.6%
% difference between FLUAM & LUCIS-plus high -5.5%
% difference between LUCIS-plus low & LUCIS-plus high -3.0%


It is possible to illustrate the difference in VMT across the scenarios according to

the type of area in which the VMT is generated. Table 5-5 shows the VMT for the three

land use scenarios according to area type. Roads in the CFRPM are classified


136









according to area type and the VMT generated by area type roads is detailed in the

model's highway assignment report. The location of roads by area type used in the

modeling of transportation demand of all the scenarios is shown in Appendix N.

Rural areas create approximately 70% of all VMT in all scenarios and CBD areas

create the least with around 2% in all scenarios. Residential areas and outlying

business districts each account for between 10-12% of VMT across all the scenarios.

Table 5-5. Vehicle-miles traveled in Lake County by transit system B comparison of
LUCIS-plus low and FLUAM scenarios by area type
FLUAM LUCIS-plus LUCIS-plus
Scenario Low Scenario High Scenario
Area Type VMT VMT VMT
CBD 252,000 268,000 254,000
CBD Fringe 733,000 726,000 744,000
Residential 1,557,000 1,356,000 1,329,000
Outlying Business Districts 1,488,000 1,396,000 1,561,000
Rural 9,192,000 9,131,000 8,600,000
Total 13,221,000 12,876,000 12,489,000


A pair-wise comparison of the three scenarios demonstrates how VMT changed

across the various area types between the different scenarios. A comparison of the

FLUAM and LUCIS-plus low scenarios shows that the largest decrease in VMT

occurred in residential areas, followed by the outlying business districts and rural areas.

A small decrease (1%) occurred in CBD fringe areas and in CBD areas VMT increased

by 6.3%.

Table 5-6. Vehicle-miles traveled in Lake County by transit System B comparison of
the FLUAM and LUCIS-plus low scenarios by area type
FLUAM LUCIS-plus
Scenario Low Scenario %
Area Type VMT VMT Difference Change
CBD 252,000 268,000 16,000 6.3
CBD Fringe 733,000 726,000 -7,000 -1.0
Residential 1,557,000 1,356,000 -201,000 -12.9
Outlying Business Districts 1,488,000 1,396,000 -92,000 -6.2
Rural 9,192,000 9,131,000 -61,000 -0.7
Total 13,221,000 12,876,000 -345,000 -2.6


137











A comparison of the least and most compact scenarios, i.e. the FLUAM and

LUCIS-plus high scenarios (Table 5-7), shows that the greatest reduction in VMT of

occurred in rural areas, followed by the residential areas. In all other areas, VMT

increases.

Table 5-7. Vehicle-miles traveled in Lake County by transit System B comparison of
the FLUAM and LUCIS-plus high scenarios by area type
FLUAM LUCIS-plus
Scenario High Scenario %
Area Type VMT VMT Difference Change
CBD 252,000 254,000 2,000 0.8
CBD Fringe 733,000 744,000 11,000 1.5
Residential 1,557,000 1,329,000 -228,000 -14.6
Outlying Business Districts 1,488,000 1,561,000 73,000 4.9
Rural 9,192,000 8,600,000 -592,000 -6.4
Total 13,221,000 12,489,000 -732,000 -5.5


A comparison of the two LUCIS-plus scenarios shows that the more compact

scenario, i.e. LUCIS-plus high, has the greatest reduction in VMT in rural areas followed

by the residential and CBD areas. The LUCIS-plus low scenario however has 165,000

fewer VMT in outlying business districts than the high scenario.

Table 5-8. Vehicle-miles traveled in Lake County by transit System B comparison of
LUCIS-plus low and LUCIS-plus high scenarios by area type
LUCS-plus LUCIS-plus
LUCIS-plus
Low Scenario High
Scenario %
Area Type VMT VMT Difference Change
CBD 268,000 254,000 -14,000 -5.2
CBD Fringe 726,000 744,000 18,000 2.5
Residential 1,356,000 1,329,000 -27,000 -2.0
Outlying Business Districts 1,396,000 1,561,000 165,000 11.8
Rural 9,131,000 8,600,000 -531,000 -5.8
Total 12,876,000 12,489,000 -387,000 -3.0


138









Single-Occupant Vehicle and Transit Trips

A reduction in single-occupant vehicle (SOV) trips and an increase in transit trips

can be measured using outputs from the mode split report generated by the CFRPM.

The report shows the daily volume of trips according to trip types, either home-based

work (HBW) trips or non-home-based (non-HBW) work trips. HBW trips reflect peak-

hour trips and non-HBW off-peak hour trips (J.Banet, FDOT, personal communication,

January 14, 2010).

Table 5-9 shows the daily volumes of trips, including HBW (peak) and non-HBW

(off-peak) trips across all scenarios. In all scenarios the HBW trips account for 16% and

non-HBW trips 84% of total trips. The LUCIS-plus high scenario has the lowest number

of total trips of all the scenarios, 1.8% fewer than the LUCIS-plus low scenario and 7.0%

lower than the FLUAM scenario. It has 3% fewer total HBW trips than the LUCIS-plus

low scenario and 11% fewer than the FLUAM scenario. The LUCIS-plus high scenario

has 2% fewer total non-HBW trips than the LUCIS-plus low scenario and 6% fewer than

the FLUAM scenario.

Single-occupant vehicle (SOV) or drive alone trips account for the largest

proportion of all trips, making up 81% of HBW trips and 44% of non-HBW trips. Shared

ride car trips with 2 people account for 13% of HBW trips and 35% of non-HBW trips,

and shared rides with 3 or more people 6% of HBW trips and 20% of non-HBW trips.

The LUCIS-plus high scenario has the fewest SOV trips, 0.8% fewer than the LUCIS-

plus low scenario and 7% fewer than the FLUAM scenario.

Transit trips account for the smallest proportion of trips comprising only 0.004% of

total HBW trips and 0.08% of total non-HBW trips. The LUCIS-plus high scenario has

2% more HBW transit trips than the LUCIS-plus low scenario and 6% fewer than the


139









FLUAM scenario. It also has 3% fewer non-HBW trips than the LUCIS-plus low

scenario and 11% fewer than the FLUAM scenario.

Table 5-9. Daily volume of home-based work and non-home-based work trips in Lake
County by mode all scenarios
FLUAM LUCIS-plus LUCIS-plus
HBW trips Scenario Low Scenario High Scenario
Drive alone 232,261 213,288 208,990
Drive Shared ride 2 37,743 35,340 33,399
Drive Shared ride 3+ 18,130 17,226 15,173
Transit 242 212 255
Total HBW Trips 288,376 266,067 257,818

Non-HBW trips
Drive alone 652,172 615,256 613,123
Drive Shared ride 2 513,772 486,554 480,950
Drive Shared ride 3+ 295,137 289,163 275,565
Transit 33 58 56
Total Non-HBW trips 1,461,115 1,391,032 1,369,695

All trips
Drive alone 884,433 828,544 822,133
Drive Shared ride 2 551,515 521,894 514,349
Drive Shared ride 3+ 313,267 306,389 290,738
Transit 275 270 311
Total Trips 1,749,491 1,657,099 1,627,513


Summary

The most compact urban growth pattern was achieved using the LUCIS-plus

method. The LUCIS-plus high scenario had the most compact urban form with 60% of

total population and 59% of total employment existing in TAZs intersected by a 3-mile

buffer of future transit routes. Population and employment densities within that area

were 6.0 people per acre and 2.2 jobs per acre. The least compact scenario is created

using the FLUAM method, which had 49% of total population and 57% of total

employment existing in TAZs intersected by a 3-mile buffer of future transit routes.

Population densities within that area were lower than the LUCIS-plus high scenario,


140









being 3.5 people per acre however the employment density was slightly higher than the

LUCIS-plus high scenario, being 2.4 jobs per acre.

The greatest reduction in VMT was achieved using the LUCIS-plus method. The

study modeled two transit systems, one with two additional routes and higher frequency

of stops than the other. The difference in VMT between both transit systems was very

small, being less than two-tenths of one percent.

The LUCIS-plus high scenario generated the fewest VMT, 5.5% fewer than the

FLUAM scenario and 2.6% fewer than the LUCIS-plus low scenario. When the most

compact and least compact scenarios are compared, i.e. the LUCIS-plus high and

FLUAM scenarios, the greatest reduction in VMT occurred in rural areas followed next

by residential areas. The VMT in all other areas increased.

The greatest potential reduction in highway trips was achieved using the LUCIS-

plus method. The LUCIS-plus high scenario generated the fewest total trips, 1.8%

fewer than the LUCIS-plus low scenario and 7% fewer than the FLUAM scenario. The

fewest SOV trips are generated by the LUCIS-plus high scenario, 0.8% fewer than the

LUCIS-plus low and 7% fewer than the FLUAM scenarios. The LUCIS-plus high

scenario also generated the most transit trips, 13% more than the FLUAM and 15%

more than the LUCIS-plus scenarios.


141









CHAPTER 6
DISCUSSION

Discussion of Findings and Methods

This study aims to assess the affect methods of forecasting future land use have

on achieving long-range planning goals. The study uses several long-range planning

goals that can be measured to determine the affect the method of forecasting land use

change has on the achievement of the stated goals. The planning goals the study uses

are ones often cited in Smart Growth initiatives, goals that aim to combat sprawling

development patterns and decrease travel demand by encouraging the use of public

transportation to reduce dependency on SOV and conserve energy by reducing VMT.

The literature shows that urban form and travel demand are inextricably linked and that

several land use characteristics, such as density, diversity of land uses, and proximity to

transit, have the potential to increase the use of public transportation and reduce

automobile dependency (TRB, 2009). This study shows that allocating urban growth

near future transit routes using a method of forecasting future land use change based in

land use suitability analysis and scenario planning techniques created a more compact

urban form and achieved greater reductions in SOV and VMT than a method that used

a gravity model to allocate future growth based on historic development trends. The

following discussion highlights how this was achieved and why the results are relevant

for the improvement of coordinated planning of land use and transportation.

Compact Development

The study shows that the LUCIS-plus method creates a scenario with a

development pattern that is more compact than a scenario created with the FLUAM

method. This was in part due to the two zones of measurement the study uses, with


142









growth being measured either inside or outside of a 3-mile buffer of future transit routes.

The scenarios created using the LUCIS-plus method specifically target these zones in

allocating urban growth, whereas the FLUAM scenario concentrates growth around

existing and proposed activity centers regardless of their proximity to future transit.

While this difference creates some bias in the study it also serves to highlight an

important point about how the different land use forecasting methods have significant

consequences for the coordination of land use and transportation in Lake County.

The FLUAM uses a gravity model that assigns attractiveness factors to TAZs

based on their proximity to adjacent activity centers and the size of those activity

centers. This method emphasizes existing growth patterns by assuming that as activity

centers grow larger they will continue to attract growth, which is likely to be true to some

extent. However, the under-emphasis the gravity model places on slower-paced areas

of growth, such as those adjacent to future transit, is problematic unless the model

compensates for this when assigning attractiveness factors. The land use change

forecasted using the FLUAM appears to show that proximity to future transit does not

influence a TAZs attractiveness, resulting in a land use pattern that is less supportive of

the long-range transportation goals the study investigated.

The LUCIS-plus method uses land use conflict and suitability analysis to find

suitable locations for urban growth that specifically addressed the long-range planning

goals of the study. In both of the LUCIS-plus scenarios proximity to future transit routes

determined the extent to which the largest proportion of population growth would be

allocated. Land use conflict and suitability analysis used in the method, undertaken at

the small-scale of one-quarter acre areas, is able to accurately determine how many


143









acres of suitable land are available for growth so that different visions of varying

densities and intensities of land uses could be created and tested. The varying

population and employment densities of both the LUCIS-plus scenarios are illustrated in

maps shown in Appendices O through R. The result of using the LUCIS-plus method is

the creation of a land use pattern that is more supportive of the long-range

transportation goals under investigation than the FLUAM method. The effectiveness of

the LUCIS-plus method in quantifying, and representing the land use planning goal of

encouraging compact development improved the achievement of the long-range

transportation goals and ultimately could enhance the coordination of land use and

transportation plans for the Lake County.

Increased Transit Ridership

The results of this study are inconclusive in determining whether the method of

forecasting land use has an effect on increasing levels of transit ridership. The method

that creates the most compact development pattern, the LUCIS-plus high scenario

projects only a very small increase in actual transit trips. Two different transit systems

are modeled to show how transit improvements in conjunction with compacting growth

might increase transit ridership. However, little difference results between the two

transit systems that show either an increase use of transit or decrease in VMT. This

was mostly due to the fact that there was not a large difference between the systems

modeled. One of the two systems includes the visionary transit routes outlined in the

Lake County 2020 transit development plan, and should have included a commuter rail

line connecting north-western Orange County to central Lake County and a bus-rapid

transit (BRT) route crossing through central Lake County. Due to time and technical

constraints of the study however it was not possible to create a transit route that


144









incorporates those modes. Measures to compensate for this deficiency in one of the

transit systems, such as including bus services along the commuter rail and BRT routes

with more frequent and numerous stops, however did not result in any significant

differences between the two transit systems across any of the scenarios modeled.

It might be possible however to speculate what difference more transit trips, and

therefore fewer highway trips may have had on reducing VMT. If the visionary transit

routes could have been modeled and an increase in transit ridership occurred then a

corresponding decrease in highway trips would have also occurred. In the study a

122,014 decrease in highway trips occurred between the FLUAM (less compact) and

LUCIS-plus high (most compact) which corresponded to a decrease of 732,000 VMT.

This represents a decrease on average of 7.6 VMT with the reduction of a single

highway trip. If transit ridership increased ten-fold from the existing route when

modeling the visionary transit route, meaning an increase from 311 to 3110 transit trips,

then highway trips would decrease by the same amount (3110 trips) and a decrease of

23,636 VMT could be achieved. This reduction however only creates a 0.20% decrease

in overall VMT for the study area indicating that for transit ridership to have any

significant effect on reducing VMT a large increase in ridership is necessary. Such an

increase in transit ridership may be possible in Lake County however, where in 2000

25% of all workers, or 20,000 people traveled to Orange County to work. Assuming that

this number would increase by 2025 then opportunities could exist for significantly

increasing transit ridership and decreasing VMT if future land use change was planned

in a way that increased accessibility to transit, like that done in the LUCIS-plus high

scenario.


145









Evidence from the literature review undertaken by the TRB in 2009 indicates that

proximity to transit has a strong influence on transit use, as does the employment

densities at trip ends (p. 47). The LUCIS-plus scenarios modeled in this study used

proximity to transit and increased employment densities in areas adjacent to transit as

guidelines for allocating growth. It is possible therefore that had Lake County's

visionary transit system been modeled in conjunction with this type of future land use

configuration that larger increases in transit ridership could have been achieved and a

further reduction in VMT observed when forecasting future land use change using the

LUCIS-plus method.

Decreased Dependency on SOV

This study showed that the method of forecasting land use change did have an

effect on decreasing SOV use. The LUCIS-plus scenarios both decreased the number

of SOV trips compared to the FLUAM scenario, with the LUCIS-plus high scenario

achieving the greatest reduction in SOV trips overall. It should be noted however that

the decrease in SOV trips was due to the overall decrease in highway trips rather than

an increase in shared highway trips. The CFRPM allocates highway trip types (SOV

and shared trips) according to set percentages within the model with approximately 81%

of all highway trips being assigned as SOV trips, 13% as two-person shared trips and

5% as three or more person shared trips.

The reduction in overall highway trips between the LUCIS-plus high and FLUAM

scenarios did not come as a result of an increase in transit trips, as discussed above.

The reduction occurred as a result of fewer trips occurring between TAZs, as the

CFRPM only measures trips that are produced in one zone and attracted to another. It

is not able to measure intra-zonal trips. Therefore, a reduction in the number of


146









highway trips between the two scenarios is a result of more trip productions and

attractions being located within the same TAZ in the LUCIS-plus high scenario than in

the FLUAM scenario. The CFRPM uses the ZDATA1 table, which includes data on

residential population, to calculate trip productions for each TAZ, and the ZDATA2 table,

which includes employment data, to calculate trip attractions for each TAZ. The LUCIS-

plus high scenario resulted in more population and employment being located within the

same TAZs because the land use conflict and suitability variables used in the LUCIS-

plus method allowed specific locations to be found where opportunities for compact

growth such as mixed use and infill development exist where population and

employment could be more closely clustered together than in the FLUAM scenario.

The approach to allocating population and employment in FLUAM relies heavily on

the assumption that the location of existing development is appropriately located for the

achievement of long-range planning goals, such as the ones under investigation in this

study. It also assumes that the future land use element of the comprehensive plan, a

document with a relatively short planning horizon of only ten years, predicts the most

appropriate distribution of land uses to achieve long-range planning goals. However, as

noted in the literature review land use change typically occurs at a slow pace and

policies targeting changes to urban form, such as a more compact development pattern,

may take a long time to become a reality. Methods of forecasting future land use

change that use land use variables based on existing development patterns and

comprehensive plans may unintentionally continue short-term trends at the expense of

encouraging long-range goals that broader strategic plans seek to achieve. The


147









following discussion regarding the reduction of VMT between the FLUAM and LUCIS-

plus scenarios illustrates this point.

Energy Conservation Reduction in VMT

This study showed that the method of forecasting future land use change had an

effect on conserving energy by reducing VMT. The LUCIS-plus scenarios both

generated fewer VMT than the FLUAM scenario, with the LUCIS-plus high scenario

generating the fewest VMT of all the scenarios. VMT is calculated as the product of

average traffic counts by roadway lengths, so the fewer highway trips generated by the

most compact scenario, LUCIS-plus high, resulted in the greatest reduction in VMT. The

5.5% reduction in total VMT and 14.6% reduction in residential VMT between the

FLUAM and LUCIS-plus high scenarios are in line with conclusions from the TRB report

(2009) which suggest that a doubling of residential densities may be associated with a 5

to 12 percent decrease in household VMT (p. 4).

Comparison of VMT across the three scenarios by area type illustrates how

differences in forecasting methods resulted in different patterns of travel demand.

Although Table 5-1 shows that the FLUAM and LUCIS-plus low scenarios had similar

overall distributions of population and employment between the two zones of

measurement (within 3-miles of future transit routes, and outside the 3-mile distance),

the number of VMT generated in residential areas is significantly different (Table 5-6).

Appendix N shows that the majority of residential roads (69%) are located within 3-miles

of future transit routes so because the LUCIS-plus low scenario generated fewer VMT in

this area type it is must be due to more trip productions and attractions occurring with

the same TAZs within this area for this scenario. Table 5.2 shows that the LUCIS-plus


148









low scenario has higher population densities within this 3-mile distance of future transit

(buffers 1, 2, 3 and 4) that would account for more trip productions in residential areas.

Table 5-2 also shows, however, that the LUCIS-plus low scenario has lower

employment densities than the FLUAM scenario and Table 5-1 shows it also has less

employment allocated to this area. The mix of employment in each scenario for this

area however makes an important impact on the number of trips attracted. Although the

FLUAM scenario has more overall jobs in residential areas, the LUCIS-plus low has a

higher percentage of service jobs which attract more trips than commercial and

industrial jobs. The ability of the LUCIS-plus method to find suitable locations for more

service jobs in residential areas therefore results in more intra-zonal trips and a greater

reduction in VMT in residential areas than FLUAM was able to achieve.

Another important difference in VMT between the three scenarios can be seen in

the levels generated in rural areas. Table 5-6 shows that both the FLUAM and LUCIS-

plus low scenarios produce similar VMT in rural areas, both more than the LUCIS-plus

high scenario which had the largest reduction in VMT occur in rural areas (Tables 5-7

and 5-8). The reason for this difference lies in how the population and employment are

distributed between the two zones of measurement in LUCIS-plus high scenario

compared with the other two scenarios. Both the FLUAM and LUCIS-plus low

scenarios allocated population and employment into future developments of regional

impact (DRIs) located in buffer 5, the area most distant from future transit and which

contains the largest percentage of rural roads (see Appendix N). The LUCIS-plus high

scenario however does not allocate into DRIs in buffer 5. After allocating population

and employment into buffers 1 through 4 the remainder of unallocated population and


149









employment was too low for the projected growth in DRIs in buffer 5 to be allocated in

the LUCIS-plus high scenario.

The LUCIS-plus method was able to allocate higher levels of population and

employment into suitable locations in a more compact development pattern than the

FLUAM method. The FLUAM scenario assumes that DRIs in buffer 5 will be developed

because Lake County's future land use plan dictates this development pattern and as a

result this method was unable to forecast a land use pattern that reduced VMT in the

areas where the greatest proportion is created, Lake County's rural areas. Similarly the

LUCIS-plus low scenario assumed that buffer 5 DRI's would be developed and it too

failed to significantly reduce VMT in rural areas when compared to the more compact

LUCIS-plus high scenario.

Regional Accessibility and the Challenges of DRIs

The challenges that DRIs in distant rural locations creates for the coordination of

land use and transportation are illustrated by this study. Ewing and Cervero (2001)

noted in their meta-analysis of literature reviews on the impacts of the built environment

on VMT that overall regional accessibility is the main driver of vehicle travel demand.

This study concurs with this finding, as the more compact development pattern created

in the LUCIS-plus high scenario produced residential areas that had greater

accessibility to employment opportunities which resulted in fewer highway trips and

greater reduction in VMT by 5.5% compared with the FLUAM scenario. The FLUAM

and LUCIS-plus low scenarios both contained regionally isolated DRIs in rural areas

that resulted in more VMT in these areas. If significant proportions of future population

growth are permitted to occur in regionally isolated areas in Lake County the population


150









and employment densities needed adjacent to transit routes would not be achievable for

a significant increase in transit ridership and reduction in SOV and VMT.

Proximity to Transit and Employment Densities

The literature review undertaken by the TRB (2009) reported that proximity to

transit and employment densities at trip ends has a stronger influence on transit use

than urban design features such as walkability and land use factors such as mixed uses

and increased residential densities. Although this study was unable to show an

increase in transit ridership the LUCIS-plus method was able to allocate greater

proportions of population and employment in closer proximity to future transit than did

the FLUAM method, and the result was a reduction in VMT due mostly to residential

and employment land uses in closer proximity to each other. If a transit route that truly

reflected the planned future transit system for Lake County was modeled it is likely

greater reductions in VMT would have occurred as higher employment densities

proximal to transit routes was achieved using the LUCIS-plus method.

Criticisms of the FSM and the CFRPM

Some criticisms of the FSM cited in the literature review are evident in this study,

as the CFRPM follows the same four-step process of modeling transportation demand.

The policy insensitivity of the FSM is apparent in this study in the way that the CFRPM

is unable to capture the potential impacts that TOD has on intra-zonal trips. This is

largely a result of another criticism of the FSM; that it relies on zonal level data. If TAZs

are too large in areas where micro-level change is being attempted, such as in areas

targeted by policies encouraging transit use and TOD then important intra-zonal activity

may not be captured by the FSM model. For example, while the FSM shows that fewer

trips may occur in TOD areas because more productions and attractions are occurring


151









with the same TAZ, it is also possible that a higher level intra-zonal travel is also

occurring for the same reason but the FSM is unable to account for it.

Complex Land Use Models vs. Deterministic Approaches to Forecasting Land
Use Change

The literature review showed that a recent trend has existed in land use modeling

that favors more complex disaggregated simulations models over simpler deterministic

and rule-based approaches to forecasting future land use change. While the FLUAM

uses a gravity model and not a disaggregated simulation model, it is based in economic

theories about land use that are mathematically calculated giving the model an

appearance of being more scientific than perhaps a deterministic approach such as the

LUCIS-plus method. A criticism of deterministic approaches is that they are too

subjective and not grounded in theory, however as Handy (2008) points out complex

and more scientific approaches to forecasting land use change and transportation

demand also incorporate subjective measures often in ways that are less apparent to

the non-technical user of the forecasts. The scientific validity often afforded to complex

forecasting methods sometimes results in them being seen more as predictions rather

than simply as possible futures, which all forecasts inherently are. When a forecast is

treated as inevitable because it was scientifically derived the planning process suffers

because equally possible forecasts may be overlooked.

The criticism that deterministic approaches to forecasting land use change are too

subjective to be useful belies one of the greater strengths of such approaches their

ability to envision different futures based on a variety of different inputs and viewpoints.

While the LUCIS-plus method used in this study focused on creating future land use

patterns that supported transportation objectives, it could easily have been used to


152









focus on agricultural preservation or environmental conservation, or hazard mitigation,

or a combination of all of these viewpoints. The strength of methods such as LUCIS-

plus lies not in that they forecast futures that will happen but instead on futures that

could happen allowing a wide variety of possible alternative to be weighed and

considered in the planning process. For the successful coordination of land use and

transportation the ability of a land use forecasting method to envision a future rather

than predict one is particularly useful, as the results of this study demonstrate.

Study Limitations

As noted earlier this study was not able to model the visionary transit route

outlined in Lake County's 2020 transit development plan due to time and technical

constraints of the study. Due to this limitation the study was not able to comment about

whether the compact development created by the LUCIS-plus method had any impact

on the transportation goal of increasing transit ridership. An opportunity for further

research exists if a TROUTE.lin file for the CFRPM reflecting the visionary transit route

was available or could be created, and this study's methodology repeated to determine

what effect the transit route makes to the achievement of the transportation goals

Another limitation of the study involves the suitability rasters used in both the land

use conflict and suitability analysis used in the combine raster, upon which the LUCIS-

plus scenarios were created. As noted earlier the conflict and suitability rasters were

originally created for a different project involving the central Florida region. The

visionary transit route for Lake County was not used in the creation of any of these

suitability rasters. GIS vector data for regional major roadways was included in the

creation of the suitability rasters however, and the visionary transit routes, for the most

part, follow these same roadways. Nonetheless, had the visionary transit routes data


153









been included in the suitability models that dealt with proximity to transportation, it

would have likely been weighted more highly than major roadways which would have

resulted in areas adjacent to transit receiving a higher suitability than they did in the

absence of this data. As a result it is likely that more areas that were favorable to TOD

would have existed than the suitability rasters used in this study were able to show. An

opportunity exists for further research into the effect proximity to future transit routes

has on the transit ridership by testing new suitability rasters that include this data

Both the FLUAM and LUCIS-plus scenarios use population estimates that were

based on 2000 Census data, or other Census estimates which also used the 2000 data.

Considering the age of this data, many assumptions were made about how population

has actually changed since that time, which must be considered a study limitation,

although a largely unavoidable one. An opportunity for future research exist once the

2010 Census data becomes available to do a study similar to this one using the latest

available data, and to compare the results of the two studies to see how they are the

same or different.


154









CHAPTER 7
CONCLUSION

Lake County's long-range transportation plan contains a public transit system that

in the next 15 to 25 years is to include a commuter rail line that links eastern Lake

County to western Orange County. It also includes a cross-county bus-rapid transit line

moving through central Lake County connecting west and east. Apart from increasing

transit ridership within Lake County the long-range plan is to capture an increasing

number of choice riders, those traveling to work, particularly those traveling to Orange

County, where in 2000 one in four Lake County residents were employed.

Research into the impact of urban form on transit use shows that proximity to

transit is an important factor in attracting transit users. An important question Lake

County planners must therefore ask is whether the future urban form their

comprehensive plans prescribe is one that will support the transit system planned in the

next two decades. Considering the slow pace at which land uses change an answer to

this question is highly pertinent to long-range plans presently under development.

Anticipating the urban form needed to support these long-range transportation goals

and beginning to encourage development in appropriate locations makes good sense.

Incremental and disjointed land use change however has made the task of

forecasting future land use patterns that support public transportation difficult in Florida.

History and research into the relationship between land use and transportation show

that as urban development patterns disperse, and land uses become more separated,

the need for mobility increases. In Florida particularly and the United States in general,

where personal mobility is highly reliant upon the automobile, a dispersion of urban

growth most often leads to increased investment in road infrastructure as a remedy for


155









the increased and dispersed travel demand forecasted. Over time regional trips grow in

length as distances between activity centers increase and higher levels of VMT occur in

areas that are more sparsely populated; a pattern of development which is not

conducive to the support of a public transit system. Such a scenario is demonstrated in

this study of Lake County which has shown that the largest volume of VMT occurs daily

in the more distant rural areas of the county where population densities are the lowest.

The aim of this study is to show how methods of forecasting future land use

change can impact the achievement of long-range planning goals. It put forth the

hypothesis that a forecasting method based in land use suitability analysis and scenario

planning techniques could improve the achievement of mutually supportive land use

and transportation goals, more so than a traditional forecasting method based on

historical development trends and a gravity model. The contention is that land use

suitability analysis used as a forecasting tool can precisely locate areas for future

growth that meet specific planning guidelines designed to enhance the coordination of

long-range land use and transportation goals. Various scenarios can be created that

meet the planning guidelines to varying degrees and each can be tested by modeling

their transportation demand to determine which scenario best reflects the achievement

of all the desired goals. The results of this study show that the LUCIS-plus method was

better able to reflect the long-range land use and transportation goals that this study

investigated, more so than the FLUAM method that has been historically used in

forecasting Lake County's transportation demand.

The value of this research is two-fold. Firstly, it highlights the difficulty of

coordinating long-range land use and transportation plans when future land use


156









forecasting methods are based on short-term planning policies such as comprehensive

plans. Because future land use elements of comprehensive plans typically forecast

land use for a ten-year period the risk of replicating short-term development patterns is

very likely when forecasting methods use this land use data as a main indicator of future

growth. The degree of land use change needed in places like Lake County to support a

public transit system will take place over a much longer period of time than the ten

years addressed in local comprehensive plans. The cumulative effects of incremental

land use change on transportation systems may not be adequately anticipated when

future land use forecasting methods are based on short term policies, making it difficult

to plan the land uses needed for longer range investments such as public transit. The

LUCIS-plus method offers a positive alternative to addressing this problem since it

anticipates future land use change based on suitability guidelines of how future land use

should be configured to meet desired planning goals, rather than following a short-term

future land use map. A major strength of the LUCIS-plus method is its ability to both

allocate land uses at a small-scale for in-depth analysis and aggregate land uses to a

macro-scale so that change needed to support long-range goals can be regionally

envisioned.

This study has also been valuable in highlighting the importance of regional

accessibility in the coordination of long-range land use and transportation plans, and in

demonstrating how methods of forecasting future land use change can either aid or abet

this process. Large-scale regionally isolated land developments have a considerable

impact on a region's ability to attract higher density and more compact development

adjacent to future transit systems and the levels of ridership needed to reduce VMT. In


157









this study two of the three land use scenarios modeled each contained two DRIs

located more than 3 miles from future transit routes. Both scenarios had higher vehicle

trips and VMT than the more compact scenario. Despite the fact that both of the DRIs

observed Lake County's prescribed design guidelines of increased densities and mixed

land uses, their regional isolation negated any reductions in VMT that were gained from

those design measures. As a result of the LUCIS-plus method used in this study, the

impact these isolated DRIs had on the region's ability to support its long-range goal of

public transit was made apparent.

As an analytical tool the LUCIS-plus method has much potential in estimating the

regional impacts of DRIs and smaller incremental changes on future travel demand. A

continuing goal of the LUCIS and LUCIS-plus research currently being undertaken at

the University of Florida is the amalgamation of all the LUCIS tools into a unified model

able to be run at a variety of scales, where data inputs and model parameters are

modified by user-friendly GUIs. These enhancements will enable non-expert GIS

planners to utilize the LUCIS-plus method to create scenarios for comparative analysis

and further modeling within other planning frameworks such as transportation and

hazard mitigation planning.

The relative ease and flexibility that the LUCIS-plus method brings to the

assessment of regional land use plans offers many potential benefits to regional and

state planning decision makers, particularly with regard to the approval of

comprehensive plan amendments. If a state-wide and regional LUCIS models were

available then the cumulative impacts of individual plan amendments could be more

easily envisioned, providing valuable input into the approval process. In light of the


158









current state of affairs in Florida with regard to the over-allocation of land uses in many

local comprehensive plans this kind of analytical input would be very helpful in guiding

decision makers.

According to Interim Report 2010-107 of the Florida Senate Committee on

Community Affairs (2009) there is controversy surrounding the issue of "needs

assessment" in the determination of comprehensive plan amendments (p. 1). The

"needs assessment" rule is used to determine whether a proposed amendment is

required to meet projected future population growth. Due to an uneven application of

the "needs assessment" rule by the Department of Community Affairs large amounts of

land have been assigned to future land uses far in excessive of what projected

population growth requires (p. 8) and have resulted in urban sprawl in many counties.

Changing the land use designation of many of these inappropriately located

developments would be difficult due to an interference of private property rights, and

their existence increases the difficulty of approving more appropriately located

developments. Report recommendations for resolving these issues include a clearer

articulation of the "needs assessment" test as it relates to the goals for planning growth.

Specifically these goals are the discouragement of urban sprawl and the efficient use of

infrastructure spending, the prevention of the fragmentation of the environment; and the

promotion of coordinated plans among adjacent local governments (pp. 8-9). The use

of the LUCIS-plus method to translate interrelated planning goals into quantifiable and

comparable regional scenarios was demonstrated in this study and could well be

applied to the "needs assessment" test used by the Department of Community Affairs in

their approval process for comprehensive plan amendments.


159









Prior to the commencement of this study in January 2009, the LSMPO began work

on updating their LRTP for the period 2015 to 2035. As part of the update process

consideration was given to the methods used to forecast future land use change. After a

comparison of the FLUAM and the LUCIS-plus methods it was decided in May 2009

that a future land use scenario derived using the LUCIS-plus method would provide the

socio-economic data for modeling the future transportation demand needed for their

2035 LRTP update. This recognition by the LSMPO of the merits of the LUCIS-plus

method helps raise awareness among other regional planning organizations and state

planning agencies that analytical tools now exist that can improve the coordination of

long-range planning initiatives at local, regional and state levels.


160










APPENDIX A
LAND USE MODELS REVIEWED BY AUTHOR AND MODEL TYPE



Miller et Zhao &
al. EPA Chang Chung
Model Name Model Type (1998) (2000) (2006) (2006)
DRAM/EMPAL/ITLUP
METROPILUS Spatial Interaction Y Y Y Y
HLFM 11+ Spatial Interaction N N N Y
LILT Spatial Interaction N N Y Y
LUTRIM Spatial Interaction N N N Y
DELTA Spatial Input/Output N Y N Y
MEPLAN Spatial Input/Output Y Y N Y
TRANUS Spatial Input/Output Y N N Y
CUFM: CUF-1/CUF-2 Rule-Based N Y N Y
SAM/LAM/SAM-IM Rule-Based N Y N Y
SLAM Rule-Based N N N Y
ULAM Rule-Based N N N Y
UPLAN Rule-Based N Y N Y
WHAT IF? Rule-Based N Y N Y
Random Utility/
METROSIM Discrete Choice Y Y Y Y
INDEX Other N Y N Y
LUCAS Other N Y N Y
Markov Model of
Residential Vacancy Other N Y N Y
SMART PLACES Other N Y N Y
IRPUD Micro-Simulation N Y N Y
MASTER Micro-Simulation N N N Y
NBER/HUDS Micro-Simulation N N N Y
UrbanSim Micro-Simulation Y Y N Y
Mathematical (Linear)
Herbert-Stevens Programming N N Y Y
Mathematical (Linear)
POLIS Programming N N Y Y
Mathematical (Linear)
TOPAZ/TOPMET Programming N N Y Y
SLEUTH Cellular Automation N Y N Y
MUSSA Bid-Rent Y N Y N


161











APPENDIX B
STRENGTHS AND LIMITATIONS OF LAND USE MODEL TYPES BY AUTHOR

Author Miller et al. (1998)
Model Type Strength Limitation
over-reliance on equilibrium no land supply
Spatial Interaction history of validation or prices spatial distribution relies on
exogenous data lacks spatial precision
framework sufficiently flexible to include
chain of demand over-reliance on equilibrium over-reliance
S tial I t/Out t (land-4buildings-activity-floor space, on trip generation elasticity to fit trip lengths *
paal npuupu density, price etc.) fully endogenous use of aggregate input/outputs restricts zone
prices explicit representation of transit size
services some validation
Rule-Based N/A N/A
Random Utility / consistent use of micro-economic ov e n iliri
over-reliance on equilibrium
Discrete Choice concepts uses traffic zone level
Micro-Simulation N/A N/A
Linear Programming N/A N/A
Cellular Automation N/A N/A
theoretically rigorous and complete *
uses traffic zone level and finer
u s traffic z e l l ad finer over-reliance on equilibrium static model
Bid-Rent resolution possible has been validated
that must be run by time step
resulting in forecasts with good face
validity
Author EPA (2000)
Model Type Strength Limitation
Spatial Interaction robust model type most often used by focuses on aggregate choice rather than
metro areas ability to introduce individual choice behavior little scope to
constraints or other influences (to introduce planning policies other than zoning *
account for local knowledge) no mechanism for simulating land-market
clearing process limited independent
variables may lead to underestimation of
infrastructure investments
Spatial Input/Output flexible design to meet needs of user static model validation potentially
(no rigid input requirements) can be problematic depending on level of observed
used for regional analysis allows data used in base year
analysis of different kinds of policies
Rule-Based flexible requirements, customizable and lacks modeling sophistication and theoretical
easy to use useful for simulating basis to examine interrelated factors such as
alternative future development scenarios transportation, fiscal and planning policies that
affect land use change
Random Utility / grounded in economic theory None stated
Discrete Choice recognition of how market forces shape
and change land use
Micro-Simulation can simulate decision making process None stated
of individual can reflect real-world
processes dynamically in time and space
high degree of spatial resolution
Mathematical (Linear) N/A N/A
Programming
Cellular Automation allows for relatively simple alternative unable to deal explicitly with population,
scenario projection can simulate policies and economic impacts on land use
temporal booms and busts change
Bid-Rent N/A N/A


162











Author Chang (2006)
Model Type Strength Limitation
conceptually simple but fails to represent unique characteristics of
comprehensive can include a variety location lacks behavioral interpretation -
Sof locations ability to represent aggregates trends w/out social or income
diverse transport components disaggregation
Spatial Input/Output N/A N/A
Rule-Based N/A N/A
Effectively addresses vocational aggregation bias unsatisfactory in addressing
responses between transport and land use
Random Utility / characteristics, including unobserved ( responses between transport and land use
Discrete Choice endogenouss vocational accessibility subject to
Discrete Choice high degree of behavioral validity of transport system one-directional instead of
transport system one-directional instead of
individual's decision-making process mutually determined).
Micro-Simulation N/A N/A
simple mathematical form combined
models able to produce endogenously gives little consideration to unique
Linear Programming determined transport costs; impedance characteristics of location fails to represent
generated by mutual adjustment decision-making process of an individual
between land use and transport
Cellular Automation N/A N/A
effectively represents unique
characteristics of locations by using does not capture the explicit interaction
Bid-Rent hedonic theory rational framework
describes the behavior of decision
making process of consumers
Author Zhao & Chung (2006)
Model Type Strength Limitation
easy to use population and lacks behavioral factors influencing choice -
Spatial Interaction employment distributed as a function of land market and prices not considered spatial
location attractiveness and travel cost detail limited
addresses regional spatial patterns of
location of economic activities real
location of economic activities real generates static equilibrium solution to changes
Spatial Input/Output estate and labor markets are in oneon to changes
in one or more inputs
considered travel demand part of
modeling process
useful for long-range scenario testing
Useful for long-range scenario testing rules not comprehensive or flexible enough to
easy to apply and data generally
Rule-Based e asy to apply and data general model complex economic and market processes
available based on market rules and in detail
in detail
economic theories
firmly rooted in economics *
Random Utility / recognizes market forces deals None stated
Discrete Choice explicitly with land use policy and land
use change
simulates the behavior of individuals
which are aggregated to form the
overall behavior of a system sensitive
Micro-Simulation to behavioral responses to changes in None stated
transport system or land use policies
because it models the decision-making
process of individuals
more manageable, understandable & does not realistically describe behavioral
*more manageable, understandable &
responses to changes in transport system or land
Linear Programming computationally easier than other r s to c s i t s o
optimization techniques use policies behavior and uncertainty difficult to
optimization techniques model
model
useful for representing interactions
Useful for representing interactions abstract representation of agents, decisions and
Cellular Automation between a location and its immediate decisis
surrounds
Bid-Rent N/A N/A


163










APPENDIX C
DATA DICTIONARY FOR LUCIS-PLUS SCENARIOS


Data Description Type Source
Population & household estimates Lake County 2007 Tabular Census Bureau
Tabular and
Traffic analysis zone data Lake County 2000 and 2012 Tabular FDOT
GIS vector
Thompson -
Tabular and projected using
Traffic analysis zone data Lake County 2007
GIS vector FDOT 2000 & 2012
TAZ data
Average of medium and high population projections Tabular BEBR
Lake County 2025
ECFRPC created
Employment projections Lake County 2005, 2010 Tabular using REMI Policy
and 2025 Insight Model
Florida Geographic
Parcel data Lake County 2007 GIS vector Florida Geographic
Data Library (FGDL)
Hydrography Lake County lakes and ponds GIS vector FGDL
Florida Managed Lands Lake County conservation
GIS vector FGDL
areas 2009
Tabular and
Developments of regional impact Tabular ECFRPC
GIS vector
LUCIS agriculture, conservation and urban suitabilities GIS raster Dr. P. Zwick UF
GIS raster Dr. P. Zwick UF
Lake County created in 2008-2009
LUCIS land use conflict Lake County created in
GIS raster Dr. P. Zwick UF
2008-2009
Visionary transit lines and stations for 2035 Lake GIS vector ECFRPC
GIS vector ECFRPC
County
School attendance zones Lake County 2008-2009 GIS vector FGDL


164









APPENDIX D
ZDATA1 AND ZDATA2 DESCRIPTIONS

ZDATA1
Attributes Description
TAZ Zone number
SFDU Single family dwelling units
% of single family dwelling units that are vacant or occupied by non-
SF PCTVNP
permanent residents
SF_PCTVAC % of single family dwelling units that are vacant
SFPOP Single family population
SF_OAUTO Single family dwelling units with 0 automobiles
SF_1AUTO Single family dwelling units with 1 automobiles
SF_2AUTO Single family dwelling units with 2 automobiles
MFDU Multi family dwelling units
% of multi family dwelling units that are vacant or occupied by non-
MF PCTVNP
C- permanent residents
MF_PCTVAC % of multi family dwelling units that are vacant
MFPOP Multi family population
MF_0OAUTO Multi family dwelling units with 0 automobiles
MF_1AUTO Multi family dwelling units with 1 automobiles
MF_2AUTO Multi family dwelling units with 2 automobiles
HMDU Hotel/motel dwelling units
HMOCC % of Hotel/motel dwelling units occupied
HMPOP Hotel/motel population
ZDATA2
Attributes Description
TAZ Zone number
INDEMP Industrial employment
COMEMP Commercial (retail) employment
SEREMP Service employment
TOTEMP Total employment
SCHENRL School enrolment
Note. Adapted from "FSUTMS Powered by CUBE/VOYAGER Data Dictionary" by FDOT, 2005,
p. 8.


165












APPENDIX E
ALLOCATION BUFFERS 1 AND 2


Buffers 1 and 2
1/4 mile surrounding
Future Transit Stations


Data Descriptions
Buffers 1 and 2 represent a zone with a 1/4 mile
radius from future transit stations. This is generally
considered a comfortable walking distance and highly
suitable for transit-oriented development.
Buffer 1 represents areas within that zone that are
a distance of approximately 400 feet of major roads.
The adjacency of these areas to traffic makes them
preferable for higher density mixed-use development.
Buffer 1 and 2 TAZs are those traffic analysis zones that
are intersected by Buffers 1 and 2


Note: Map courtesy of author.


166












APPENDIX F
ALLOCATION BUFFER 3


Buffer 3
Commercial Corridors within
3 miles of Future Transit



Data Descriptions
Buffer 3 represents a zone of 1/2 mile wide commercial
corridors that are within a distance of 3 miles to future
transit lines. Areas identified in Buffer 1 & 2 are excluded.
The commercial corndors selected were those identified in
the Lake County 2025 Comprehensive Plan Future Land Use
Map as major commercial corridors.
Buffer 3 TAZs are those traffic analysis zones that
are intersected by Buffer 3 and include Buffer 1 and 2 TAZs.


Note. Map courtesy of author.












APPENDIX G
ALLOCATION BUFFER 4


Buffer 4
Traffic Analysis Zones within
3 miles of Future Transit


Data Description
Buffer 4 represents a zone that is 3 miles from future transit
lines and excludes those areas already identified in
Buffers 1, 2 and 3.
This buffer represents a reasonable distance that
potential transit users might drive to access
park-and-ride and kiss-and-ride facilities.
Buffer 4 TAZs are those traffic analysis zones
intersected by Buffer 4 and includes Buffer 1, 2
and 3 TAZs.


Note. Map courtesy of author.


168












APPENDIX H
ALLOCATION BUFFER 5


Buffer 5
Traffic Analysis Zones outside
of Buffers 1, 2, 3 and 4






Data Description
Buffer 5 includes the traffic analysis zones that are outside
of Buffers 1, 2, 3 and 4.
This buffer represents areas within Lake County that are
greater than 3 miles from future transit. Access to transit
diminishes as locations become more distant, and these
areas are considered to be less attractive to potential
transit users.


Legend
Future Transit Stations o
Major Highways -
Future Transit Lines -
Buffer 5 [

I I I Miles A
0 5 10


Note. Map courtesy of author.


169











AGGREGATION OF FDOR


APPENDIX I
LAND USE CODES IN THE PARCEL DATA TO TAZ LAND
USE CODES


Land Uses Land Uses
FDOR TAZ FDOR TAZ
Vacant residential VACRES Grazing land IND
Single family SF Poultry, bees, fish, rabbits IND
Mobile homes MF Dairies, feed lots IND
Multi-family MF Orchards, groves, citrus IND
Condominia MF Ornamentals, misc. agriculture IND
Cooperatives MF Mining, petroleum, gas lands IND
Multi-family < 10 units MF Subsurface rights IND
Vacant commercial VACCOM Sewage disposal, borrow pits IND
Stores one-story COM Retirement homes SER
Mixed use (store/office) COM Boarding homes SER
Department stores COM One-story non-prfssnl offices SER
Supermarkets COM Multi-story non-prfssnl offices SER
Regional shopping malls COM Professional service buildings SER
Community shopping centers COM Airports, marinas, terminals SER
Restaurants, cafeterias COM Financial institutions SER
Drive-in restaurants COM Insurance company offices SER
Repair service shops COM Camps SER
Service stations COM Golf courses SER
Auto repair, service, sales COM Hotels, motels SER
Parking lots, mobile home sales COM Churches SER
Florist, greenhouses COM Private/public schools, colleges SER
Drive in theaters, open stadiums COM Private/public hospitals SER
Encl. theaters, auditoriums COM Homes for aged SER
Night clubs, bars, lounges COM Mortuaries, cemeteries SER
Tourist attractions COM Clubs, lodges, union halls SER
Race horse, auto, dog tracks COM Sanitariums, convalescent SER
Vacant Industrial VACIND Cultural organizations SER
W/sale manufacture/processing IND Military SER
Light and Heavy Manufacturing IND Forest, park, recreation areas SER
Lumber yards, sawmills IND Other county/state/federal SER
Fruit, veg and meat packing IND Gov. owned and leased SER
Canneries, distilleries, wineries IND Utilities SER
Other food processing IND Outdoor recreational SER
Mineral processing IND Acreage not zoned for agric. VAC
W/houses, distribution centres IND Vacant Institutional VACCOM
Industrial storage (fuel, equip) IND Undefined No use
Improved agriculture IND Rights-of-way, streets, roads No use
Cropland IND Rivers, lakes, submerged land No use
Timberland IND Centrally Assessed No use


170










APPENDIX J
AGGREGATION OF ECFRPC EMPLOYMENT CATEGORIES TO TAZ LAND USE
CODES

ECFRPC Employment Category TAZ Land Use Code
Wholesale trade COM
Retail trade COM
Forestry, fishing, other IND
Mining IND
Construction IND
Manufacturing IND
Farm IND
Utilities SER
Transportation, warehousing SER
Information SER
Finance, insurance SER
Real estate, rental, leasing SER
Professional, technical services SER
Management of companies SER
Administration, waste services SER
Educational services SER
Health care, social assistance SER
Arts, recreation, entertainment SER
Accommodation, food services SER
Other services (excluding government) SER
State and local governments SER
Federal civilian SER
Federal military SER


171













APPENDIX K

LAKE COUNTY 2009 2020 TRANSIT DEVELOPMENT PLAN MAP AND
DESCRIPTION OF MODES AND ROUTES


1.10 LXlo. 1 Cou ComrvCiomcr (Opwd an is ain 2012) ial Rsa.
1320 LX Roue 2 -IAtrng CiItlar (OpBedn a isil 2012) Fiaid Ro
1.30 LX a u* 3 Mwa= DWIa Ckcxlaf (Opuzn a is VM 2012) Fe= Rom
1.40 ZELLWOOD CONECTO(GRANT 2009) hald I=
111 sv L% Roao* 1 C Couy Coaw (SaIm i 2012) ied Rom
S1-21 SLE ESBTR RmTLAN)D PARK CIRCULATOR Cicuata
131 GODN TRIAMNGLE CIECULATOR Cizcalam
141 ELLWOOD CCWNNCTOR AM H RW Firmd Io=0




____ ( AA(i TSwgrc "2 o~ t~tmrsscij~~fiii s.i(^iw 0u-iiC


Source. "Lake County Transit Development Plan 2008 Major Update"
Associates, 2008, p. 9-26 and p. 9-29.


by Wilbur Smith


172


ALTERNATIVE 1


Legend
AJZRIATs COaMlD .

C-ROS OU 4NT OMllCl TOi uNdv,. CbnwdUnM. Sd.lI
U5ilHUA OMCIAATDR (ORCLited Bnds a |


COMtONA LTZANArL a CiaM S=dtd
S 2LVLWooC C O nt. O .e te "d .epd 9on. RLXtS I
_coummon__ruarwascaInm
LrME IS a417


~uu, ulugvr~ rut:







~'
"`"''"'
""'" `""'












APPENDIX L
TOD DESIGN STANDARDS USED AS STUDY GUIDELINES


T3


bn.. Dws.. Sksk .

uItal Deny Dwelung tUkA per Are 3 to 30 Dwe0 g Unlhs/Ac.

Populio Damily Paem per Aper I 15 180 Pr0iom/Aaw

InplomrHeem y ly Eo per Acr so 40 Jotb/Acr

h. m ou/Duowa olfz u

Mn Moor Am* lRafo PAM 2.0 3.0

Mnaume Wn wd tal Devty 35 to 60 Dwilng khil/Acr

A**t- ling Bd i 0l006 3 or nwe Storm

"I-M to& r oCta 80%

wmen... are" hws 8D%
bMoinmn ftrunng0 80%


Mauinw iaR lal Parlan Spcuts per Uni) 2 SpacktA*t
Manuum OH/elal Prlaang iSpo per I0OO 3 ces/1,000 .ft

Mane Surte Piulp f(% Of TOd Spoe0) 20%

SEd 'vs. Single*4 Padrg Pod, y Shared

Pa4k &ld and ua Yes

MkImd Use A D0out
M041001 Houm of 'S1miNcW AdMy I 4Hamun

Awrag. Jobtx/Hu- Wft.o I Jobs: I DOwllIg A

Mix of Uses 1% 20d i %) NoPe4dku0 70% ienlderabal and 30% iNoo.sidewial


Grid Dmiry Potygonm per Square Mk) brycl. MM c
PudAm and bout Mawsk
Average Sloc& S (t POe) 200' 800x




$V t We son
V. I 9 1 1%0%0
"" Si t~i
II~
q! hIrrrr


-- -


ommI .N q utmANNurama mrAr n89,


I Iawul I I .l l

Source. "Transit Oriented Design Guidelines" by Florida Department of Transportation, 2009b,
p. 9.


173











GENERAL SELECTION


APPENDIX M
CRITERIA FOR ALLOCATION INTO BUFFERS 1 TO
SCENARIOS


5 FOR LOW AND HIGH LUCIS-PLUS


Buffer Selection 1 Selection 2 Selection 3 Selection 4 Selection 5
Redev areas with VAC, VACRES, VACCOM or
med to high MU VACINST with med to high
1 opptnty in areas >= MU opptnty in areas >= 1
acre @ MU acre @ MU densities
densities
v as w Infill opptnty & incr empl
medv to high M density VACCOM and
med to high MU VACRES to SER @ MU
2 opptnty in areas >= VACRES to SER @ MU
acre @ MU density; exist SER @ MU
lacre @ MU
densities density; MOB to SER @ MU
density
Opptnty to incr res density in Opptnty to incr empl SF opptnty select Emp opptnty outside
redev areas remaining density in redev areas VACRES and using redev select high
Redev areas with
ed ih redev (areas < 1 ac) select remaining redev areas conflict & sliced SFSUIT urban pref (conflict =
med to high MU
e t in MU SF,MF & VACRES allocate (< 1 ac) select SER, allocate to most suitable 113, 112, 123, 213, 223)
3 opptnty in areas >=
1 acre @ MU @ trend MF density IND, VACCOM -incr (conf = 113, 112, 123, allocate VAC,
densities SER to MU density; 213 or 223 and highest VACCOM, VACINST to
change VACCOM & IND SFSUIT) 75% SER and 25%
..................................................................to S E R @ M U density;...... C O M trend densities
Outside of DRIs MF & SF
DRls select
DRI alocat to VACRES @ trend density;
DRI>0; allocate
ased o clic SER to VACRES @ trend
based on conflict
4 (med-high urb pre density (excessive amounts
of VACRES in buff 4); COM
and sitiiy @ to VACCOM and IND to
DRI densities
DRI densities VACIND @ trend density
DRIs select Outside of DRIs MF & SF
DRI>0; allocate to VACRES @ trend density;
based on conflict SER & COM to VACCOM
5
(avoid CON conf) and IND to VACIND @ trend
and suitability @ density any shortfalls
DRI densities allocate to VAC











APPENDIX N
STUDY AREA ROADS USED IN CFRPM BY AREA TYPE


Study Area Roads % within % outside
Average Total 3-milesof 3-miles of %of
Distance Distance future future Total
Area Type (miles) (miles) transit transit Roads
CBD 0.27 30.01 52% 48% 2%
CBD Fringe 0.25 62.38 100% 0% 4%
Residential 0.33 304.06 69% 31% 20%
Outlying Business District 0.33 86.3 85% 15% 6%
Rurual 0.54 1065.99 31% 69% 69%








~-


I I 8 i
0 4 8 Miles


Note. Map courtesy of author


175


- CBD roads
- CBD Fringe roads
Outlying Business District roads
- Residential roads
- Rural roads
Development of Regional Impact (DRIs)
TAZ's outside 3 mile buffer
TAZ's intersected by 3 mile buffer


71-,










APPENDIX O
LUCIS-PLUS LOW SCENARIO POPULATION DENSITY


LUCIS-plus Low Scenario

Population Density


Note. Map courtesy of author.


176


Population Density
(People per acre)
0 30 to 60
0 to 3 60 to 90
3to15 90to185
15 to 30 | 3 mile buffer










APPENDIX P
LUCIS-PLUS HIGH SCENARIO POPULATION DENSITY


LUCIS-plus High Scenario

Population Density


Note. Map courtesy of author.


Population Density
(People per acre)
0 30 to 60
Oto3 60 to 90
; 3to15 90to185
15 to 30 1 3 mile buffer









APPENDIX Q
LUCIS-PLUS LOW SCENARIO EMPLOYMENT DENSITY


LUCIS-plus Low Scenario

Employment Density


Note. Map courtesy of author.


178


Employment Density
Jobs per acre
0 10 to 25
Oto2 25to 50
2 to 10 50 to 85
1E13 mile buffer









APPENDIX R
LUCIS-PLUS HIGH SCENARIO EMPLOYMENT DENSITY


LUCIS-plus High Scenario

Employment Density


179


Employment Density
Jobs per acre
0 10 to 25
0 to2 25 to 50
2 to 10 50 to 85
FJ1 3 mile buffer









LIST OF REFERENCES


Bartholomew, K. (2005). Integrating land use issues into transportation planning:
Scenario planning. Retrieved March 19, 2010, from
http://faculty.arch.utah.edu/bartholomew/SP SummaryRpt Web.pdf

Bartholomew, K., & Ewing, R. (2009). Land use-transportation scenarios and future
vehicle travel and land consumption: A meta-analysis. Journal of the American
Planning Association, 75(1), 13-27.

Bento, A. M., Cropper, M. L., Mobarak, A. M., & Vinha, K. (2005). The effects of urban
spatial structure on travel demand in the United States. Review of Economics &
Statistics, 87(3), 466-478.

Borning, A., Waddell, P. & Forster, R. (2006). UrbanSim: Using simulation to inform
public deliberation and decision-making. Retrieved November 25, 2008, from
http://www.cs.washington.edu/homes/borning/papers/borning-urbansim-case-
study-2006.pdf

Carbon Dioxide Information Analysis Center. (unknown). Top 20 emitting countries by
total fossil-fuel C02 emissions for 2006. Retrieved February 10, 2010, from
http://cdiac.esd.ornl.gov/trends/emis/tre tp20.html

Chang, J. S. (2006). Models of the relationship between transport and land-use: A
review. Transport Reviews, 26(3), 325-350.

Collins, M.G., Steiner, F.R., & Rushman, M.R. (2001). Land-use suitability analysis in
the United States: Historical development and promising technological
achievements. Environmental Management, 28(5), 611-621.

Dalton, L. C. (2001). Thinking about tomorrow: Bringing the future to the forefront of
planning. Journal of the American Planning Association, 67(4), 397.

Deka, D. (2004). Social and environmental justice issues in urban transportation. In S.
Hanson, & G. Guiliano (Eds.), The geography of urban transportation (3rd ed., pp.
332-355). New York: The Guildford Press.

Environmental Systems Research Institute. (2010). ArcGIS desktop 9.3 help. Retrieved
March 27, 2010, from
http://webhelp.esri.com/arcqisdesktop/9.3/index.cfm?TopicName=welcome

Florida Department of Transportation. (unknown). Florida highway mileage reports -
glossary. Retrieved April 1, 2010, from
http://www.dot.state.fl. us/planning/statistics/m ileage-rpts/glossary. shtm

Florida Department of Transportation. (2005). FSUTMS powered by CUBE/VOYAGER
data dictionary. Retrieved March 25, 2010, from
http://www.fsutmsonline. net/images/uploads/mtf-files/datadictionary. pdf


180









Florida Department of Transportation. (2009a). MPO program management handbook -
Chapter 4: Long-range transportation plan. Retrieved February 8, 2010, from
http://www.dot.state.fl. us/planning/policy/metrosupport/mpohandbook/ch4.pdf

Florida Department of Transportation. (2009b). Transit oriented development design
guidelines. Retrieved March 27, 2010, from
http://www.floridatod.com/docs/Products/TODGuide041409.pdf

Gannett Fleming Inc. (unknown). CFRPMv45 Cube Voyager conversion technical
memorandum 1 model conversion procedural guide. Orlando, FL: Florida
Department of Transportation District 5.

Handy, S. (2005). Smart growth and the transportation-land use connection: What does
the research tell us? International Regional Science Review, 28(2), 146-167.

Handy, S. (2008). Regional transportation planning in the US: An examination of
changes in technical aspects of the planning process in response to changing
goals. Transport Policy, 15(2), 113-126.

Hanson, S. (2004). The Context of Urban Travel: Concepts and recent trends. In S.
Hanson, & G. Guiliano (Eds.), The geography of urban transportation (3rd ed., pp.
3-29). New York: The Guildford Press.

lacono, M., Levinson, D., & EI-Geneidy, A. (2008). Models of transportation and land
use change: A guide to the territory. Journal of Planning Literature, 22(4), 323-340.

Isserman, A. (1985). Dare to plan: An essay on the role of the future in planning practice
and education. Town Planning Review, 56, 483-491.

Johnston, R. A. (2004). The urban transportation planning process. In S. Hanson, & G.
Guiliano (Eds.), The geography of urban transportation (3rd ed., pp. 115-140).
New York: The Guildford Press.

Kahyaoolu-KoraCin, J., Bassett, S. D., Mouat, D. A., & Gertler, A. W. (2009). Application
of a scenario-based modeling system to evaluate the air quality impacts of future
growth. Atmospheric Environment, 43(5), 1021-1028.

Kelly, E. D. (1994). The transportation land-use link. Journal of Planning Literature, 9(2),
128-145.

Kenworthy, J. R. (2003). Transport energy use and greenhouse gases in urban
passenger
transport systems: A study of 84 global cities. Paper presented at the International
third conference of the regional government network for
sustainable development, Notre Dame University, Fremantle, Western Australia.
Retrieved February 10, 2010, from
httD://cst.uwinnirea.ca/documents/TransDort Greenhouse.odf


181









Knaap, G. (unknown). A requiem for smart growth. Retrieved March 17, 2010, from
http://www.smartgrowth.umd.edu/research/pdf/Knaap Requiem 022305.pdf

Koomen, E., Stillwell, J., Bakema, Aldrik, Scholten, & Henk J. (Eds.). (2007). Modelling
land-use change: Progress and applications. The Netherlands: Springer.

Lake County Department of Growth Management. (2009). Lake County, Florida -
Comprehensive plan as amended through ordinance #2009-32. Retrieved January
26, 2010, from
http://www.lakecountvfl.qov/pdfs/2025/Comp Plan as%20of Ordinance 2009-
32.pdf

Lake-Sumter Metropolitan Planning Organization. (2006). Lake-Sumter 2025 long range
transportation plan prepared for: Lake-Sumter metropolitan planning organization.
Retrieved January 26, 2010, from
http://www.lakesumtermpo.com/pdfs/lrtp/lake sumter Irtp complete 052307.pdf

McHarg, I. L. (1992). Design with nature (25 anniversary ed.). New York: J. Wiley.

McNally, M. G. (2007a). The activity-based approach. In K.J. Button & D. A. Hensher
(Eds.), Handbook of transport modelling (2nd ed., pp. 53-69). Amsterdam: Elsevier
Science.

McNally, M. G. (2007b). The four step model. In K.J. Button & D.A. Hensher, (Eds.),
Handbook of transport modelling (2nd ed., pp. 35-52). Amsterdam: Elsevier
Science.

Meyer, M. D., & Miller, E. J. (2001). Urban transportation planning: A decision-oriented
approach. New York: McGraw-Hill.

Myers, D., & Kitsuse, A. (2000). Constructing the future in planning: A survey of theories
and tools. Journal of Planning Education and Research, 19(3), 221-231.

Miller, E. J., Kriger, D. S. & Hunt, J. D. (1998). Integrated urban models for simulation of
transit and land-use policies. Retrieved October 10, 2008, from
http://www.nap.edu/openbook.php?record id=9435&page=R1

Muller, P. O. (2004). Transportation and urban form: Stages in the spatial evolution of
the American metropolis. In S. Hanson, & G. Giuliano (Eds.), The geography of
urban transportation (3rd ed., pp. 59-85). New York: The Guildford Press.

Smith, S. K. (2005). Florida population growth: Past, present and future. Retrieved
January 30, 2010, from http://www.bebr.ufl.edu/system/files/FloridaPop2005 0.pdf


182









State of Florida Senate Committee on Community Affairs. (2009). Interim report 2010-
107 Population need as a criteria for changes to a local government's future land
use map. Retrieved March 9, 2010, from http://www.fl-
counties.com/Docs/Legislative%20Division/Growth%20Management/Population N
eed as a Criteria for Changes to a Local Government s Future Land Use
Map.pdf

Transportation Research Board. (2009). Special Report 298: Driving and the built
environment the effects of compact development on motorized travel, energy
use, and CC2 emissions. Washington DC: The National Academies.

Transportation Research Board. (2010). About TRB. Retrieved March 9, 2010, from
http://www.trb.org/AboutTRB/Public/AboutTRB.aspx

United Nations Economic Commission for Europe. (2008). Passenger vehicles (number)
at 31 December by vehicle category, country and year. Retrieved March 9, 2010,
from
http://w3.unece.org/pxweb/Dialol/varval.asp?ma=ZZZ TRvehiclePsqFlit r&ti=Pas
senger+Vehicles+%28number%29+at+31 +December+by+Vehicle+Category%2C
+Country+and+Year&path=../DATABASE/Stat/40-TRTRANS/02-
TRRoadFleet/&lanq=1

United States Census Bureau. (2000). Profile of selected housing characteristics: 2000
Census 2000 summary file 3 Lake County, Florida. Retrieved February 2,
2010, from http://factfinder.census.gov/servlet/QTTable? bm=y&-
geo id=05000US12069&-qr name=DEC 2000 SF3 U DP4&-
ds name=DEC 2000 SF3 U&- lang=en&- sse=on

United States Census Bureau. (2000). Profile of selected housing characteristics: 2000
Census 2000 summary file 3 Orange County, Florida. Retrieved February 2,
2010, from http://factfinder.census.gov/servlet/QTTable? bm=y&-
geo id=05000US12095&-qr name=DEC 2000 SF3 U DP4&-
ds name=DEC 2000 SF3 U&- lang=en&-redoLoq=false&- sse=on

United States Census Bureau. (2000). Profile of selected housing characteristics: 2000
Census 2000 summary file 3 Osceola County, Florida. Retrieved February 2,
2010, from http://factfinder.census.gov/servlet/QTTable? bm=y&-
geo id=05000US12097&-qr name=DEC 2000 SF3 U DP4&-
ds name=DEC 2000 SF3 U&- lana=en&-redoLoa=false&- sse=on


183









United States Census Bureau. (2007). Selected housing characteristics: 2005- 2007
American community survey 3-year estimates Lake County, Florida. Retrieved
October 21, 2009, from
http://factfinder.census.qov/servlet/ACSSAFFFacts? event=&qeo id=05000US12
069& qeoContext=01000USI04000US12105000US12069& street=& countv=Lak
e+County& cityTown=& state=& zip=& lanq=en& sse=on&ActiveGeoDiv=& use
EV=&pctxt=fph&pqsl=050& submenuld=factsheet 1&ds name=DEC 2000 SAF
F& ci nbr=null&qr name=null&req=null%3Anull& keyword=& industry=

United States Census Bureau. (2008). Journey to work and place of work Census
2000 data 2000 County to county worker flow files. Retrieved January 30, 2010,
from http://www.census.qov/population/www/cen2000/commuting/index.html#FL

United States Census Bureau. (2008). Selected housing characteristics: 2008 2008
American community survey 1-year estimates Lake County, Florida. Retrieved
February 2, 1010, from http://factfinder.census.gov/servlet/ADPTable? bm=v&-
geo id=05000US12069&-qr name=ACS 2008 1YR GOO DP4&-context=adp&-
ds name=&-tree id=308&- lanq=en&-redoLoq=false&-format=

United States Census Bureau. (2008). Selected housing characteristics: 2008 2008
American community survey 1-year estimates Orange County, Florida.
Retrieved February 2, 1010, from
http://factfinder.census.gov/servlet/ADPTable? bm=y&-context=adp&-
gr name=ACS 2008 1YR GOO DP4&-ds name=ACS 2008 1YR GOO &-
tree id=308&-redoLoq=true&- caller=qeoselect&-geo id=05000US12095&-
format=&- lang=en

United States Census Bureau. (2008). Selected housing characteristics: 2008 2008
American community survey 1-year estimates Osceola County, Florida.
Retrieved February 2, 1010, from
http://factfinder.census.gov/servlet/ADPTable? bm=y&-context=adp&-
gr name=ACS 2008 1YR GOO DP4&-ds name=ACS 2008 1YR GOO &-
tree id=308&-redoLoq=true&- caller=qeoselect&-geo id=05000US12097&-
format=&- lang=en

United States Environmental Protection Agency, Office of Research and Development.
(2000). Projecting land-use change: A summary of models for assessing the
effects of community growth and change on land-use patterns (Report No.
EPA/600/R-00/098). Cincinnati, OH: U.S. EPA.

United States Environmental Protection Agency. (2009). Human-related sources and
sinks of carbon dioxide. Retrieved February 10, 2010, from
http://www.epa.aov/climatechanae/emissions/co2 human.html


184









United States Government Accountability Office. (2007). Crude oil: Uncertainty about
future oil supply makes it important to develop a strategy for addressing a peak
and decline in oil production (Report No. GA-07-283). Washington DC: United
States Government Accountability Office.

University of Florida Bureau of Economic and Business Research, & University of
Florida Population Program. (2009). Florida population studies bulletin 153 -
Florida county population projections, 2008 2035. Gainesville, FL: Bureau of
Economic and Business Research, College of Business Administration, University
of Florida.

Wachs, M. (2001). Forecasting versus envisioning: A new window on the future. Journal
of the American Planning Association, 67(4), 367-72.

Wachs, M. (2004). Reflections on the planning process. In S. Hanson, & G. Guiliano
(Eds.), The geography of urban transportation (3rd ed., pp. 141-162). New York:
The Guildford Press.

Waddell, P., & Ulfarsson, G. F. (2004). Introduction to urban simulation: Design and
development of operational models. In P. Stopher, K. Button, K. Haynes & D.
Hensher (Eds.), Handbook of transport volume 5: Transport geography and spatial
systems (pp. 203-236) Pergamon Press.

Weitz, J. (1999). From quiet revolution to smart growth: State growth management
programs, 1960 to 1999. Journal of Planning Literature, 14(2), 266-337.

Wilbur Smith Associates. (2008). Lake County transit development plan 2008 major
update. Retrieved February 4, 2010, from
http://www.lakesumtermpo.com/pdfs/tdp/Lake County 2020 TDP adopted0808.p
df

Zegras, C., Sussman, J., & Conklin, C. (2004). Scenario planning for strategic regional
transportation planning. Journal of Urban Planning and Development, 130(1), 2-
13.

Zhao, F., & Chung, S. (2006). A study of alternative land use forecasting models (Final
Report No. BD015-10). Tallahassee, FL: Florida Department of Transportation.


185









BIOGRAPHICAL SKETCH

Liz Thompson was born in 1963 in Camp Hill, Pennsylvania. Her family emigrated

to Perth, Western Australia in 1971 where she lived until returning to live in the United

States in 2003. Liz is currently pursuing a Master of Arts degree in Urban and Regional

Planning at the University of Florida, in Gainesville. She has a Bachelor of Arts degree

in geography and history from the University of Western Australia. As a graduate

student at the University of Florida, Liz worked as a research assistant for Dr. Paul

Zwick assisting in the continuing development of the Land Use Conflict Identification

Strategy (LUCIS), work that she will continue after her graduation in summer 2010. Liz

began working for the Shimberg Centre for Housing Studies in April 2010 where her

proficiency in geographic information systems is assisting in the development of the

Florida Neighborhood Data Initiative (FNDI) which aims to support affordable housing

and community revitalization.

Liz met Kevin Thompson in 1994 during a five-year working holiday that began in

1992 and traversed Japan, the United States, Taiwan, and New Zealand. In 1996, they

moved to Liz's home town of Perth, Western Australia. In 2002, they married and later

that year their beautiful daughter Amelie Helen was born. They returned to live in the

USA in 2003, moving to Gainesville, Florida in 2007 where they continue to reside.


186





PAGE 1

1 ENVISIONING URBAN GROWTH PATTERNS THAT SUPPORT LONG RANGE PLANNING GOALS A COMPARATIVE ANALYSIS OF TWO METHODS OF FORECASTING FUTURE LAND USE CHANGE By ELIZABETH A. THOMPSON A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN URBAN AND REGIONAL PLANNING UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Elizabeth A. Thompson

PAGE 3

3 To Amelie Helen Thompson, my beloved daughter and guiding li ght

PAGE 4

4 ACKNOWLEDGMENTS My sincere thanks and gratitude are bestowed upon my generous and supportive committee members Dr. Paul Zwick and Dr. Ruth Steiner. As chair of my committee, Dr. Zwick has been instrumental in nurturing and guiding my research journey in a way that contin ually challeng es hat I might learn to see more than what is first apparent I could not have reached this important personal goal, and so to him I am truly indebted. As my committee co chair Dr. Steiner has been e qually important by helping me navigate the often murky waters of understanding the relationship between land use and transportation. Without her vast knowledge, advice and guidance I would have certainly stumbled along the way far more than I did, so to her I am also very thankful. Thanks also to Dr. Zhong Ren Peng for introducing me to the world of transportation demand modeling and for gladly giving ass istance whenever it was needed. To my fellow LUCI S ists, Iris Patten, Emily Stallings, Naser Arafat, Christen Hutton and Yuyang Zou, I owe a large measure of thanks, not only for their technical support and advice at all hours of the night and day but also for their sincere friendship which has meant a g reat deal to me. The love and support of my family made this whole journey possible, without their encouragement I surely would not have had the strength to reach this goal. I shall always be indebted to my parents Marshall and JoAn Wrightstone, and my pa rents in law, Keith and Marilyn Thompson who have helped me in more ways than I can ever mention here. And above all others I owe the most thanks to my adoring husband Kevin and precious daughter Amelie who have endured my lengthy distractedness, frequen t

PAGE 5

5 frustrations and occasional bad temper, but love me nonetheless. I could not have done this without you both, nor would I ever want to.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 10 LIST OF FIGURES ................................ ................................ ................................ ........ 12 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 The Challe nge of Forecasting the Future ................................ ................................ 17 Research Argument and Study Objective ................................ ............................... 18 Study Outline ................................ ................................ ................................ .......... 19 2 LITERATURE REVIEW ................................ ................................ .......................... 21 Introduction ................................ ................................ ................................ ............. 21 The Land Use Transportation Relationship ................................ .......................... 23 The Spatial Evolution of Urban America ................................ ........................... 24 Walking Horsecar era (1800 1890) ................................ ......................... 24 Electric streetcar era (1890 1920) ................................ ........................... 24 Recreational automobile era (1920 1945) ................................ ............... 25 Freeway era (1945 present) ................................ ................................ .... 26 Mobility and Accessibility Key Concepts ................................ ........................ 27 Trends in Planning Approach and Research ................................ .................... 29 Meta Analysis of Selected Literature Reviews ................................ ................. 34 Badoe and Miller (2000) ................................ ................................ ............. 36 Ewing and Cervero (2001) ................................ ................................ ......... 37 Handy (2 005) ................................ ................................ ............................. 41 Transportation Research Board (2009) ................................ ...................... 43 Summary ................................ ................................ ................................ .......... 45 Transportation and Land Use Modeling ................................ ................................ .. 46 Transportation Modeling The Four Step Model ................................ ............. 47 Overview ................................ ................................ ................................ .... 47 Sub models ................................ ................................ ................................ 48 Land Use Modeling ................................ ................................ .......................... 53 Model types an d operational characteristics ................................ .............. 53 Meta analysis of selected literature reviews ................................ ............... 55 Scenario Planning ................................ ................................ ................................ ... 59 History and Planning Applications ................................ ................................ .... 61

PAGE 7

7 Strengths and Limitations ................................ ................................ ................. 63 Land Use Suitability Analysis ................................ ................................ .................. 66 Overview ................................ ................................ ................................ .......... 66 Land Use Conflict Identification Strategy (LUCIS) ................................ ............ 69 Summary ................................ ................................ ................................ ................ 74 3 STUDY AREA AND TIME FRAME ................................ ................................ ......... 77 Study Area Selection ................................ ................................ .............................. 77 Study Time F rame ................................ ................................ ................................ .. 86 4 METHODOLOGY ................................ ................................ ................................ ... 88 Methodology Overview ................................ ................................ ........................... 88 Step One ................................ ................................ ................................ ................. 90 Florida Land Use Allocat ion Method Scenario Data ................................ ......... 90 Land Use Conflict Identification Strategy Planning Land Use Scenarios Data ................................ ................................ ................................ .............. 91 Step Two ................................ ................................ ................................ ................. 92 Transportation Analysis Zone Data Overview ................................ ............... 92 FLUAM the Florida Land Use Allocation Method ................................ ........... 93 Development of control totals ................................ ................................ ..... 93 Determination of developable lands ................................ ........................... 94 Input of approved developments, manual adjustments and overrides ....... 94 Allocation of growth to TAZs ................................ ................................ ...... 96 LUCIS plus The LUCIS Planning Land Use Scenarios Method .................... 97 Creation of combine raster to identify areas suitable for growth .............. 100 Establishment of existing gross densities ................................ ................ 110 Establishment of control totals for growth ................................ ................ 114 Allocation of existing population and employment using gross densities 114 Establishmen t of guidelines for distribution of growth .............................. 115 Allocation of growth to suitable areas based on established guidelines ... 120 Step Three ................................ ................................ ................................ ............ 126 FLUAM Scenario TAZ Data and Transit System ................................ ......... 127 LUCIS Plus Scenarios TAZ Data and Transit System ................................ 128 Central Florida Regional Planning Model Model Run for Each Scenario .... 129 5 FINDINGS ................................ ................................ ................................ ............. 132 Compact Development ................................ ................................ ......................... 132 Energy Conservation ................................ ................................ ............................ 135 Single Occupant Vehicle and Transit Trips ................................ ........................... 139 Summary ................................ ................................ ................................ .............. 140 6 DISCUSSION ................................ ................................ ................................ ....... 142 Discussion of Findings and Methods ................................ ................................ .... 142 Compact Development ................................ ................................ ................... 142

PAGE 8

8 Increased Tr ansit Ridership ................................ ................................ ............ 144 Decreased Dependency on SOV ................................ ................................ ... 146 Energy Conservation Reduction in VMT ................................ ...................... 148 Regional Accessibility and the Challenges of DRIs ................................ ........ 150 Proximity to Transit and Employment Densities ................................ ............. 151 Criticisms of the FSM and the CFRPM ................................ ........................... 151 Complex Land Use Models vs. Deterministic Approaches to Forecasting Land Use Change ................................ ................................ ....................... 152 Study Limitations ................................ ................................ ............................ 153 7 CONCLUSION ................................ ................................ ................................ ...... 155 APPENDIX A LAND US E MODELS REVIEWED BY AUTHOR AND MODEL TYPE ................. 161 B STRENGTHS AND LIMITATIONS OF LAND USE MODEL TYPES BY AUTHOR ................................ ................................ ................................ ............... 162 C DATA DICTIONARY FOR LUCIS PLUS SCENARIOS ................................ ......... 164 D ZDATA1 AND ZDATA2 DESCRIPTIONS ................................ ............................. 165 E ALLOCATION BUFFERS 1 AND 2 ................................ ................................ ....... 166 F ALLOCATION BUFFER 3 ................................ ................................ ..................... 167 G ALLOCATION BUFFER 4 ................................ ................................ ..................... 168 H ALLOCATI ON BUFFER 5 ................................ ................................ ..................... 169 I AGGREGATION OF FDOR LAND USE CODES IN THE PARCEL DATA TO TAZ LAND USE CODES ................................ ................................ ...................... 170 J AGGREGATION OF ECFRPC EMPLOYMENT CATEGORIES TO TAZ LAND USE CODES ................................ ................................ ................................ ......... 171 K LAKE COUNTY 2009 2020 TRANSIT DEVELOPMENT PLAN MAP AND DESCRIPTION OF MODES AND ROUTES ................................ ......................... 172 L TOD DESIGN STANDARDS USED AS STUDY GUIDELINES ............................ 173 M GENERAL SELECTION CRITERIA FOR ALLOCATION INTO BUFFERS 1 TO 5 FOR LOW AND HIGH LUCIS PLUS SCENARIOS ................................ ........... 174 N STUDY AREA ROADS USED IN CFRPM BY AREA TYPE ................................ 175 O LUCIS PLUS LOW SCENARIO POPULATION DENSITY ................................ ... 176

PAGE 9

9 P LUCIS PLUS HIGH SCENARIO POPULATION DENSITY ................................ ... 177 Q LUCIS PLUS LOW SCENARIO EMPLOYMENT DENSITY ................................ 178 R LUCIS PLUS HIGH SCENARIO EMPLOYMENT DENSITY ................................ 179 LIST OF REFERENCES ................................ ................................ ............................. 180 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 186

PAGE 10

10 LIST OF TABLES Table page 2 1 Number of journal articles and book chapters resulting from literature search of ScienceDirect database ................................ ................................ ...................... 30 2 2 Selected federal policies related to urban transportation planning ......................... 32 2 3 Meta analysis of selected literature reviews summary of findings ....................... 45 2 4 Selected literature reviews for meta analysis of land use models .......................... 55 2 5 The six step LUCIS process. ................................ ................................ .................. 70 2 6 Example of a subset o f goals, objectives and sub objectives for the agriculture category ................................ ................................ ................................ .............. 71 3 1 percentage change per decade, 1900 to 2000 ................................ ................................ ................................ ....... 77 3 2 crease (per thousand people) by region, 2010 to 2035 ................................ ................................ ................................ ............... 79 3 3 2010 to 2035. ................................ ................................ ................................ .................. 81 3 4 Residence and work counties for workers in Lake, Osceola and Orange Counties for 1990 and 2000 ................................ ................................ ............... 82 3 5 Median values of owner occupied houses in 2000 and 2008 ................................ 83 4 1 Total acreage by TAZ land use categories. ................................ .......................... 111 4 2 Census Bureau estimates for 2007 for Lake Count y population by residential TAZ land use categories ................................ ................................ ................... 111 4 3 TAZ estimates for 2007 for Lake County population by residential TAZ la nd use categories ................................ ................................ ................................ ......... 112 4 4 Existing gross residential densities for study area ................................ ............... 112 4 5 ECFRPC estimates for 2007 for Lake County employment by TAZ land use categories ................................ ................................ ................................ ......... 113 4 6 TAZ estimates for 2007 for Lake County employment by TAZ land use categories ................................ ................................ ................................ ......... 113 4 7 Existing gross employment densities for study area ................................ ............ 114

PAGE 11

11 4 8 Population and employment control totals ................................ ........................... 114 4 9 NEW_LU codes ................................ ................................ ................................ .... 115 4 10 Transit systems A and B routes, modes, and headways ................................ 127 5 1 Population and employment distribution FLUAM and LUCIS plus scenarios compared to 2007 TAZ data ................................ ................................ ............. 134 5 2 Average TAZ population and employment densities by buffers ........................... 135 5 3 Daily volume of vehicle miles traveled in Lake County comparison of transit systems by land use scenario ................................ ................................ ........... 136 5 4 Daily volume of vehicle miles traveled in Lake County comparison of all scenarios for transit system B ................................ ................................ ........... 136 5 5 Vehicle miles traveled in Lake County by transit system B comparison of LUCIS plus low and FLUAM scenarios by area type ................................ ........ 137 5 6 Vehicle miles traveled in Lake County by transit System B comparison of the FLUAM and LUCIS plus low scenarios by area type ................................ ........ 137 5 7 Vehicle miles traveled in Lake County by transit System B comparison of the FLUAM and LUCIS plus high scenarios by area type ................................ ...... 138 5 8 Vehicle miles traveled in Lake County by transit System B comparison of LUCIS plus low and LUCIS plus high scenarios by area type .......................... 138 5 9 Daily volume of home based work and non home based work trips in Lake County by mode all scenarios ................................ ................................ ........ 140

PAGE 12

12 LIST OF FIGURES Figure page 2 1 Example of mutually supportive land use and transportation planning goals. ...... 22 2 2 Methods of interpreting planning goals and their relationship. ............................... 23 2 3 The Four Step Model ................................ ................................ ............................ 48 2 4 E xample of simple raster analysis ................................ ................................ .......... 68 2 5 Combining stakeholder preferences to show where conflict occurs ....................... 73 3 1 ution by region, 1900 to 2000 ................................ ...... 78 3 2 tion growth from 2010 to 2035 ............... 80 3 3 ..................... 85 4 1 Diagram of study methodology. ................................ ................................ .............. 89 4 2 Determination of developable lands in FLUAM ................................ ...................... 95 4 3 ................................ ................................ ........................ 102 4 4 Model of combin e raster for allocation buffer 1 ................................ .................... 104 4 5 The region group process ................................ ................................ .................... 106 4 6 Calculation of mixed use densities for the low LUCIS plus scenario. ................... 117 4 7 Calculation of mixed use densities for the high LUCIS plus scenario. ................. 118 4 8 Guideli nes for allocation of growth for low LUCIS plus scenario .......................... 119 4 9 Guidelines for allocation of growth for high LUCIS plus scenario ........................ 120 4 10 Calculation of remaining growth to be allocated ................................ ................. 121 4 11 Calculation of school enrolmen t for 2025 for the low LUCIS plus scenario ........ 126 4 12 Transit system A routes and stops. ................................ ................................ 128 4 13 Transit system B routes and stops. ................................ ................................ 130 5 1 TAZs within and outside a 3 mile distance of future transit. ................................ 133

PAGE 13

13 LIST OF ABBREVIATION S BRT Bus Rapid Transit CAA Clean Air Acts of 1970 and 1977 CAAA Clean Air Act Amendments of 1990 DOT United States Department of Transportation DRI Development of Regional Impact(s) ECFRPC East Central Florida Regional Planning Council EIS Environmental Impact Statement ( s ) FDOR Florida Department of Revenue F DOT Florida Department of Transportation FGDL Florida Geographic Data Library FLUAM Future Land Use Allocation Model FSM Four Step Model FSUTMS Florida Standard Urban Transportation Model Structure GHG Greenhouse Gas GIS Geographic Information System ( s ) GUI Graphical User Interface(s) ISTEA Intermodal Surface Transportation Efficiency Act of 1991 LRT Light Rail Transit LRTP Long range Transportation Plan LSMPO Lake Sumter Metropolitan Planning Organization MPO Metropolitan Planning Organization NEPA Natio nal Environmental Policy Act of 1969 PUD Planned Unit Development(s) SOV Single occupant vehicle ( s )

PAGE 14

14 TAZ Traffic Analysis Zone (s) TOD Transit Oriented Development TDM Transportation Demand Model TDP Transit Development Plan TPO Transportation Planning Organization

PAGE 15

15 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Master of Arts in Urban and Regional Planning ENVISIONING URBAN GROWTH PATTERNS THAT SUPPORT LONG RANGE PLANNING GOALS A COMPARATIVE ANALYSIS OF TWO METHODS OF FORECASTING FUTURE LAND USE CHANGE By Elizabeth A. Thompson August 2010 Chair: Paul D. Zwick Cochair: Ruth L. Steiner Major: Urban and Reg ional Planning Anticipating the future underlies much of the work of urban and regional planners who often rely on forecasts created by models attempting to simulate the real world. For planners concerned with the coordination of land use and transportation t he complex and interrelated nature of the two presents many challenges to the development of models that realistically capture the intricacies of the land use and transportatio n relationship Because land use change effects transportation de mand, land use forecasting methods may thus have a significant effect on forecasts of transportation demand and ultimately influence t he extent to which land use and transportation plans can be successfully coordinated. This study is based upon the principle that the method used to forecast future land use change in a region is influential to the achievement of long range transportation planning goals. A set of interrelated land use and transportation planning goals are used as guidelines f or the creation of three future land use scenarios for the study area, Lake County, Florida for the period 2007 to 2025 These goals focus on the discouragement of urban sprawl through a compact development pattern, and an

PAGE 16

16 increase in energy conservation through a reduction in single occupant vehicle trips, vehicle miles of travel (VMT) and increased transit ridership. Two different methods of forecasting future land use change are used in this study to model three future land use scenarios for the stud y area. One method uses the Florida Land Use Allocation Method (FLUAM) and is based on historical development trends, comprehensive plan policies and a mathematically derived gravity model to allocate future population and employment growth. The other met hod uses the Land Use Conflict Identification Strategy Planning Land Use Scenarios method (LUCIS plus) and is one based in land use suitability analysis and scenario planning techniques The transportation demand of each scenario is modeled and several tr ansportation variables are compared to determine the effect each future land use forecasting method has in the achievement of long range planning goals. The results of this study show the LUCIS plus method can create a future land use scenario where 11% more population reside within 3 miles of future transit routes and at a density 71% greater than in a scenario created using FLUAM. A more compact LUCIS plus land use scenario can result in greater energy conservation through a potential decrease in vehic le miles of travel of 5.5% when compared to the FLUAM scenario. Using the LUCIS plus method to forecast future land use change can result in a regional urban form that is more supportive of long range transportation goals than a land use pattern forecaste d using the FLUAM method. The results of this study highlight the need for planning professionals and responsible government agencies in Florida to recognize the effect different land use forecasting methods can have on the coordination of long range land use and transportation plans.

PAGE 17

17 CHAPTER 1 INTRODUCTION The Challenge of Forecasting the F uture Much of the work of urban and regional planners involves anticipating the future, not simply how it will look but specifically how the actions of the present might change what is yet to come. Invariably planners rely on forecasts of the future that most often are created u sing mode ls that attempt to simulate the real world For planners concerned with the coordination of land use and transportation, forecasting models are integral tools for the creation and assessment of long range plans aim ed at minimizing the negative impacts of urban growth. The relationship between land use and transportation is complex however and its interrelated nature creates many challenges to the development of models that can accurately capture its complexity. The dimension of t ime adds to these challenges. Considering that the ramifications of change in both land use and transportation systems often take s many years before becoming apparent, the difficulty of creating realistic forecasts using models grows exponentially. The ongoing development of future land use and transportation demand models has taken place over the past five decades. During this period several federal policies recognizing the need to anticipate the impacts of major transportation investments, have manda ted the use of forecasting models. Over time forecasting models have been increasingly complex and s cientific in their calculations lending an air of inevitability to their outputs with associated unintended consequences for the coordination of land use a nd transportation plans. When a forecast of future land use change is regarded as a certainty alternative future scenarios with perhaps greater potential for minimizing long

PAGE 18

18 term transportation impacts may be overlooked in the planning process and opport unities for better coordination are missed. Research A rgument and S tudy O bjective The idea that the method of forecasting future land use change is influential in the achievement of long range land use and transportation planning goals is the principle upo n which this study is based. To test this idea, a set of interrelated land use and transportation planning goals will be used as guidelines for the creation of three future land use scenarios. Two different methods of forecasting future land use chan ge will be used to model three future scenarios. The transportation demand of each scenario will then be modeled and several transportation variables will be compared to determine the effect each future land use forecasting method had in the achievement o f the long range planning goals under investigation Current research interest in the field of land use and transportation planning has centered on the relationship between urban form and transportation demand. This study extends that research interest by testing forecasting methods against the achievement of long range goals that specifically relate to the land use and transportation relationship. Lake County, in the rapidly growing central region of Florida, was chosen as a study area, and selected g oals from its comprehensive and long range transportation plans are used as guidelines for the creation of the scenarios to be tested in this study. The two methods of forecasting future land use change differ on several levels. One is the forecasting me thod historically used in Lake County, and is based on historical development trends, comprehensive plan policies and a mathematically derived gravity model to allocate future population and employment growth. The other method is one based in land use sui tability analysis and scenario

PAGE 19

19 planning techniques that allocates future growth according to user defined guidelines and considerations of land use suitability. It is the contention of this study that the latter of these two methods will result in a more satisfactory achievement of the long range planning goals under investigation in this study than the former method. The results of this study will highlight the need for planning professionals and responsible government agencies to recognize the effect d ifferent land use forecasting methods can have on the coordination of long range land use and transportation plans. Should the difference in methods be significant it may be necessary to re evaluate the current methodologies for forecasting land use at s tate, regional and local planning levels In the event that the results are inconclusive however this study will serve as a starting point for further research into ways in which the coordination of land use and transportation plans can be improved. Stud y O utline The following chapters detail the research path taken in this study. Beginning in Chapter Two, background information is assembled in a review of the literature on topics relating to the relationship between land use and transportation, models of forecasting transportation demand and land use change, scenario planning and land use suitability analysis The choice of study area and time frame are described in Chapter Three, followed by an explanation of the study methodology detailing the planni ng goals, forecasting methods, and transportation model used in testing the research argument in Chapter Fo ur. Chapter Five describes the findings of the study and Chapter Six discusses those results in light of the literature reviewed, opportunities for further research and limitations of the study. Chapter Six concludes the study with a

PAGE 20

20 summary of its overall findings and a discussion of improving the coordination of land use and transportation planning in the state of Flori da.

PAGE 21

21 CHAPTER 2 LITERATURE REVIEW Introduction The interdependent nature of the relationship between land use and transportation is a central underlying theme in this study. Both land use and transportation are demand driven human activities and each has the potential to effect change in the other. This is of particular importance to urban and regional planners as they formulate and amend long range plans that aim to minimize negative social, economic and environmental impacts of urban growth. To fulfill that aim planners endeavor to create land use and transportation plans that include complementary goals and objectives that support mutually positive change in each of these key elements of their long range plans Thi s study focuses on such a set of complimentary goals; the encouragement of a compact urban development pattern and a reduc tion of single occupant vehicle (SOV) use increased transit ridership and increased energy conservation. Successful achievement of any of these goals has the potential to positive ly impact the achievement of any of the remainder Figure 2 1 represents these interactions. The effective coordination of land use and transportatio n goals thus has a significant bearing on the success or failure of long range planning initiatives and highlights the importance of better understanding the relationship between land use and transportation. Of similar importance to the effective coordina tion of planning goals are the methods used to translate, quantify and represent them. For example, the method used to translate a land use goal, such as encouraging a compact development pattern, into a spatial representation like a map or geographic dat aset may have a direct bearing on how effectively the goal is interpreted and in turn the extent to which interrelated goals

PAGE 22

22 like increasing transit ridership may be achieved (Figure 2 2) The effectiveness of methods used to translate, quantify, and repr esent planning goals thus also plays an important role in the attainment of long range planning goals and the coordination of transportation and land uses. Figure 2 1. Example of mutually supportive land use and transportation planning goals. Note. Diagram courtesy of author. The following literature review addresses these two broad topics; the relationship of land use and transportation, and methods used to represent land use and transportation goals in long range plans Beginning with th e land use and transportation relationship the review will narrow its focus to literature relevant to methods of

PAGE 23

23 forecasting future land use change and transportation demand. Trends in both land use and transportation modeling will be reviewed as these to ols are used in methods of translating planning goals and are specifically used in this study. Finally literature concerning scenario planning and suitability analysis will be reviewed as they form the basis of one of the methods of forecasting land use c hange this study compares Figure 2 2. Methods of interpreting planning goals and their relationship. Note. Diagram courtesy of author. The Land Use Transportation Relationship Land use and transportation have evolved together to produce the urban form and travel patter n s evident in the United States today. Muller (2004) describes this evolution as taking place in four stages each predominantly driven by technological

PAGE 24

24 advancements that have enhanced human mobility. The following overview bro adly highlights this interaction between transportation and land use and provides some historical background to the discussion that follows. The Spatial Evolution of Urban America Walking Horsecar e ra (1800 1890) The introduction of the horse drawn stre etcar in the mid 19 th century enabled middle income city residents to move to the urban fringes away from the overcrowding and pollution that accompanied the industrialization of American cities (Muller, 2004, p. 64 67). Prior to this transportation innov ation the urban area was limited to the extent to which people could comfortably walk resulting in the clustering of people and activities within close proximity of each other (Muller, 2004, p. 64). The horse drawn streetcars moved along rails which slight ly improved their speed compared to moving along unpaved roadways, attracting city residents that could afford the fare the opportunity to move to the narrow strip of land on the cities periphery (Muller, 2004, p. 66). Initially the streetcars followed ra dial routes but the demand for housing on the urban fringe saw the construction of cross town lines and inf ill development soon followed. Electric streetcar e ra (1890 1920) The invention of the electric traction motor was to revolutionize mobility and Mu ller (2004) considers it one of the most important innovations in US history. By attaching motors to streetcars speeds of up to 15 miles per hour could be achieved bringing a much wider area on the urban fringe into commuting distance to the city (Muller, 2004, p. 67). As streetcar lines ran further away from downtown areas development along them changed the overall shape of cities from a circular to a star shaped pattern

PAGE 25

25 Commercial development occurred alongside trolley tracks with residential streets f orming a grid pattern in between lines (Muller, 2004, p. 67). The relatively low fare charged by streetcar companies coupled with extensive networks of lines increased mobility for city residents and resulted in a greater separation of land uses as people could now live further from centers of commerce and industry (Muller, 2004, p.69) Specialized land use zones quickly emerged with the central business district becoming the predominant location for such activities. The invention of the elevator also enabled greater clustering as buildings grew in height further concentrating commercial development downtown (Muller, 2004, p. 69). The later years of this era saw the introduction of electric commuter trains in larger cities such as N ew York that either superseded the streetcar systems or in the case of new cities such as Los Angeles were adopted outright (Muller, 2004, p. 69) Recreational automobile era (1920 1945) The advent of the automobile was to have the most drastic impact on urban form beginni mass production techniques quickly made the car an affordable mode of transportation for the majority of Americans (Muller, 2004, p. 70). Many of the earliest roads were constructed in rural areas where farmers badly needed better access to local services. City dwellers initially used their cars for recreational trips but quickly came to realize their potential for personal daily travel. As early as 1922 the number of car dependent households had grown to 135,000 across 60 cities (Muller, 2004, p. 70). The increasing mobility that cars provided caused an increase in possible commuting distances an d spurred further development on urban fringes and bet ween suburban rail axes. The subsidization of the streetcar system by home building companies was no longer

PAGE 26

26 necessary as potential new homeowners provided their o wn means of transportation. T he demise of the suburban transit system soon bega n and was hei ghtened during the Depression years of the 1930s (Muller, 2004, p. 71 72). Freeway era (1945 present) In contrast to the preceding eras that were spurred by some innovation in transportation technology, this period has been driven by a different force, the freeway. Enabled by the massive highway building building effort of the post war economic boom, the pattern of urban development rapidly dispersed as mobility increased along the high speed expressways that allowed even further separation of land uses (Muller, 2004, p. 75 81). Car ownership was no longer a luxury; it was a household necessity for working, shopping and entertainment. Just as the streetcar system produced a network shaped development pattern, so too did the freeway system (Muller, 2004, p. 76) ; extending development into increasingly more distant locations and resulting in the widely dispersed patterns of urban development common across the United States today. The regional advantage that city central business districts had in attracting business and employment was mostly eliminated by the network of expressways that now connected any location along its path (Muller, 2004, p. 76). Lower cost locations for business a long expressway routes attracted commercial, retail and light industrial development to highway intersections in the outer city areas; creating suburban downtowns which in turn attracted residential development (Muller, 2004, p. 79). With the ever increasing distance and separation of land uses, mobility for m ost Americans had become dependent upon car ownership and the ability of the road network to take them to the places the needed to go.

PAGE 27

27 Mobility and Accessibility Key Concepts The above overview highlights the affect transportation innovations have had on urban form in the United States and the patterns of development that have evolved within urban areas. Two key concepts emerge from this historical background that are helpful in understanding the relationship between land use and transportation ; mobility and accessibilit y. The following section provides an overview of these concepts so the changing emphasis each has received in planning theory and pr actice may be better observed. At a very broad level the link between land use and transportation can be understood in simple terms. Transportation can be described as the network of routes taken to move from one location to another using various modes of transport (Meyer and Miller, 2001, p. 128). How this network is organized is largely dependent on the c onfiguration of the land uses it serves as these dictate where people and the activities they undertake are located and directly affects their mobility, or in other words their ability to move from place to place Similarly, land use patterns are influenced by the transportation system which determines how acces sible one location is to another (Meyer and Miller, 2001, p. 128) and is dependent upon the available routes and modes for travel As the above descriptio n highlights m obility and accessibility are two concepts central to understanding the land use and transportation relationship In this light mobility can be seen as being principally a function of the characteristics of the transportation network and the modes it employs, or more simply the means by which people can travel Accessibility however is linked to location and is characterized by how easily a location can be reached using the transportation network as well as by the

PAGE 28

28 number and type of activitie s a location offers and when these are available (Hanson, 2004, p. 5 6). Thus accessibility is dependent on mobility because it determines how e asily a location can be reached and perhaps more importantly by what modes. Hanson (2004) points out that acc essibility can be measured in different ways. F or example personal accessibility often refers to the number of activity sites within a and how easily the y can be reached whereas location accessibility refers to the number of activity sites within a given distance of a specific place (p. 6). Both are measured in similar ways however there is an important distinction between them. Measure s of location accessibility treat all peopl e within the zone of measurement as being the same, so that those without access to a private vehicle are considered to have the same accessibility as those who do (p. 6). Thus a person living in walking horsecar era town where people and activities were clustered closely together would have had higher levels of personal accessibility than someone living in a freeway era home without owning a private vehicle because the majority of activity sites can only be reached by car. Although such measurements and examples are overly simplified they do highlight the role mobility, and particularly available modes of transportation, play in how accessible people and places are to each other. T he historical background of the spatial evolu tion of urban America highli ghted that as the demand for mobility increased the need for mobility increased simultaneously as land uses became more segregated and people more car dependent In keeping with this demand, increasing mobility has been a paramount goal for transportation planners who have often equate d an increase in mobility with an increase in accessibility ( Hanson, 2004, p. 4 5 ) As Handy ( 2005) p oints out this planning

PAGE 29

29 approach is problematic as it perpetuates a cycle of continually planning for mobility, where increasing the capacity of road systems to reduce congestion and increase mobility eventually leads to more travel, mounting congestion and a need to further incr ease mobility (p. 11). While accessibility may be increased for those who own a car, those unable to drive due to their age, disability or economic hardship are significantly disadvantaged (p. 11) The likelihood of increasing transportation costs in the relatively near future due to declining global oil reserves may create economic hardship and declining accessibility for large numbers of people living in urban fringe areas where driving distances are typically longer than average commutes Trends in Pl anning Approach and Research Interest in understanding the relationship between land use and transportation is not new. As early as 1930, research was underta ken to investigate the effect of a new subway line on land values in surrounding areas (Meyer and Miller, 2001, p. 133). The past several decades however have seen an increase in research interest in better understanding the relationship. Meyer and Miller (2001) summarized 36 studies that examined the land use and transportation relationship, spanni ng the years 1930 to 1997; 6 were from prior to 1970, 12 were from the 1970s, 3 from the 1980s, and 15 from the 1990s. To substantiate whether this increase in studies represents an increase in research interest in studying the relationship between land u se and transportation, a l iterature search was undertaken using the online ScienceDirect database. The ScienceDirect database contains approximately 9.5 million journal articles and book chapters from over 2,500 peer reviewed journals and books (Elsevier, 2010). Several different parameters regarding land use and transportation were queried and the results are summarized in Table 2 1. The results of this literature search reinforce the review

PAGE 30

30 undertaken by Meyer and Miller (2001) by highlighting that sinc e 1970 research interest has grown increasingly over the past four decades Depending on the search parameters used, the increase in the number of published journal articles and book chapters on the topic of land use and transportation and related issues increased from five to ten fold between 1970 and 2010. Table 2 1. Number of journal articles and book chapters resulting from literature search of ScienceDirect database Title, Abstract or Keyword = 1961 to 1970 1971 to 1980 1981 to 1990 1991 to 2000 2001 to 2010 1970 to 2010 % incr. "land use" & "transportation" 2 67 70 129 398 494 "land use" & "transportation" & "transit" 1 18 22 42 134 644 "land use" & "transportation" & "density" 1 43 36 69 247 474 "land use" & "transp & "accessibility" 0 19 25 35 108 468 "land use" & "transp & "mobility" 0 8 11 35 105 1213 "land use" & "transp ." & "VMT 1 11 7 23 67 509 Note. The ScienceDirect online database was search ed via the University of Florida Library system at http://www.sciencedirect.com.lp.hscl.ufl.edu/ on March 9, 2010. The search included all journals and books, from all sources and all subject areas. The research trend demonstrated in Table 2 1 may be explained in the context of changes to the land use and transportation planning process that began in the late 1960s with the introduction of several key federal policies that dictated a need for a better understanding of the land use and transportation relationship. Table 2 2 lists these policies and some of their key points that were to effect the transportation planning process The passage of these Acts collectively reflects changing American attitud es towards their society and the environment. Issues of social justice within urban areas came to the forefront of planning discussions in the US during the 1960s and 1970s with an emphasis on creating redistributive policies that favored minorities and those under represented in the

PAGE 31

31 planning process becoming popular (Deka, 2004, p. 334). Urban unrest in many American cities was the driving force behind this movement, with the Watts riot in Los Angeles in 1965 typifying the tensions during this period. The McCone Commission that investigated the Watts riot concluded that the social problems that led to the rioting were in part due to a lack of adequate transportation services restricting the mobility and accessibility of inner city residents (Deka, 2004 p. 335). Three key policies reflect a government response to these issues, the Federal Aid Highway Acts of 1970 and 1973, and the National Mass Transportation Assistance Act of 1974, all of which direct government spending on mass transit. The observed increase in literature published on the topics of land use, transportation, transit, density, accessibility and mobility (Table 2 1) may be attributed to the introduction of the se three pieces of legislation and the effect they were to have on the planning process, spurring a need to better understand the interrelationship of these topics. The 1960s and 1970s was also the period at which environmentalism was at the forefront of planning discussions, with environmental planning emerging as a profession in its own right around this time (Deka, 2004, 345). Of greatest concern to environmental planners was the protection of the nat ural environment from polluting industries (Deka, 2004, 345). Such concern is reflected in the enactment of severa l important pieces of legislation that were to have significant impacts on transportation planning. They were the National Environmental Policy Act of 1969 (NEPA) the Clean Air Acts of 1970 and 1977 (CAA) and the National Energy Act of 1978 (NEA) Each of these acts was to place requirements on responsible agencies to monitor environmental impacts of transportation investments.

PAGE 32

32 Table 2 2. Selected federal policies related to urban transportation planning Year Act Key Points 1969 National Environmental Policy Act (NEPA) (EIS) for major federal actions to include direct and indirect effects (present & future) induced development effect and mu st be forecasted 1970 Clean Air Act Amendments Created Env. Protection Agency authorized to set ambient air quality standards state implementation plans (SIPs) 1970 Federal Aid Highway Act assure adverse economic, social and env. e ffects are 1973 Federal Aid Highway Act Allowed expenditure of federal systems money to be spent on mass transp. projects allowed withdraw a l of interstate segments and substitution of mass transit projects 1974 National Mass Transportation Assistance Act Authorized federal operating assistance for urban transit systems 1977 Clean Air Act Amendment s of national air quality standards develop transp. control plans to reduce mobile (from transp. sources) emissions 1978 National Energy Act All phases of transp. planning and project development were to encourage fuel conservation as car pooling programs 1990 Clean Air Act Amendments Projected emissions associated with transp. projects and programs must be reconciled emission reductions of SIPs 1991 Intermodal Surface Transportation Efficiency Act transp. planning, relating to mobility and access for people and goods, system performance and preservation and environment and quality of life Note. Adapted from 629.

PAGE 33

33 The substantial air quality problems being experienced in many US cities during the 1960s and 1970s led to the amendments of the Clean Air Act (see Table 2 2) which had originally been passed in 1955 in response to growing concerns about the health proble ms associated with vehicle emissions (Wachs, 2004, 141). Requirements to meet air quality standards placed greater emphasis on understanding how the interaction between land use and transportation influence travel behavior, the major driver of the demand for mobility and automobile use. Increased interest in transit as a means to reduce congestion, improve air quality and save energy is reflected in the enactment of the Federal Aid Highway Act amendments of 1970 and 1973 and the National Mass Transportati on Assistance Act of 1974. Seen together, the enactment of the federal policies from the 1960s and 1970s listed in Table 2 2 help explain the increased research interest during that period that the literature search details in Table 2 1 depicts. The early 1990s saw two influential acts of federal legislation also have marked affects on how transportation investments were planned. The Clean Air Act of 1970 (CAA) had already brought transportation planners into the sphere of environmental planning by linkin 2004:24). The 1990 Clean Air Act Amendments (CAAA) strengthened that link by requiring the integration of clean air planning and transportation planning at the regional level, mandating tha t transportation programs must meet air quality standards. The need to forecast future travel demand, or predict vehicle miles of travel (VMT), as a means of assessing impacts to air quality for a transportation project thus became a legislative requireme nt for local, regional and state transportation planning agencies.

PAGE 34

34 One particular conformity rule (40 CFR 93.122[b][1]) required Metropolitan Planning Organizations (MPOs) to adopt some kind of land use forecasting model or committee that could account fo r the regional impacts of transportation plans on land development (Johnston, 2004:119). Similarly the 1991 Inter modal Surface Transportation Efficiency Act (ISTEA) had an impact on the quantitative assessment of transportation plans. The Act required State Departments of Transportation (DOTs) and MPOs to have coordinated long range transportation plans (LRTPs) and transportation improvement programs (TIPs) that were now to include as considerations land use, inter modal connectivity and transit service enhancement methods (Bureau of Transportation Statistics, unknown). In particular, the requirement to consider land use in transportation planning would compel agencies to better understand the relationship between the two. When analyzed in conjunction with an overview of several key federal policies relating to urban transportation (Table 2 2), the increased research interest observed in the literature search ( Table 2 1 ) suggests that a trend exists that reflects firstly a gap in knowledge about how la nd use and transportation interact and secondly a recognition that a better understanding of it may be necessary to address the social and environmental problems urban planners seek to resolve. Meta A nalysis of S elected Literature R eviews The two fold i ncrease in literature published on topics relating to land use and transportation between 1990 and 2010 (Table 2 1) can be furthered explained in the context of two global concerns that have a direct relevance to these topics ; adverse climate change due to global warming and uncertainties about future global oil supplies. For a country like United States, which had over two billion cars in 2005 (United Nations

PAGE 35

35 Economic Commission for Europe, 2008), these issues have serious implications for the future mobi lity of a population who depend almost entirely on personal vehicles for transportation. According to the US Environmental Protection Agency (EPA) transportation is the second largest contributor to greenhouse gas (GHG) emissions (US EPA, 2009). Transpo rtation is also the largest consumer of petroleum products in the United States according to the US Government Accountability Office (GAO) (p. 9) which also reported in 2007 that global oil supplies are predicted to begin declining within the next 30 to 40 years (p. 4). Strategies to reduce GHG emissions and the consumption of petroleum products involve technological advancements to improve fuel efficiency in cars, increased use of alternative energy sources, and reducing f uel consumption. A reduction in fuel consumption, or VMT has thus become a serious goal of transportation pla nners in the past decade and provides the motivation for many of the more recent research studie s in land use and transportation In response to the recognized n eed to reduce fuel consumption the US Department of Energy requested the Transportation Research Board (TRB), a private non profit institution that provides services to the government and public on transportation matters (TRB, 2010) to undertake a study i nto the relationships between VMT, development patterns and energy consumption (TRB, 2009, p. xi). This report and the literature reviews of three other authors will be examined and a meta analysis of their findings will be reported in the re mainder of thi s section. Table 2 3 details the authors being reviewed, the year of their publication, and a broad summary of their findings.

PAGE 36

36 Badoe and Miller (2000) In their study into the transportation land use interaction, Badoe and Miller (2000) identify the curr ent state of knowledge with a particular focus on potential policy impacts and also to id entify gaps in the knowledge to guide future research (p. 236). Their overall goal is to generate discussion for the development and application of integrated models of land use and transportation (p. 236). Their empirical review focused mostly on studies published after 1994 and analyzed the data based on two categories; literature on the impact of urban form on travel behavior, and literature on the impact of transp ortation (transit in particular) on urban form (p. 236). Their review of the literature on the impact of urban form on travel behavior was further categorized according to five main topics ; residential density, accessibility, neighborhood design, car owne rship, and transit supply Their analysis of findings related to residential density was that the evidence was very mixed. Although some studies showed that density had a significant effect in increasing transit use or decreasing VMT, other studies found that the significance decreased when other factors such as socio economic status and car ownership were considered (p. 248). Findings related to accessibility were also mixed due to the fact that most studies asses s its importance relative to other fact ors and these factors varied across the studies (p. 251). Most studies overlooked the significance of connectivity (of the road network) which is often a major determinant of accessibility (p. 252). Studies concerning neighborhood design provided a mixtu re of findings with the principal flaw of most relating to the scale of their investigation. Because the geographic range of activities for most people extends far beyond the neighborhood scale the complex

PAGE 37

37 relationships between neighborhoods and regions m ay be overly simplified by this type of analysis (p. 253). Findings related to car ownership were consistent across the studies reviewed, with the common result showing that higher residential densities are associated with lower car ownership and household s with fewer cars tend to use transit more than those with more cars (p. 253). Few studies investigated the impact of transit use on urban f orm but those that did showed transit played a significant role in explaining mode choice, VMT, and effects of resi dential density (p. 254). Studies investigating the impacts of transit on urban form focused mostly on rail transit (subway, light and commuter rail) and resulted in variety of findings. Most the overall conclusion that rail development facilitates rather than generates new development (p. 259). The overall finding of the literature review by Badoe and Miller (2000) was that there was a wide variety of conclusions being drawn across various st udies about the strengths and weakness of the relationship between land use and transportation (p. 260 261). Methodological and data weaknesses were largely responsible for a lack of clarity with resp ect to potential policy impacts. They conclude that an integrated urban model that accounts for all actors and factors in the urban system should be an area for further research (p. 261). Ewing and Cervero (2001) In their study of travel and the built environment Ewing and Cervero (2001) set out to provide a generalization across a large number of previous studies into the relationship between travel behavior and urban form Many existing studies focus mainly on research findings without elaborating on methodological details making it

PAGE 38

38 difficu lt to assess their reliability and validity (p. 87 ). Ewing and Cervero aimed to collectively assess the body of literature with a focus on providing more det ail of how studies were done so differences between them could be identified (p. 87). They take a n empirical approach by reviewing 46 studies focusing on how each explains four types of travel variables; trip frequencies, trip lengths, mode choices, and cumulative person miles traveled or vehicle miles traveled or vehicle hours traveled (p. 87). The studies are categorized according to the general characteristics of the built environment they investigate; neighborhood and activity center designs, land use patterns, transportation networks, urban design features, and composite transit or pedestrian o riented design indices. A table for each category summarizes the studies it includes and the land use transportation relationships identified. Studies that compared neighborhood and activity center designs were furthered categorized as being contemporary, or traditional, car or pedestrian oriented, and urban or suburban (p. 88). Across all these sub categories of the built environment trip frequencies differed very little, with socio economic characteristics of households being a major determinant. Altho ugh evidence was limited trip lengths appeared to be shorter in traditional neighborhoods which would be expected due to the finer land use mixes and grid networks characteristic of this type of development (p. 88). Walking and transit use are also more p revalent in traditional settings but this could be due to self selection; those who prefer these modes choose to live in settings where it is possible (p. 88). Studies that tested land use variables were far more prevalent than any other type of study (p. 92). These have generally focused on residential and employment densities, measures of land use mix, and measures of accessibility that reflect the

PAGE 39

39 number of a ttractions with a specific distance of households (p. 92). Overall these studies showed vehicular travel is mainly a function of regional accessibility, even when the effects of local density and land use mixed are controlled. This indicates that higher density, mixed use developments that are regionally isolated from other activity sites may be of only modest regional travel benefit (p. 92). Trip frequencies were mostly dependent upon socio economic characteristics rather than land use variables (p. 92) Mode choice was the travel variable that was most affected by local land use variables with transit use being firstly dependent on local densities and secondly on land use mix, and walking equally dependent on both (p. 92). Employment densities at trip destinations was possibly more important at destinations than population densities at origins for both transit and walking modes, meaning the preoccupation with residential densities by proponents of transit oriented development (TOD) may be misguided (p. 92). Many studies focused on density but whether the impact of density on travel behavior is due to density itself or other variables is still not determined (p. 92 93). Many transportation network variables can affect travel times by different modes and can potentially affect travel decisions. Some include street connectivity, routing directness, block size and sidewalk continuity (p. 100). Walking and transit access are improved by grid patterns of streets but so is car access, so it is difficult to d etermine which mode gains the most advantage from this configuration (p. 100). The attractiveness of network types to particular modes depends on design and scale, with grid patterns of narrow streets being more attractive to walking than car travel (p. 1 00 101). Evidence showing a relationship between network design and travel are mixed so firm conclusions could not be drawn (p. 101).

PAGE 40

40 There were relatively few studies that tested urban design variables such as crosswalks, sidewalks, parking supply and building orientation Individually design variables were seldom significant in impacting travel and those that appeared to such as crosswalks near bus stops more likely were capturing some other unmeasured feature of the built environment (p. 102). Colle ctively however urban design variables may have some impact on travel. Studies of composite land use design features, those focusing on transit or pedestrian oriented design may show an interactive effect between land use and transportation variables (p. 106). For example a high traffic area with no sidewalks may lower accessibility whereas individually each of these factors has little impact on travel (p. 106). Different studies used different composite measures with some being more subjective than othe rs, and some arbitrarily weighting variables and others using statistical estimates to base weights according to their associations with other variables (p. 106). Overall these disparate approaches to measuring for example transit friendliness or walking quality have resulted in inconclusive results about the relationship between composite design measures and the impact on travel (p. 106). Generalizing across all the studies Ewing and Cervero conclude that trip frequencies are mainly a function of socio economic characteristics and then the built environment, whereas trip lengths are a function of the reverse, firstly the built environment and then socio economic characteristics (p. 106). Mode choices are dependent equally on both the built environment a nd socio economic characteristics, and vehicle or person miles traveled is most significantly affected by the built environment (p. 107) The authors call for more transparent ways of reporting results of studies on the relationship between land use and t ransportation and suggest as an

PAGE 41

4 1 example an approach involving the measurement of the elasticity of VMT in relation to land use and design variable s (p. 107). Handy (2005) Handy (2005) sets out to test four often stated propositions of advocates for Smart Growth approaches to plan ning that encourage urban development to be compact in form, pedestrian friendly, accessible to transit and diverse in land uses (p. 147 148). Handy reviews the studies of the relationship between land use and transportation to un cover evidence that support the following propositions: Building more highways will contribute to sprawl (p. 148) Building more highways will lead to more driving (p. 148) Investing in light rail will increase densities (p. 148) Adopting new urbanism desig n strategies will reduce automobile use (p. 148), W ith regard to the first proposition, e vidence from the literature shows that highway building does contribute to sprawl by influencing where growth occurs Rather than generating growth the research showed the effects are redistributive and overall highway building influences where and at what densities growth occurs (p. 152). Handy points out that while this appears to be true the converse is probably not. In other words, not building highways will not slow sprawl, (p. 153). For example the expectation of building a highway may be sufficient to induce new development. S tudies in the 1990s showed that with regard to the second proposition, a statistically significant relationship between increasing hig hway capacity and increased travel demand existed (p. 154). Their results suggested that in economic terms the

PAGE 42

42 travel time savings gained by increased road capacity caused an increase in travel consumption (p. 154). However more recent studies using dis aggregate approaches and modeling techniques that better identify causalities suggest that the earlier studies overestimated impacts and that highway building has only a limited affect on induced travel demand (p. 154). Handy concludes that it has yet to be concluded as to whether or not increasing highway capacity contributes to a growth in VMT (p. 154). Just as the literature showed that highway building had a redistributive rather than a generative effect on new development, the evidence in support of t he third proposition that light rail transit (LRT) will increase densities was similar (p. 156 157). The literature did show however that under certain circumstances LRT may increase densities although it was not assured These circumstances included s ignificant growth occurring in a region, a transit system that significantly increases accessibility, station locations sited in areas conducive to development, and supportive land use policies and capital investments (p. 159). Handy points out that with regard to the fourth proposition that new urbanism design strategies will reduce car use, researchers are challenged by the difficulty of separating out the relative importance of socio economic characteristics from the effects of design (p. 161). The pr oblem of self selection, that people who prefer to drive less seek out neighborhoods where it is possible to drive less, was addressed by a few residences (p. 162). Handy concludes that new urbanism design strategies may reduce car use by a small amount by addressing the unmet needs of these wanting to live in such neighborhoods (p. 162).

PAGE 43

43 In conclusion Handy finds that for all the propositions questions remain as to how st rong the link is between land use and transportation and the direction of causality in factors affecting the relationship (p. 164. She summarizes her conclusion as follows: New highway capacity will infl uence the location of growth (p. 163) N ew highway capacit y may slightly increase travel (p. 163) LRT can encourage higher densi ties under certain conditions (p. 163) New u rbanism strategies make driving less easie r for those wish to drive less (p. 163). Transportation Research Board (2009) The TRB report investigating the effects the built environment has on driving devotes a chapter to a review of the literature on the impacts of land use patterns on VMT and draws on the studies covered by the three literature reviews summarized above, Badoe and Miller ( 2000), Ewing and Cervero (2001) and Handy (2005) (p. 39). As such focus here will be g iven to the more recent studies, published after 2005, that TRB investigates in their literature review. In an effort to account for the problem of self selection so me of the more recent studies have carefully controlled for a wide range of socio economic variables to test for their effect on VMT (p. 43). While some have investigated the effect of only one factor, density, on VMT, a thorough study by Bento et al. (20 05) controlled for several variables, population centrality, jobs housing balance, city shape, road density and rail supply and found each to have a significant albeit statisti cally small effect (p. 43). The work of Bento et al. suggests however that when these factors are considered simultaneously VMT could be lowered by as much as 25 percent (p. 46).

PAGE 44

44 The effects of TOD on travel were investigated by several recent studies which indicate that transit supply and accessibility in combination with land use a re important variables affecting mode choice and thus VMT (p. 46). Of particular importance appears to be the location of a TOD within a region because this affects its accessibility to desired locations, as does the quality of connecting transit services which were found to be more influential in travel patterns than the actual design characteristics of the TOD itself (p. 46). With regard to land use and design features, proximity to transit and employment densities at trip ends appear s to have a stronge r influence on transit use than urban design features to enhance walkability and land use factors such as mixed uses and increased residential densities (p. 47). The literature review of this report concludes that few of the studies consider the potential a group of policies that combine increased density with higher concentrations of employment, improved accessibility to a variety of land uses and a good transit network can have on reducing VMT (p. 31). The authors consider two case studies where such a c ombination of factors has been fostered by different policy approaches, Portland Oregon, and Arlington County Virginia (p. 51 53). They conclude that the dramatic changes to the built environment and travel patterns require a significant political commitm ent and substantial transportation investments over a long period of time which will be challenging for many metropolitan areas (p. 54).

PAGE 45

45 Table 2 3. Meta analysis of selected literature reviews summary of findings Year Author(s) Summary of Findings 2000 Badoe & Miller Findings of studies very mixed with some showing a strong link between urban form and travel behavior while others show a weak link similarly mixed results for transit impacts on urban form. 2001 Ewing & Cervero Trip frequencies are a function of 1) socio economic characteristics and 2) built environment trip lengths are a function of 1) built environment and 2) socio economic characteristics mode choices are a function of both for VMT the built environment is most significa nt. 2005 Handy New highway capacity will influence the location of growth new highway capacity may slightly increase travel light rail can encourage higher densities under certain conditions new urbani sm strategies makes it easier to drive less for those wish to drive less. 2009 TRB Widely varying results across studies prevent any conclusive finding on the importance of changes in land use and the magnitude of their effects on travel recent studies controlling for confounding variables find modest effects of the built environment on VMT however there is some evidence that multiple factors such as population centrality, jobs housing balance, regional transit supply when implemented together may achieve greater reductions in VMT. Summary The relationship between land use and transportation is complex with each having the ability to influence change in the other. The land use patterns and transportation systems seen in the United States today are a result of over one hundred years of urban development that has been shaped by several innovations in transportation technology. Of these t he automobile has by far been the most significant in effecting change in development patterns by affording Americans an unprecedented level of person al mobility that has had positive and negative outcomes for people and their environments. In attempts to control some of the negative impacts of transportation government policies have been introduced that have placed specific requirements on

PAGE 46

46 planning, i mplementation and management of the transportation system. The literature reveals that while research into better understanding the land use and transportation relationship has been influenced by government poli cies and public concerns about uncontrolled urban development, recent global concerns about climate change and crude oil shortages have lead to even more concerted research in this area. While much has already been learned the literature shows that there is still a considerable gap in the knowledge of how land use and transportation systems should be managed in order to avoid negative social, economic and environmental impacts and still provide an adequate level of mobility a nd accessibility for everyone. Transportation and Land Use Mod eling The use of computer models in the preparation of transportation plans has been common practice since the 1960s following the introduction of Federal Aid Highway Act of 1962 which required all federally funded highway projects to be based on a contin uing, comprehensive and cooperative planning process involving state and local planning agencies (Meyer and Miller, 2001, p. 619). An integral part of that process is the forecasting of transportation demand and how that will impact the transportation net work This is typically performed using computer models as are forecasts of land use change that determine the location of people and activities upon which transportation demand is estimated. Both model types form the basis of a framework generally used by transportation planners in the creation of long rang transportation plans. This section will review the transportation and land use models commonly used in this framework and how that has changed over time, with s pecial attention focused on the model types used in this study

PAGE 47

47 Transportation Model ing The Four Step Model A model can be described as a simplification or abstraction of a real world system that can be used to test what might happen to the system if changes occur within it (Meyer and Mille r, 2001, p. 256). In the history of transportation planning one model type has been dominant since the early 1950s ( McNally, 2007 b p. 35 36). The urban transportation modeling system, most commonly referred to as the four step model (FSM) was originally developed for evaluating large scale infrastructure projects at the regional and sub regional scale (McNally, 2007 b p. 38). Despite its limitations which will be reviewed later it remains the primary tool used for forecasting future trans portation demand and performance (McNally, 2007 a p. 54). O verview The FSM is made up of four sub models that use data input s concerning the characteristics of the transportation system and the characteristics of the land use system (Figure 2 3). The tr ansportation system is represented graphically in the model as a network of links and nodes which have information such as length, speed, and capacity for links, and turn prohibitions and penalties for nodes associated with them (McNally, 2007 b, p. 38). T he land use system is aggregated to traffic analysis zones (TAZs) containing socio economic and land use data typically comprised of census data and/or data collected from household and workplace travel surveys, vehicle intercept station surveys or onboard transit surveys ( Meyer and Miller, 2001, p. 191 199). The data typically reflects activities performed over a 24 hour period and includes activity type, location, duration, and arrival and departure times (McNally, 2007b, p. 40). Following is an overvie w of the four sub models that typically make up the FSM.

PAGE 48

48 Figure 2 3. The Four Step Model. Note. A The Four Step Model 2007 b p. 39. S ub models Trip generation. In the first sub model the amount of total daily travel is determined for various activities or trip purposes. These are typically categorized as home based work (HBW) trips, home based other (non work) trips (HBO ), or non home based (NHB) tr ips. Each trip has two ends, an origin and a destination and each trip

PAGE 49

49 end is classified as a production or an attraction. Productions are considered to be the home end of a home based trip, or the origin of a NHB trip, while attractions are the non home end of a home based trip, or the destination of a NHB trip (Meyer and Miller, 2001, p. 271). S eparate models are used to define productions and attractions using socio economi c data at the TAZ level. For the calculation of productions data such as house hold income, size and employment car ownership, residential densities and distance to major activity centers are used. W orkforce data such as the number of employees, floor space areas and accessibility to workplaces are used for the calculation of attra ctions (Meyer and Miller, 2007, p. 271). The production and attraction models basically calculate a measure of attractiveness for each TAZ which serves to scale the trips for the next stage of the modeling process, trip distribution. Where certain land u ses generate untypical activities within a TAZ, such as a university or hospital, special generator factors are included in the production and attraction models to account for this activity (McNally, 2007b, p. 43) Trip d istribution. The objective of this sub model is to distribute or connect the zonal trip ends, the productions and attractions estimated for each zone in the trip generation model, in order to predict the flow of trips from production zone s and to attraction zone s (Meyer and Miller, 2001, p 278). While there are many types of trip distribution models, the most commonly used type is the gravity model This type of model estimates a travel impedance or friction factor (in terms of travel time or cost) between two zones as a function of both the land use and transportation systems characteristics (McNally, 2007b, p. 45). The transportation model used in this study, the Central Florida Regional Planning Model (CFRPM) follows the Florida Standard Urban

PAGE 50

50 Transportation Model Structure (FSUTMS) w hich specifies the use of a gravity model for calculating trip distribution (Gannett Fleming, unknown, p. 45). In this gravity model the desirability of traveling to a zone is directly related to its potential as a destination based on the activities with in that zone, and is inversely related to the spatial separation (friction factor ) between production and attraction zones (Gannett Fleming, unknown, p. 45). The final stage of trip distribution involves calibrating the model to the observed data and often involves the manual adjustment of friction factors so a best fit can be made between the observed and estimated trips (Meyer and Miller, 2001, p. 280). Mode c hoice. The purpose of the this sub model is to factor the trip tables from the trip distribution model to produce trip tables specific to each mode being modeled (McNally, 2007b, p. 48). The most commonly used model type for this step is the nested logit model, which is used in the CFRPM. Zone to zone person trips are allocated by trip purpose across the various modes and coefficients are used that quantify the sensitivity or elasticity of each mode choice to changes in service (Gan nett Fleming, unknown, p. 71). Traffic a ssignment. In th e final sub model the predicted flows between origin destination pair s for each mode are assigned to actual routes on the transportation network In determining route selection most models assume that individuals seek out the most cost effective route, in time or money. The CFRPM employs an equilibrium technique that runs several iterations of a minimum path capacity constraint assignment with travel time delays for each iteration being incorporated into subsequent ones (Gannett Fleming, unknown, p. 77). The final result of this model step is an estimate of trip volumes and speeds on the network links. In the CFRPM highway and transit trips

PAGE 51

51 are assigned in two different sub models. The outputs from the highway assignment include statistics on the number of lane and system miles, VMT, VHT (vehicle hours traveled), as well as original and congested speeds. The transit assignment outputs include transit boards by mode, route and operator, as well as boarding and departure statistics for each stop by route (Gannett Flemming, unknown, pp. 85 88). Once the model is calibrated to reflect observed data it can be run for future year estimates based on projected population and employment data and new land use configurations. This study involves the modeling o f three future land use scenarios using the CFRPM. Limitations of the FSM An often cited criticism of the FSM is that it was originally designed for evaluating large scale highway projects and is policy sensitive only with regard to major capacity impro vements (McNally, 2007a, p. 54). As such the FSM is not considered to be very effective in analyzing policy sensitivity because it fails to a dequately capture the complexities of travel, particularly at smaller scales. The FSM focus on trips rather than activities results in the two not being linked during the modeling process which is problematic because it is the underlying activity behavior that generates trips (M cNally, 2007a, p. 55). Also the FSM does not reflect temporal const raints and other depen dencies of activity scheduling such as trip chaining, meaning the sequencing of several trips to maximize time and cost savings (McNally, 2007a, p. 55). These limitations, among others, have led to continued research and development of alternative modelin g approaches. Alte rnative approaches to the FSM. The FSM is an aggregate level model, meaning it uses zonal data to model predictions about individual trips. An alternative

PAGE 52

52 method is the disaggregate approach that begins by simulating the decision proces s of individuals and summing them up to calculate aggregate travel demand (Meyer and Miller, 2001, p. 290). Often referred to as discrete choice models, these typically are based on the assumption that trip makers always seek to maximize utility, that is, they rank trips according to their relative desirability, time/money cost, comfort and/or convenience (Meyer and Miller, p. 290). Discrete choice models may utilize different concepts to assume how trip decisions are made, such as random utility or nested logit models like those commonly used in the mode choice step of the FSM Activity based models are another approach which seeks to account more for the behavioral aspects of trip making decisions. Models of this type typically fall into two broad categories, econometric models and hybrid simulation models (Meyer and Miller, 2001, p. 306). Econometric models contain systems of equations that calculate activity travel choices as probabilities of possible outcomes or decisions. Hybrid simulation mod els typically consist of rule based algorithms that attempt to replicate actual decision making processes (Meyer and Miller, 2001, pp. 301 307). Integrated land use and transportation models are another approach to improving the predictive capabilities of travel demand forecasting. Such an approach involves the creation of feedback mechanisms between both model types within a modeling framework which acknowledges the influence that land use, or human activities, has on travel behavior which determines the transportation network and in turn influences the location and intensities of land uses (Miller et al. 1998, p. 1). Different land use models and how they have been incorporated into the transportation modeling framework will be discussed in more detail in the following section.

PAGE 53

53 Land Use Modeling At the basis of integrated land use and transportation modeling frameworks is the idea that the relationship between the two can be simulated in a way that meaningfully represents reality. Since the late 1950s a number of theoretical frameworks have been developed by researchers that have significantly influenced the current research in land use and transportation modeling ( Iacono et al. 2008, p. 323). In a broad way these frameworks can be categorized into t hree main areas, spatial interaction models, econometric model s and micro simulation models. Model types and operational c haracteristics Spatial i nteraction m odels. Many early land use models employed the theory of gravity to explain the distribution of p eople and activities across regions. In 1964 Lowry was the first to use a gravity model as a basis for explaining the spatial interaction of regions and their land uses (Iacono et al. 2008, p. 325). The main assumption of gravity models is that the larg er a city or region the greater its ability to attract population, employment and commerce (Koomen et al. 2007, p. 7). A large number of spatial interaction models have been developed over the past four decades that continue to use gravity theory. One of the two methods of forecasting future land use compared in this study uses a spatial interaction model based in gravity theory to allocate projected population and employment growth. It will be reviewed in more detail in the Chapter 3. Econometric m ode ls. Models that employ econometric frameworks generally contain two model types to predict locations of population, employment, commercial activities and at times transportation flows. These models typically contain land market and regional economic sub

PAGE 54

54 variety of models have developed using this framework, with many using different theories to explain spatial distributions and actor behaviors. Random utility theory and discrete choice theory are t wo methods used to improve the capabilities of these models to simulate individual behavior which has resulted in improved measures of accessibility and travel d emand forecasting (Iacono et al. 2008, p. 328 331). Micro simulation models. As the name infe rs, micro simulation models attempt to model at a much finer resolution than the other model types. They focus on the level of the individual and typically include all the actors who influence land use changes. Although some econometric models incorporat e individual behavior they differ from micro simulation models in the way they estimate these behaviors. Rather than using a cross sectional or average of a population or market to estimate an individual behavior, micro simulation models take a bottom up approach and estimate the behavior of populations and markets based on the cumulative behavi or of individuals (Koomen et al. 2007, p. 12). Like econometric models, micro simulation models employ a variety of theories to explain the location of people, ac tivities and commerce, including random utility, discrete choice, multi agent rule based, and cellular automata approac hes to name a few (Iacono et al. 2007, pp. 332 335). Operational characteristics. Apart from theoretical differences, models can also d iffer by their operational characteristics. For example, a model may be characterized as static, meaning it calculates a set of circumstances at a given point in time; or dynamic, meaning it contains intermediate time steps that become beginning points fo r subse quent occurrences (Koomen et al. 2007, p. 3). The variety of theoretical and operational characteristics used to describe models often makes it difficult to categorize

PAGE 55

55 them into uniform groups. The three frameworks described above however are a starting point to understanding land use modeling and provide a wide context within which the following meta analysis of literature reviews can be analyzed Meta analysis of selected literature reviews Four literature reviews on land use models used in t ransportation modeling frameworks were selected for analysis to determine whether any trends in the literature could be observed. The selection of reviews was based on several criteria. As the United States and Florida in particular are the locations of the study area for this research, literature reviews that were relevant to both these geographic regions were selected. To offer an international perspective it was also considered important to include at le ast one review from outside the United States. Additionally, it was also considered important to include reviews that were written from an academic or theoretical standpoint, as well as reviews that aimed to fulfill practical purposes or specific project goals, and to focus on most recently published works Table 2 4 lists the reviews by region, year, author and title. T able 2 4. Selected literature r eviews for meta analysis of land use models Region Year Author(s) Title USA 1998 Miller, Kriger and Hunt Integrated Urban Models for Simulation of Transit and Land Use Policies USA 2000 U.S. Environmental Protection Agency Projecting Land Use Change: A Summary of Models for Assessing the Effects of Community Growth and Change on Land Use Patterns FL 2006 Zhao and Chung A study of alternative land use forecasting models KOREA 2006 Chang Models of the Relationship Between Transport and Land use: A Review In the literature models are typically characterized by the different methodologies they employ. Often models use multiple methodologies which can make categorization

PAGE 56

56 difficult. In this meta analysis the method used to categorize model types is the same as that used in the literature review by Zhao and Chung (2006) which reviews the largest number of models of all the selected reviews. Their approach was to classify models based upon the methodology most emphasized by each model ( Zhao and Chung, 2006, p. 3 ). A table listing th ese categories and which models were reviewed by each author is listed in Appendix A. Model strengths and weaknesses. In each of the selected literature reviews the authors assess the strengths and weaknesses of each model type. Ap pendix B provides a table summarizing these by author and model type. Over the time period covered by these literature reviews (1998 to 2006) there appears to be a trend toward favoring the more dynamic and disaggregate model types capable of a finer spatial resolution, over the static and aggregate ones which typically have much coarser spatial resolution. Disaggregate models attempt to simulate the behavior of individuals then make assumptions about the behavior of large groups such as markets or populations. Conversely, aggregate models simulate the behavior of markets o r populations and make assumptions about the behavior of individuals. Additionally, a model is considered to be dynamic if the outputs of an earlier iteration of the model form part of the inputs of subsequent iterations, in a way that reflects the cyclic al nature of transportation and land use. Static models on the other hand are not cyclical and require a stepped process to occur if a feedback is required between iterations of the model. With the exception of the review by Zhao and Chung (2006) the li terature reviews did not intend to identify a particular model or type that is superior to the rest. In all of

PAGE 57

57 the reviews however strengths and limitations are mentioned for the majority of model types. When these factors are summarized ( Appendix B) a general trend appears that correlates with a characterization of the aggregate and static model types being limited by those factors (i.e. by being static and aggregate), and the disaggregate and dynamic model types as being str engthened by those characteristics (i.e. by being dynamic and disaggregate). In addition, the majority of models reviewed in the earliest study by Miller et al. (1998) were static and aggregate (spatial interaction and spatial input/out models), with on ly one dynamic and disaggregate model (UrbanSim a micro simulation model) being reviewed. The more recent study by Zhao and Chung (2006) reviewed the most models, including the largest number of dynamic and disaggregate models of all the literature review s, finding those to be the least limited of all model types. Observed trend in the literature. A trend that favors the dynamic, disaggregate model types over the static, aggregate ones has been recognized by others (Johnston 2004, Waddell 2004, Iacono et al. 2008 and Koomen et al. 2007) who have put forth explanations for this development. The literature suggests that two influential factors have contributed to this trend, firstly two key federal policies concerning the environment and transportation, and secondly the emergence of Smart Growth initiatives. The influence that the CAAA (1990) and the ISTEA (1991) had on the land use and transportation planning process has already been discussed in an earlier section, Trends in P lanning Approach and Rese arch so only the topic of Smart Growth initiatives will discussed here. Smart Growth Initiatives. Beginning in the mid 1980s with increasing numbers of states in the US overhauling their growth management legislation in an attempt to

PAGE 58

58 combat sprawling dev elopment patterns (Weitz, 1999, p. 268), consensus within and outside the planning community was building in support of a new approach to land use planning. Although the interdependent nature of land use and transportation had long been recognized (Kelly 1994 p. 129) the idea that transportation investments might be (Knaap, unknown p. 12). The passa ge of ISTEA in 1991 led the US EPA to create the Urban Economic Develo pment Division (UEDD) which formed the Smart Growth Network, to provide funding to a variety of Smart G row th projects (Knaap, unknown, p. 9). Of the many principles that characterize Smart Growth development, three stand out as most relevant to this study By advocating mixed land uses, a provision of a variety of transportation choices, and directing development toward existing communities, Smart Growth initiatives have changed the way in which transportation planners have traditionally viewed land use an d development Until thi s time transportation planners an d the land use models they used assumed land use change resulted in changed transportation demands but that the transportation system itself did not influence land use change (Borning, 2006, p. 2). These particular Smart Growth concepts introduced a new level of complexity to travel demand modeling, and in turn land use change forecasting, that highlighted the spatial, temporal and behavioral limitations of the earlier models. The trend towards mo re dynamic and disaggregate model types identified in this meta analysis can be seen as an outcome associated with the rise of Smart Growth initiatives.

PAGE 59

59 Criticisms of complex models. The more complex models favored in the trend observed by this review are not without some criticisms and drawbacks. Their complexity often requires expert technical assistance for their implementation and operation, making them potentially expensive and not accessible to those who may benefit from them most, in particular metropolitan planning organizations Additionally, they often lack transparency, with the underlying assumptions that form critical model parameters being embedded within the model process in such a way that the non expert user may not fully understand t he information the model produces. These challenges however present opportunities for alternative directions in the development of land use models. One such direction is to pursue a middle ground between the ex models simpler models, such as rule based scenario analysis models, can be created that incorporate more so phisticated modeling techniques. Such an approach is investigated in this study which uses a land use forecasting method base d in scenario analysis and suitability analysis for the creation of two of the three land use scenarios this study compares. These two techniques will be reviewed in more detail in the following sections. Scenario Planning Much of what planners do involve s planning for the future which invariably requires some estimation of what the future might look like. As noted in the previous section transportation planners use models as tools to forecast the performance of transportation systems and an integral part of that process involves the use of models to forecast land use change. Also, as reported in previous sections several federal policies, such as NEPA and CAAA require state and local government agencies to

PAGE 60

60 assess impacts of transportation related initiat ives using forecasts of future urban growth and transportation demands. Over the past five decades forecasts have come to play a more prominent role in land use and transportation planning than in previous eras when plans had a more visionary purpose as o ppo sed to their contemporary focus of mitigating possible future problems (Wachs, 2001, p. 369). Some are critical of this the role of planning were simply to accommodate wh at is forecast and ignore the fact More recently Handy (2008 ) has also been critical of the emphasis on forecasts because often they are viewed in the planning process as a prediction of the future rather than what they really are, just one possible outcome among many (p. 122). The literature reviewed in the previous section revealed over the past 15 years a trend in land use modeling that has seen more complex, dynamic and disaggregate models being favored over simpler modeling techniques. Simpler land use models such as r ule based or deterministic models which produce future scenarios based on user inputted data, preferences and assumption s to reflect a particular hypothetical future, are often considered to be lacking in theory and sophisticated modeling techniques (USEPA, 2000; Zhao and Chung, 2006). Conflicting views exist about the strengths and weaknesses of each approach, leading to the question of which method is better complex/ technica l or simple/hypothetical To present a context for understanding this argument and provide a background for this research project that compares a scenario based forecasting approach with a standard la nd use forecasting method based in

PAGE 61

61 gravity theory, a review of the literature o n scenario planning a nd its strengths and weaknesses follows. History and P lanning A pplications A scenario is simply a version of what the future might be based on a particular set of assumptions (Bartholomew, 2005 p. 5; Myers and Kitsuse, 2000, p. 228) Scenario planning typically involves the conception of a range of possible scenarios followed by a process of analysis and evaluation to n arrow down possibilities so that an appropriate course of action can be identified (Bartholomew and Ewing 2009, p. 14). Scenario planning was first used in strategic warfare with some dating its origins as far back as the sixth century (Bartholomew, 2005, p. 4 5). In contemporary times its use was first popularized for military purposes by the RAND Corporation in the 1950s when they strategically assessed potential nuclear conflicts in the years following World War II (Bartholomew, 2005, p. 5). Scenario planning has also been exte nsively used in business applications, with the Royal Dutch Shell Company being among the first to use it in the late 1960s and early 1970s to develop strategic plans that were to enable them to anticipate and prepare for the global oil crisis of 1973 (Zeg ras et al. 2004, pp. 2 3). Scenario techniques have been used in land use planning since the late 1940s et al. 2009, p. 1022). Since then t he passage of NEPA in 1969 saw an increase in scenario techniques being used in transportation planning due to the requirement that environmental impact statements include alternative s to project plans (Bartholomew, 2005, p. 7). In military and business applications of scenario planning the focus is typically placed on assessing the interrelated causal relationships between external factors, such as economic conditions and environmental resources and how they limit or enhance

PAGE 62

62 possible strategies under consideration The approach to transport ation planning that NEPA helped to create however focuses not on how external factors impact a scenario but largely the reverse. Analysis of i nternally specified actions are used to create a scenario that is the n assessed based on its impact on external r esources and conditions for example, how particular changes to the transportation network may impact congestion. I nteraction betwee n internal and external factors, like how changes to the network might impact land uses, however is largely ignored (Bartho lomew, 2005, p. 7). T ransportation planning has been highly dependent on computer modeling systems that treat future land use data as an external factor that serve s as an input to the modeling process but remain s constant acr oss all scenarios being modeled By only considering one possible future land use pattern opportunities to assess the effects other alternatives might have on the transportation system are lost (Bartholomew, 2005, p. 7) and a plan is produced that is designed to address the problems fo recasted for only one possible vision of the future (Zegras et al. 2004, p. 4). Recent trends in land use modeling described in a previous section of this review are largely a response to this acknowledged shortcoming of the traditional transportation p lanning approach. Rather than using scenario planning to produce several future land use patterns for analysis with transportation models, research and practical emphasis continues to be given to more complex land use models for the production of a single forecast that serves as the only land use input into the transportation modeling process (Bartholomew, 2005, p. 7)

PAGE 63

63 Strengths and Limitations Apart from modeling criticisms that scenario planning lack s sophisticated techniques and a strong theoret ical basis (EPA, 2000), and poor capabilities to model complex economic and market processes ( Zhao and Chung, 2006) other more process oriented criticisms have been identified. Bartholomew (2005) notes a certain amount of bias may exist when too few scen arios are analyzed. For example, analyzing only two scenarios often results in them being categorized in stark terms of good and bad, and similarly three scenarios as low, medium and high risk ( p. 14). While each scenario should be objectively analyzed o n its strengths and weaknesses when too few are compared it is easier for simple categorizations like good and bad to be made The consideration of too many scenarios can also be problematic because of the excessive amounts of information often needed fo r analysis. This can cause particular difficulties when scenario analysis is linked to other modeling frameworks such as transportation demand models which consume large amounts of data and time for analysis (Zegras et al. 2004, p. 10). Difficulties in linking scenario analysis to other quantitative tools also make it difficult to empirically evaluate impacts and compare them across scenarios (Zegras et al. 2004, p. 11). Scenario planning often takes place in a collaborative environment with many stak eholders involved in the process, leading to claims that it is too subjective and idealistic to reflect the real world Similarly the idealistic nature of the process may lead to the selection of scenarios that may not be possible to implement in the real world ( Myers and Kitsuse, 2000, p. 222 ) Some argue however that it is this idealistic nature that gives sc enario planning an advantage over technically generated forecasts because it lends itself better to capturing macro trends that can be overlooked w hen

PAGE 64

64 analysis is too narrowly focused to meet more precise modeling parameters (Zegras et al. 2004, p. 3). Many of the strengths that scenario planning has to offer to the coordination of land use and transportation are drawn from the weaknesses of the t raditional transportation planning approach that has favored the use of complex and technical models to fo recast future land use change. All forecasts of future land use change are based on assumptions that are inherently subjective because the future cannot be known While s impler approaches to modeling change, such as scenario planning are often criticized for being too subjective, excessive weight is often given to forecasts derived from complex and technical computer models beca use they have the illusion of being scientifically objective. With regard to transportation planning and demand models in particular, Handy (2008) points out that not only do models produce forecasts based on assumptions they also use forecasts that were also based on assumption all of which are subjective in nature (p. 122). She cites as examples the use of K factors in gravity models to achieve a best fit with observed trip distribution data and which modes are included in the mode choice ste p in trav el demand models. Both examples embody a subjectivity that is often less t han apparent to non technicians involved in the planning process, often the public and their representatives (p. 122). The creation of forecasts of land use change using models ba sed on historical economic trends and short range planning policies, such as future land use elements of comprehensive plans with a horizon of ten years, can lead to narrow views of possible futures. Myers and Kitsuse (2000) note that political and econom ic pressures on planners are often translated into plans that emphasize short range planning goals at

PAGE 65

65 the expense of more visionary long range goals (p. 222). The use of forecasts such as these is particularly problematic in transportation planning. Land use change occurs slowly and incrementally so that policies targeting major changes to development patterns may take many years to become apparent as trends. When forecasts are based on inputs such as short range comprehensive plans long range future tr ends may not be adequately captured. The result of using forecasts derived in this way can be the implementation of transportation projects that are designed for short rather than long range future land use pattern s This can be problematic because histo ry has shown that transportation projects have a strong tendency to in fluence urban form once they are built. However, because land use change typically happens at a slower pace there is the danger that undesirable short range development patterns will be u nintentionally continued by using forecasts based on short range future land use trends represented in comprehensive plans which are most typically used in traditional approaches to transportation planning Scenario planning may be better placed to capt ure wide macro trends missed by statistically d erived forecasts due to its more transparent and collaborative process of incorporating many points of view and encouraging visions of the future that differ greatly from the present. Suggestions have been made that transportation planning would benefit from the use of forecasts derived using both scenario planning and more sophisticated land use modeling techniques ( Lemp et al., 2008; Myers and Kitsuse, 2000; Wachs, 2001; Zegras, et al. 2004). This study u ses a forecasting method that employs both scenario analysis and land use suitability analysis to assess the impact different methods of forecasting land use change have on the attainment of long range planning goals. The

PAGE 66

66 following section discusses the u se of suitability analysis in understanding how future land use change may occur. Land Use Suitability Analysis Overview As the name implies land use suitability analysis is a process of identifying the most appropriate location and distribution of futu re land uses (Collins, 2001, p. 611; Malczewski, 2003, p. 4). Various layers of spatially referenced information about selected characteristics of a study area can be analyzed collectively when overlain together, providing a cumulative assessment of a loc ation making decisions about its future use. A simple example would be combining information regarding soils, slope, and hydrology to appropriately locate a building site and decide if that is the most appropriate location a mong other potential sites Land use suitability analysis is used in a wide variety of settings, from fields such as ecology for defining habitat suitability, agriculture for crop suitability, landscape architecture for site analysis, and environmental sc iences where impact assessments may be required (Malczewski, 2003, p. 4). Its applicabilit y to planning related fields stems from its strength as a tool to aid decision makers when formulating policies about site specific development proposals, their en vironmental impacts and their potential cumulative effects over time (Collins et al., 2001, p. 611) The use of computer models to perform land use suitability analysis has become more common in the past two decades as off the shelf geographic informati on systems (GIS) software programs have become more readily available, affordable and user friendly. The first use of computers can be traced back to the early 1960s although the history of land use suitability analysis dates back to a much earlier time. The use of

PAGE 67

67 sun prints, hand drawn overlays or maps held up to a window, by early twentieth century landscap e architects Olmstead and Eliot are credited with being some of the first known uses of the technique (Collins et al., 2001, p. 612). Perhaps the m ost well known early advocate of land use suitability analysis however was the landscape architect Ian McHarg who in the 1960s was responsible f or a significant methodological advancement McHarg introduced a weighting system to his suitability analysis whereby he would shade individual transparent overlays of a study area, each representing a unique site characteristic, using lighter and darker colors to denote increasing or decreasing suitability. When viewed co llectively the degree to which a location was more or less suitable could be determined by the cumulative shade of color at that location ( McHarg, 1992, pp. 31 42). The introduction of GIS to suitability analysis was to have dramatic effects on the speed and accuracy of analyses vector data, that is represented as unique points, lines or polygons, the development of raster based GIS programs enabled not only a greater number of data layers to be analyzed but increased computat ional speed and spatial accuracy (Collins et al., 2001, pp. 613 614) Raster data is represented using a grid or raster with each individual grid cell containing a unique value representing some attribute in the real world. Figure 2 4 shows the concept o f raster analysis and how raster data can be combined to create new aggregate datasets. suitability, raster analysis combines grid cells to determine values of varying le vels of suitability. Using the example shown in Figure 2 4 the following analysis could be

PAGE 68

68 undertaken for a study area represen ted by the raster extent shown to determine the most suitable location for a new school based on two site characteristics, dis tance to major roads where closer distances are less suitable and distance to existing residential areas where closer distances are more suitable. When the two input rasters are combined the analysis shows the most suitable location for a school would be located in the lower left grid cell of the output raster with a value of 6. Figure 2 4. Example of simple raster analysis Note. Diagram courtesy of author. Decisions on how to assign suitability values must be made at two different levels during th e analysis. Firstly each dataset in the analysis must have a range of suitability values assigned to it, and secondly each dataset must be weighted according to its

PAGE 69

69 relative importance in the overall analysis. In the example shown in Figure 2 4 an equal importance or weight is given to each dataset in the analysis, however this may not be so in a process involving several decision makers where each has a differing viewpoint on the relative importance of the dataset s A typical procedure to overcome this is to assign an agreed upon weight to each dataset so their relative importance is captured during the analysis. Several methods exist for developing a weighting system from using a Delphi method where group consensus among decision makers is tallied using a voting system (Collins et al., 2001, p. 614) or simply by the use of expert advice (Carr and Zwick, 2007, p. 59 60) Another alternative is pair wise comparison where each dataset is paired with another and the re lative preference of one over the other is assigned by each decision maker until all possible combinations have been compared. Values are then normalized and averaged to determine a weight for each dataset (Carr and Zwick, 2007, pp. 62 66). These techniq ues can also be used in decisions regarding the assignment of value ranges to individual dataset. One of the methods of forecasting future land use employed in this study uses a form of land use suitability analysis that was designed as a suite of analyt ical models to determine where potential conflicts may occur between competing land uses. Developed by Margaret Carr and Paul Zwick from the University of Florida during the 2000s the process known as LUCIS, or Land Use Conflict Identification Strategy w ill be described in more detail in the remainder of this section. Land Use Conflict Identification Strategy (LUCIS) GIS is a powerful tool planners use to better understand the incremental land use decisions that cumulatively affect humans and their natural environment. LUCIS was developed with the main purpose of highlighting where potential conflicts may occur in

PAGE 70

70 the future between competing uses of the land (Carr and Zwick, 2007, p. 7). LUCIS raster based land use suitability analysis (Carr and Zwick, 2005, p. 60) generally following the method describe d in the preceding para graphs. It also uses role playing techniques where major stakeholders ac t as teams of experts who are charged with rating all available land according to relative suit abilities that best s upport a specific lan d use they are advocates for (Carr and Zwick, 2005, p. 60) LUCIS uses three generalized land use categories that groups of stakeholders represent; urban, agriculture and conservation. Once each group of stakeholders rates the available lands according to their particular preferenc e, the three results are compared to see where potential conflicts may occur. More specifically LUCIS follows a six step process which is represented in Table 2 5. Table 2 5. The six step LUCIS process. Step Description 1 Define goals and objectives tha t become the criteria for determining suitability 2 Inventory data resources relevant to each goal and objective 3 Analyze data to determine a relative suitability for each goal 4 Combine the relative suitabilities of each goal to determine preference 5 Normalize and collapse the preferences for each land use into three categories high, medium and low 6 Compare the ranges of land use preference to determine likely areas of conflict The development of goals and objectives is an important part of the LUCIS process as it is with many other goal driven activities such as the development of comprehensive plans by local governments. Not only do goals and objectives give

PAGE 71

71 direction to plans and processes they are also useful for the creation of alternative plans or scenarios that can be measur ed and compared (Carr and Zwick, 2007, p. 84). LUCIS uses a hierarchical set of goals, objectives and sub objectives to drive the analytical process, with a unique set of goals created for each of the three land use categories. Table 2 6 shows an example of a subset of the goals, objectives and sub objectives that could be used for the agriculture category Table 2 6. Example of a subset of goals, objectives and sub objectives for the agriculture category Agriculture Goals Goal 1 Row Crops Identify the suitability for row crops 2 Livestock Identify the suitability for livestock 3 Specialty Farms Identify the suitability for specialty farms 4 Nurseries Identify the suitability for nurseries 5 Timber Production Identify the suitability for timber production Goal 1 Row Crops Identify the suitability for row crops Objective 1.1 Physical suitability Sub objective 1.1.1 Identify soils suitable for row crops 1.1.2 Identify land uses suitable for row crops Objective 1.2 Economic suitability Sub objective 1. 2.1 Identify land values suitable for row crops 1.2.2 Identify lands proximal to markets 1.2.3 Identify lands proximal to major roads Source pp. 86 87. Once goals and objectives have been identified and an inventory of data to be used is collected, suitability analysis is performed for all of the go als (Carr and Zwick, 2005, p. 62 ) The next step invo lves assigning preference to the suitability rasters, meaning each group of stakeholders must rate the individual goals according to their relative importance. LUCIS uses the technique of pair wise comparison, described in preceding paragraphs to assign w eights to the different goals within each land use category (Carr and Zwick, 2007, p. 63) For example, using the agriculture goals shown

PAGE 72

72 in Table 2 6, row crops may be the most valued type of agriculture so it would be assigned a higher weight than lives tock which may be less valued by the agriculture stakeholders. The end result of this step is the creation of three separate rasters one each for agriculture, urban and conservation, with grid cell values that represent the individual preference stakehol ders assigned to them. So that the three rasters may be compared each one is normalized and their values reclassified into three intervals reflecting, high, medium and low suitability. The final step of LUCIS, represented in Figure 2 5, involves the comb ination of the three preference grids so that each grid cell in the output raster has a unique value that represents the level of conflict between land uses in that location This information can be useful in a wide variety of ways to both planning professi onals and advocates for special interest groups, such as those supporting environmental conservation or agricultural preservation. By knowing the location and type of conflict that is likely to occur groups can more specifically target their efforts to pr eserve, conserve or change land uses in the future (Carr and Zwick, 2007, p. 164). For urban planners, LUCIS provides information that can be used at varying levels of complexity. At a broad level LUCIS can help to calculate whether appropriate levels of undisputed land uses exist to accommodate projected future growth. For example, if population growth is projected to increase by 200,000 people in the next 10 years, then using the existing gross urban population density the additional acres needed to ac commodate growth at the existing rate can be calculated. If that acreage is more than what LUCIS indicates is available in the undisputed urban category (where urban wins over agriculture and conservation) then urban land uses will likely encroach others

PAGE 73

73 and result in conflict. This knowledge is invaluable to planners as the formulate policies aiming to manage growth to minimize the loss of agricultural land and preserve future areas for conservation. Figure 2 5. Combining stakeholder preferences to show where conflict occurs Note. Diagram courtesy of author. A more complex use of LUCIS is as a tool for generating future land use scenarios that form the basis of a scenario planning process for identifying the impacts of different urban growth patt erns. This study use s LUCIS as the basis of a method for forecasting future land use scenarios that will subsequently be compared to a forecast generated using a different method. Projected population and employment growth for the period 2007 to 2025 wil l be allocated based on levels of land use conflict identified using

PAGE 74

74 LUCIS for Lake County, Florida. A detailed description of the allocation method follows in Chapter Three. Summary The relationship between land use and transportation in the United States is longstanding and complex. While history has shown that innovations in transportation technology have enabled increasing levels of mobility for most Americans their desire to live in low density suburban locations has simultaneously driven their need for greater mobility. At the centre of the land use and transportation relationship is the automobile and the high level of dependency most Americans have developed for it. Public concerns about the negative social and environmental impacts of trans portation and the urban growth it has fostered have led to government policies aimed at controlling and mitigating these effects. Research into the relationship between land use and transportation has been largely driven by these concerns. Despite a tren d of increasingly focused research into understanding this relationship there still remains a considerable gap in the understanding of dynamics. A coordinated approach to planning land us e and transportation may offer the greatest oppo rtunity for positive change yet a best approach to achieve coordination eludes both practitioners and researchers. Research into the relationship between land use and transportation helps guide decision makers in formulating policies aimed at minimizing ne gative effects of urban growth. The use of transportation and land use models plays an important role in understanding potential impacts of future changes to both the transportation system and urban growth patterns. The quantitative demands required of t hese models are predominantly a consequence of federal policies that more closely monitor the

PAGE 75

75 environmental impacts of transportation investments and their associated increases in automobile use. As more emphasis has been placed on the quantification and forecasting of potential future impacts land use and transportation models have become increasingly more sophisticated and complex, relying hea vily on computer simulations based on a large number of assumed variables. Based in mathematical and economic t heories, complex models often give the impression of being scientifically superior to less complex or rule based models. As a result, over time forecasts generated using more complex methods have become the basis from which future l ong range plans are dev eloped rather than being used as an assessment of outcomes of proposed plans. Instead of planning for a desired vision of the future emphasis is centered on planning for a future forecasted on assumptions about variables that cannot be known Emphasis given in the urban planning process to complex models and their resulting forecasts has come at the expense of simpler models that offer equally possible forecasts of the future The use of other methods of forecasting such as scenario planning techniques offer an opportunity to capture broader macro trends tha t more statist ically derived forecasts are unable to capture due to their narrower focus on micro level simulations. When supplemented with more sophisticated approaches to forecasting future land u se change, such as land use suitability analysis, simpler methods such as scenario planning may offer better opportunities to successfully coordinate long range land use and t ransportation plans than the more complex mathematically driven models in common practice today. That hypothesis is tested in this study which compares two methods of forecasting future land use change One

PAGE 76

76 method incorporates scenario planning techniques and suitability analysis to create a vision based on a set of broadly defined m utually supportive policy goals. The other method is mathematically derived used a gravity model and is based on historical development trends and prescriptive land use guidelines from existing policy. The following chapter details the methodology used to carry out this comparison.

PAGE 77

77 CHAPTER 3 STUDY AREA AND TIME FRAME Study Area Selection Florida has a history of rapid population growth. While the national population growth rate for the USA in the twentieth century was between 10 and 20 percent, the state of Florida experienced rates of between 20 and 80 percent (Smith, 2005, p. 2). astern and southwestern regions of Florida have experienced the most rapid rate of growth in the twentieth century, with decade averages of 77% and 61% respectively (Table 3 1), the t ion since the 1960s (Figure 3 1 ). This trend is projected to continue. The central region is expected to grow the most in terms of absolute population between 2010 and 2035, with 45% of new Floridians predicted to reside in the central region (Table 3 2) Table 3 percentage change per decade, 1900 to 2000 Region Northern Central Southeastern Southwestern Years Increase (%) Increase (%) Increase (%) Increase (%) 1900 to 1910 31.1 60.6 70.0 89.9 1910 to 1920 13.6 36.9 120.2 79.0 1920 to 1930 16.5 74.6 171.3 68.3 1930 to 1940 19.3 23.1 74.6 14.6 1940 to 1950 31.5 43.4 79.3 34.1 1950 to 1960 36.9 89.7 113.5 105.5 1960 to 1970 18.4 37.0 48.4 61.6 1970 to 1980 24.9 48.3 44.5 75.8 1980 to 1990 23.8 38.9 26.5 52.0 1990 to 2000 21.6 22.9 23.1 31.2 Average Increase 23.8 47.5 77.1 61.2 p. 23.

PAGE 78

78 Figure 3 1 2000. Adapted from Florida population growth: Past, present and future 22.

PAGE 79

79 Table 3 2010 to 2035 Region 2010 2035 Increase 2010 to 2035 Total Growth 2010 to 2035 Central 7,620 11,735 4,115 45% Northern 3,833 5,820 1,988 22% Southeastern 5,823 7,398 1,574 17% Southwestern 1,987 3,436 1,449 16% bulletin 153 Florida county population projections, 2008 Bureau of Economic and Business Research, & University of Florida Population Program, 2009. Within the central region Orange County is expected to grow by the largest amoun t for the period 2010 to 2035, increasing by approximately 736,000 people and accounting for almost 18% of total growth in the region (Figure 3 2). Figure 3 2 shows adjacent counties, such as Osceola, Polk, Brevard and Lake, will also experience substanti al population increases.

PAGE 80

80 Figure 3 2 Note. B population studies bulletin 153 Florida county population project ions, 2008 University of Florida Population Program, 2009 A comparison of projected gross population densities for counties in the central region provides insight into the effect population growth will have in specific locations in this region. For example, although Figure 3 2 shows Orange County will have the largest increase in absolute numbers of new residents between 2010 and 2035, its gross population density will only incre ase by 64% (Table 3 3). Conversely, Osceola County will have only approximately half the amount of growth as Orange County, and

PAGE 81

81 Lake County approximately one third, yet their respective increases in gross population densities are 117% and 84%. The differ ences in densities are even more pronounced when viewed for the period 2000 to 2035, with Lake County increasing by 163% and Osceola County by 263% (Table 3 3). This is a result of both Lake and Osceola Counties having relatively lower existing population s than neighboring Orange County and explains how a relatively low amount of population growth can still have dramatic effects in terms of changing population densities. Table 3 2010 to 2035. Gross Density (people per acre) Increase (%) Increase (%) Increase (%) County 2000 2010 2035 2000 to 2010 2010 to 2035 2000 to 2035 Sumter 0.15 0.25 0.68 90 139 353 Osceola 0.21 0.35 0.75 67 117 263 St. Lucie 0.54 1.21 1.71 125 42 218 Lake 0.33 0.47 0.86 43 84 163 Hernando 0.42 0.57 0.99 33 75 133 Marion 0.25 0.33 0.56 31 69 122 Indian River 0.36 0.47 0.78 29 66 115 Pasco 0.73 0.95 1.56 30 65 114 Orange 1.55 1.99 3.26 28 64 110 Citrus 0.30 0.38 0.59 25 56 95 Levy 0.05 0.06 0.09 24 57 94 Polk 0.41 0.50 0.78 24 54 90 Hillsborough 1.49 1.83 2.79 23 53 88 Brevard 0.73 0.88 1.26 19 43 71 Volusia 0.62 0.73 1.01 17 39 63 Okeechobee 0.07 0.08 0.11 14 33 52 Pinellas 5.30 5.43 5.95 2 10 12 Seminole 1.84 0.76 1.19 59 58 35 Note. bulletin 153 Florida county population projections, 2008 Bureau of Economic and Business Research, & University of Florida Population Pro gram, 2009. Gross density was calculated as gross area (land area less open water) divided by projected population. The significant increases in gross densities for Lake and Osceola Counties may be opportunities may be greater yet housing opportunities less affordable to specific groups

PAGE 82

82 of work ers Census Bureau data regarding travel patterns for journeys to work between these counties support this theory. Table 3 4 shows the residence and work counties for workers in Lake, Osceola and Orange Counties for 1990 and 2000. In particular it show s the number of workers travelling to Orange County from its adjacent counties, and from Lake and Osceola Counties to their adjacent counties. Although most workers live and work in the same county, a large number of Lake and Osceola residents travel to O range County for work, with Lake County experiencing the greatest relative increase of 135% between 1990 and 2000. In 2000 25% of workers in Lake County and 45% of workers in Osceola County traveled to Orange County to work. Table 3 4. R esidence and work counties for workers in Lake, Osceola and Orange Counties for 1990 and 2000 Residence County Work County 1990 2000 Change 1990 to 2000 % Change 1990 to 2000 Orange Orange 317,493 376,709 59,216 19 Osceola Orange 19 495 34,085 14,590 75 Lake Orange 8 516 20,009 11,493 135 Seminole Orange 70,609 80,875 10,266 15 Polk Orange 5,496 11,823 6,327 115 Brevard Orange 2,953 6,122 3,169 107 Lake Lake 42,777 51,842 9,065 21 Lake Orange 8,516 20,009 11,493 135 Lake Seminole 1,261 2,979 1,718 136 Lake Volusia 1,406 1,536 130 9 Lake Sumter 510 1,214 704 138 Lake Osceola 457 1,110 653 143 Lake Marion 658 629 29 4 Osceola Osceola 29,323 38,416 9,093 31 Osceola Orange 19,495 34,085 14,590 75 Osceola Lake 66 1,628 1,562 2367 Osceola Polk 362 560 198 55 Osceola Brevard 364 330 34 9 Osceola Indian River 49 35 14 29 Osceola Okeechobee 5 24 19 380 Census 2000 data 2000 County to by United States Census Bureau, 2008.

PAGE 83

83 surrounding lower density counties can be found in housing data for 2000 and 2008 from the United States Census Bureau. Table 3 5 shows the median value of owner occupied houses for Lake, Osceola and Orange County. Median values for both 2000 and 2008 were highest in Orange County and lowest in Lake County, with rates of increase similarly highest in Orange and lowest in Lake County. T h e median value of homes in Osceola was 8% les s than in Orange County in 2000 and by 2008 homes were more affordable in Lake County where their median value was 23% less than those in Orange County. Although housing affordability is only one potential fac tor this trend continue it may result in an increase in the number of Lake and Osceola County residents traveling to Orange County to work in the future. Table 3 5. Median values of owner occupied houses in 2000 and 2008 County Median Value ($) 2000 Median Value ($) 2008 Orange 107,500 243,400 Osceola 99,300 207,500 Lake 100,600 187,800 United States States Census Bureau, 2000. An increasing number of residents traveling into Orange County to work will have important implications for long range t ransportation planning in adjacent Lake and Osceola Counties. In Florida long range transportation plans (LRTPs) are prepared at a regional scale by agencies known either as a metropolitan planning organization (MPO) or transportation planning organizatio n (TPO). These organizations are responsible for the creation of 20 year transportation plans which are updated every five years that

PAGE 84

84 include strategies consistent with local and state planning objectives for the municipalities within their region (Florid a Department of Transportation, 2009, p. 4 2). Osceola and Orange Counties are within the same MPO boundary, Metroplan Orlando, meaning the transportation needs for these counties are planned jointly under the same LRTP. Transportation planning for the projected needs of Lake County however is address ed by the Lake Sumter MPO (LSMPO) under their own LRTP. How well these two MPOs coordinate their long range plans is particularly relevant to the enhancement of public transportation in Lake County which seeks to increase the proportion of choice riders, or those traveling to work. With one in four residents of Lake County traveling to Orange County to work, the need to coordinate transit plans is especially important if increased ridership is to be achieved. A MPOs long range transportation plan is statutorily required to be consistent with the future land use element and policies of the comprehensive plans of the local governments within its jurisdiction (FDOT, 2009a, p. 4 14). In the case of Lake County and the LSMPO, its LRTP must be consistent w ith one county and fourteen municipa l comprehensive plans. This represents a large number of plans to be consistent with considering the relatively low number of people the LSMPO serves, around 211,000 in 2000 (UFBEBR, 2009). Orange, Seminole and Osceola Counties which fall under 2000 but only 26 comprehensive plans with which to co ordinate. Figure 3 3 shows municipality, MPO and TPO boundaries for the central Florida region. Planning consistently across many jurisdictional boundaries may be particularly challenging for the LSMPO if the number of residents traveling to Orange County is isolated to just a

PAGE 85

85 few local governments rather than being spread more evenly across the region. Local governments with few potential choice riders may have less interest in creating future land use plans that support a future transit plan aimed specifically at increasing this type of ridership. Figure 3 3. Municipality, MPO and TPO Note. Map courtesy of author. Lake County was chosen as the area for this study f or several reasons. Firstly the county is situated in the most populou s region of the state and is projected to receive

PAGE 86

86 adjacency to rapidly growing Orange County creates opportunities for population growth r ange transportation plans both within Lake County and with Orange County. And thirdly, Lake County has a relatively high number of local governments within its MPO region, which may increase the potential for uncoordinated urban growth and reduce the abil ity of the LSMPO to reach some of its long range planning goals, particularly those concerning public transportation. These factors highlight a long range planning situation in Lake County that could benefit from the use of scenario analysis in the formul ation and coordination of its land use and transportation plans. Study Time Frame The time frame for this study was the period 2007 to 2025. This was determined according the time period for which the data used was relevant and for the planning horizons covered by the policy documents used in the study. It was necessary to select a point in time from which population and employment projections would be calculated for the scenarios to be compared in the study. The year 2007 was chosen as the beginning o f the time period because it was the year for which the most current property parcel data was available for Lake County. The end year of 2025 was chosen based on several factors. Firstly, of the two methods of forecasting land use change that are compa red in this study, one is that 2025. Secondly, the transportation demand model (TDM) used in the study is that used by FDOT in their state wide modeling of transportation demand The model can be run for selected future years of which 2025 is one. Thirdly, at the commencement of this

PAGE 87

87 study the most current comprehensive plan and therefore future land use map for Lake County was also for the period 2005 to 2025. And lastly, pro jected population and (BEBR) for several years including 2025. Since two of the main policy documents guiding this study describe plans up to the year 2025 and that a TDM specific to that year which uses data created by one of the methods under investigation, 2025 was deemed most appropriate year to end the study period.

PAGE 88

88 CHAPTER 4 METHODOLOGY Methodology Overview Thi s study uses land use and transportation modeling techniques to compare three future urban growth scenarios for Lake County, FL in the year 2025. Two different methods are used to allocate projected population and employment growth for the period 2007 to 2025 to specific locations within Lake County. One growth scenario uses the method adopted by the Lake Sumter Metropolitan Planning Organization (LSMPO) in the preparation of its 2025 Long Range Transportation Plan (LRTP). Two alternative growth scenari os use a method that combines the use of the Land Use Conflict Identification Strategy (LUCIS) and land use suitability analysis, and concentrate urban growth at varying densities around future transit routes. This study determine s which of the three scen arios best reflects two long range planning goals Lake County seeks to achieve; specifically a more compact urban form, and increased energy conservation by reducing single occupancy vehicle (SOV) dependency and increasing transit ridership. The findings of this study demonstrate which of the two methods results in a future land use pattern that was most supportive of the specified long range planning goals. By highlighting how different methods of forecasting future land use effect the achievement of lon g range planning goals this study will alert state, regional and local government planning agencies of the benefits alternative methods ca n have in improving the coordination of land use and transportation plans. Figure 4 1 is a diagram explaining the me thodology used in this study that will be discussed in detail in the following sections of this chapter.

PAGE 89

89 Figure 4 1. Diagram of study methodology.

PAGE 90

90 Step One The diagram of the stu dy methodology shown in Figure 4 1 is a generalized representation of the overall process and is divided into two halves One half refers to the creation of a scenario using the Florida land use allocation method (FLUAM) which is the forecasting method adopted by the LSMPO in the creation of the ir 2025 LRTP. The other half of the diagram refers to the creation of two scenarios using the LUCIS plus allocation method. LUCIS plus (LUCIS planning land use scenarios) is the name tentatively given to a population and employment allocation meth od currently under development in the Department of Urban and Regional Planning at the University of Florida by Dr. Paul Zwick (P. Zwick, personal communication, March 19, 2010). The methodology is further div id ed into three steps, of which this section d iscusses the first. Florida Land Use Allocation Method Scenario D ata It should be noted that the FLUAM scenario was not created in this study It is an accompanying dataset to the CFRPM version 4.5 provided by the FDOT District 5 for use in this study. The 2025 land use and socio economic data for Lake County used in the CFRPM was supplied to FDOT by the LSMPO (J. Banet, FDOT, personal communication, February 9, 2010). Chapter Five of the Lake Sumter 2025 LRTP describes the methodology used to forecast future land use and socio economic data for their 2025 plan (Lake Sumter Metropolitan Planning Organization, 2006, p. 5 1). Datasets used in the forecast are included in Chapter 5 of the LRTP (p p. 5 2 5 3) and are listed as follows : University of Florida BEBR population projections average of medium and high projections for 2025

PAGE 91

91 Vacant parcels from the year 2000 ; The Future Land Use Element from the Lake County Adopted Comprehensive Plan; Traffic analysis zones used in the CFRPM; Locations of master planned unit developments (PUDs) and developments of regional impact (DRIs) ; Environmentally sensitive land, such as the Green Swamp, Wekiva Basin, and the Ocala National Fo rest; Lakes and rivers; and Other lands that are not able to be developed. L and Use Conflict Identification Strategy Planning Land Use Scenarios D ata The LUCIS plus scenarios were created using similar types of data as the FLUAM scenario such as populatio n and emp loyment projections, parcel and traffic analysis data, and environmental data A comprehensive list of data used in the creation of the LUCIS plus scenarios is shown in Appendix C. A general list of data used is as follows: Population and house hold estimates for 2007 ; TAZ data for the years 2000 and 2012 ; A verage of medium and high population projections for 2025 ; Employment project ion s for 2005, 20 10 and 2025; Parcel data for 2007 ; Lakes and ponds ; Conservation areas ; LUCIS suitability and conflict rasters ;

PAGE 92

92 Visionary transit lines and stops for 2035 ; and Projected s chool enrolment data for 2025 The list shown in Appendix C shows the tabular and GIS data used, however other types of information were also used in the constructio n of the LUC IS plus scenarios such as policy guidelines. As t hese additional sources of information are used in the methodology they will be discussed at that point in more detail. Step Two The second step of the methodology involves the allocation of projected po pulation and employment to specific locations across the study area. This is achieved using different methods for the FLUAM scenario and the LUCIS plus scenarios The datasets created for each of these scenarios are input s to the third step of the method ology which involves the modeling of transportation demand for each scenario using the CFRPM. The CFRPM requires the datasets to contain a particular set of attributes necessary to calculate travel demand in the model. An overview of those attributes beg ins this section on Step Two and is followed by a more detailed explanation of both FLUAM and LUCIS plus and how they are used to create the three scenarios compared in this study Transportation Analysis Zone Data Overview The CFRPM requires socio econo mic data for each TAZ in the study area often referred to generally as TAZ data The TAZ data are both a reflection of forecasted land uses and the distribution of population and employment across those land uses. The TAZ data for the CFRPM are further divided into two sub sets. O ne dataset called ZDATA1 is used to calculate trip production s and accounts for residential land uses, their population and number of automobiles. The other dataset called ZDATA2 is used

PAGE 93

93 to calculate trip attractions and accounts for employment land uses and the number of employees in each location The CFRPM requ ires each dataset, ZDATA1 and ZDATA2, to be in a database file, or .dbf format Appendix D lists the attributes and their descriptions for each dataset. In this study, after population and employment have been allocated for each scenario the data is tabu lated so that a ZDATA1 .dbf and a ZDATA2.dbf exists for each of the three scenario s to be modeled with the CFRPM FLUAM the Florida Land Use Allocation Method FLUAM is the method used by the LSMPO for the creation of socio economic data used in the prep aration of their 2025 LRTP, and thus is the method that was used for the FLUAM scenario of this study, which uses the same dataset. Data processing occurs using both GIS vector and tabular data, with final data compiled in a tabular format. FLUAM is compr ised of four general steps (LSMPO, 2006, p. 5 1): 1. Development of control totals; 2. Determination of developable lands; 3. Input of approved developments, manual adjustments, and overrides; and 4. Allocation of growth to traffic analysis zones. Development of contr ol totals Tabular data are used to develop control totals for population, dwelling units and employment totals for the county which are based on historical growth rates and an average of medium and high projections from BEBR for 2025 (LSMPO, 2006, p. 5 2). School enrollment forecasts also are included in the control totals as well as hotel/motel dwelling units from approved developments (LSMPO, 2006, p. 5B 6).

PAGE 94

94 Determination of developable lands Developable lands are determined using parcel data in a GIS ve ctor format for the year 2000 as this is date of the last Census and therefore the most accurate population data available. Using GIS processes the parcel data, which contains land use codes used by the Florida Department of Revenue ( F DOR) is queried to locate vacant parcels and a GIS layer is created containing these parcels. This layer is combined with GIS data for environmentally sensitive lands, lakes and rivers, and other undevelopable land, which are subsequently selected and remove d from the layer. PUDs and DRIs are also removed as they are added in a later step of the process. TAZ and future land use data are combined with the layer so that all vacant parcels can be summarized according their future land use by TAZ number. A tab le is then created using the attributes of this GIS layer and the total acreage of vacant developable land uses available in each TAZ is summarized (LSMPO, 2006, p 5 3). Figure 4 2 shows a graphic representation of this part of the FLUAM process. Input of approved developments, manual adjustments and overrides For the development of factors used in the allocation process the study area is delineated into ten planning areas that have similar development characteristics based on the following (LSMPO, 2006, pp.. 5B 12 5B 13): Existing land use; Future land use; Existing population and employment; Location of cities; Major roadway corridors;

PAGE 95

95 Character of areas; and Functional relationship of land uses. Figure 4 2 Determination of developable lands in FLUA M Sumter 2025 Long by LSMPO, 2006, p. 5 4 Activity centers are then located for each planning area by calculating the amount of employment and dwelling units per TAZ. Depending on their amount of employment and d welling units each TAZ is assigned a weight and these weights are then combined for each planning area to determine where the centers of most activity are located (LSMPO, 2006, p. 5B 13). Densities and intensities of land uses are required by FLUAM to distribute population and employment into the developable lands layer. These are calculated for each planning area by comparing the existing built out densities for each land use and the maxi mum allowable densities contained in the Lake County Comprehensive Plan

PAGE 96

96 (LSMPO, 2006, p. 5B 10). For example, an existing urb an area that is fully developed, or built out, may have a residential density of four dwelling units per acre, whereas the Compreh ensive Plan indicates the maximum density allowed was five dwelling units per acre. By dividing the built out density by the maximum allowed density, a multiplier factor can be calculated for guiding density in future locations of that land use. In this example the density multiplier would be 0.8. Densities are estimated for each TAZ based on their Comprehensive Plan designation and multiplier factors applied to the unoccupied developable land. If for example a TAZ has five acres of developable land fo r residential uses and an approved density of four dwelling units per acre, then using a density multiplier of 0.8 the maximum number of dwellings for the vacant developable land in this TAZ is eight (i.e. 5 x 4 x 0.8 = 8). Employment density factors are calculated in the same way (LSMPO, 2006, p. 5B 11). By calculating the maximum allowable dwelling units and employment that vacant developable land can accommodate, the allowable growth for each TAZ is calculated. The maximum development each TAZ can acc ommodate is then calculated by adding the allowable growth to the existing development which is then used to determine if the allocated growth is physically possible within each TAZ. If the growth is not possible the model re allocates it to other TAZs (L SMPO, 2006, p. 5B 12). Allocation of growth to TAZs The allocation of dwelling units and employment is based on the control totals and the maximum allowable development for each TAZ, and is performed using a spreadsheet Allocations for PUDs and DRIs ar e input manually into the spreadsheet as the population and employment numbers are known for these areas, based on the information included in their development approvals. The remainder of the population

PAGE 97

97 and employment totals is allocated using a gravity model which assigns each TAZ an attractiveness factor (LSMPO, 2006, p. 5B 14) A gravity model is based on the concept that activity centers within a study area exert an influence on surrounding locations based on the activity centre size and number of a vailable activities. An attractiveness factor for each TAZ is calculated using the size of the nearest activity centre divided by the square of the distance of the TAZ to the activity centre TAZs with higher attractiveness factors are considered to deve lop faster than those with lower factors (LSMPO, 2006, p. 5 4) The attractiveness factor is used in combination with how close each TAZ is to other TAZ Growth is then allocated first to, and filling, TAZ s closest to the planning area centre in a way that simulates compact urban development (LSMPO, 2006, p. 5B 14) School enrollment is entered manually into the TAZs where each school is located. The school enrollment from the prior LRTP is reviewed and where determined, due to forecasted increased enrollments in specific locations, additional school locations are assigned. Approved developments that include new schools are added to the appropriate TAZ and the approved number of students is all ocated (LSMPO, 2006, p. 5B 15). Hotel/motel units are also allocated manually and are based on the number and location of units in 2000. Units created as part of approved developments are added, and likely locations of new units is determined through consultation with County and m unicipal staff (LSMPO, 2006, p. 5B 15). Finally, all of the socio economic data is tabulated into two .dbf fil es, ZDATA1 and ZDATA2 for inputting into the CFRPM. LUCIS plus The LUCIS Planning Land Use Scenarios Method LUCIS plus is a GIS raster based method for allocating regional population and employment growth. It is currently being developed by Dr. Paul Zwick at the University

PAGE 98

98 of Florida as a scenario planning tool for use in a wide range of planning fields, including environmental and conservatio n planning, transportation planning and hazard mitigation planning (P. Zwick, personal communications, 2010) The method employs the use of the ESRI software ArcMap and its associated ModelBuilder application A suite of models are created that produce r asters capable of representing different future land use scenarios Research and development of LUCIS plus has also included the creation of several complimentary tools that work in conjunction with the suite of models that enable planners to input scenar io parameters via simple GUIs (graphical user interfaces) containing drop down menus for the selection of desired variables (A Arafat, personal communication, March 15, 2010) Additionally, programs using Python and VB (Visual Basic) scripts have been de veloped as part of the LUCIS plus research so that the raster data can be summarized into a manageable tabular format for easier analysis and integration into other modeling frameworks, such as transportation demand and hazard simulation models This study uses LUCIS plus as a method for creating two scenarios however it does not use the complimentary tools or additional programs noted above Instead data inputs and variable p arameters are entered manually into models allocation fields within th e rasters are calculated manually, and output rasters are also manually summarized for the creation of ZDATA tables. There are differences between the FLUAM and the LUCIS plus methods however it is important to note one major difference between them befo re describing the study methodology further. The scale at which growth is allocated in both methods is vastly different. FLUAM allocates growth at the TAZ level, meaning future population and

PAGE 99

99 employment are distributed across zones that vary in size from approximately 20 to 40,000 acres The average size of a TAZ is approximately 2,800 acres. LUCIS plus allocates growth at a much smaller scale as it uses rasters containing cells that are 31 meters square, approximately one quarter acre in size to refle ct the average sized residential building lot In the LUCIS plus scenarios growth is allocated to raster cells and then aggregated to a TAZ level at the end of the allocation process. It is possible to aggregate data from the smaller scale LUCIS plus sc enarios up to the larger scale TAZ level so that it can be compared with data from the FLUAM scenario. However, it is not possible to disaggregate the larger scale FLUAM data down to a smaller scale for comparison with the LUCIS plus data. S maller scale data used in the creation of the FLUAM scenario, such as the specific parcel and land use data, was not available to this study. In the absence of such information any attempt to disaggregate t he larger scale data would require a large measure of estimati on that would nullify any meaningful comparison between the FLUAM and LUCIS plus scenarios. The comparison of scenarios in this study is therefore undertaken at the larger TAZ scale rather than the smaller quarter acre scale. F ollowing is a general list of steps used to create the two LUC IS plus scenarios for this study; each will subsequently be described in more detail Not all of the steps listed are strictly speaking a part of the LUCIS plus method. F or example the second, third and fifth steps invo lve the calculation of gross land use densities, control totals for population and employment growth, and specific growth patterns such as land use intensities These are variables needed for input into models used in LUCIS plus but

PAGE 100

100 the steps themselves are not a part of the allocation method and could be developed using a variety of different method s The steps used for the LUCIS plus scenarios are: 1. Creation of a combine raster that includes all the variables needed to identify specific areas suitable for growth; 2. Establish ment of existing gross densities ; 3. Establishment of control totals for population and employment growth; 4. Allocation of existing population and employment at gross densities; 5. Establish ment of specific guidelines for how population and em ployment growth will be spatially distributed ; 6. Allocation of population, employment and school enrollment to specific locations identified as suitable for growth. Creation of combine raster to identify areas suitable for growth The LUCIS plus method center s on the use of a final GIS raster that represents the study area and into which future population and employment is allocated. The fi nal raster, referred to as a combine raster, is a combination of several other independently generated rasters each conta ining information about the study area that are useful to planners when forecasting where future land use change might occur. processing tool. A description of the ArcMap geo processing e model used in the LUCIS plus method. o create a single output raster. All GIS vector and raster files contain an associated table in which information or attributes about the geographic space t he files represent is stored. The attribute table of any GIS raster contains rows and columns, with each row containing a

PAGE 101

101 unique record relating to geographic locations or cells in the raster and each column a unique attribute or item of information about the raster cells The output raster of a combine process contains columns that represent a unique item for each input raster and rows that contain unique combinations of the values o f the input rasters. Figure 4 3 illustrates how the values of mul tiple input rasters are attributed to an output raster during a combine process. The value column in the output raster contains a record for each unique combination of cells from the input ras ters. For the example shown in Figure 4 3 the combination of values of both the input rasters creates in the output raster a unique value of 1, seen in both the top left and bottom right cells of the output raster. In the attribute table of the output r aster this creates a row containing information about cells tha t contain the unique value of 1 (value = 1), of which there are two cells (count =2) and the original value from each input raster is retained in the columns bearing the names of the input rasters. Where a raster cell contains no data, meaning the locatio n the cell represents in the raster contains no information in the geographic space it relates to, it has a value of NoData. When any input raster cell is combined with a cell containing NoData the corresponding cell in the output raster will also contain NoData

PAGE 102

102 Figure 4 3 Allocation b uffers. In this study five combine rasters are created for each LUCIS plus scenario. In total 10 combine rasters are created; five fo r a low density LUCIS plus scenario and five for a high density LUCIS plus scenario. For each scenario the five combine raster s represents an allocation of population and employment into a specific zone or allocation buffer The combine rasters contain 3 1 meter by 31 meter cells, creating a high spatial resolution and increasing file size as the extent of each allocation buffer increases. To increase computational speed of the models and the allocation process it was determined that a separate combine ra ster for each allocation buffer would be necessary. Data from the five combine rasters are collated once all allocations have been made for each scenario to create the ZDATA1 and ZDATA2 .dbf files necessary for step three of the methodology. The five allocation buffer s are illustrated in Appendices E through H and include a description and rationale of the buffer distances. The location of transit stops used in the creation of the buffer zones are those contained in a GIS vector file created by the

PAGE 103

103 ECFRPC of visionary future transit routes and stations used in modeling future growth for their 2060 Strategic Regional Policy Plan (P. Laurien, personal communication, January 20, 2010). Combine raster model. Each comb ine raster is created using a sim ilar LUCIS plus model M odel inputs and out puts are the same for each of the combine rasters with the exception of the buffer raster which determines the spatial extent of the a llocation and the inclusion of DRI (developments of regional impact) rasters in buffers 4 and 5 Figure 4 4 shows the model of the combine raster for the allocation of buffer 1 and will serve as a n example for explaining the functioning of all the combine raster models. For ease of explanation the illustration of the model highli ghts three phases that will be described separately. When the model is run the processing occurs in the order the phases follow. Combine model phase one. Data inputs to the first phase of the model include a raster of the extent of buffer 1 and a ras ter of the opportunities for mixed use development. The buffer 1 raster was created by converting a vector shapefile of the area to a raster and then reclass ifying the cells so that only cells representing the location of buffer 1 contain a value and all other cells are as signed a NoData value. When this raster is combined with all other input rasters in the combine model all cells that overlap the NoData cells will also be assigned a NoData value, thereby masking those cells out of the output raster. Th is process is referred to as creating a mask. In the combine models of this study the buffer rasters act as masks for delineating where specific configurations of urban growth will occur.

PAGE 104

104 Figure 4 4. Model of combine raster for allocation buf fer 1 Note. Diagram courtesy of author.

PAGE 105

105 The raster representing the opportunities for mixed use development was created in a separate model that combines individual preference rasters for each of the following land uses; commercial, institut ional, multi family, retail and service which represent the mix of uses that are desirable in TOD areas. Each cell in the mixed use opportunities raster contains a 5 digit value that represent s the level of conflict or opportunity that exists between the different land uses in that cells location. For example a cell with a value of 33333 represents a good opportunity for mixed use as all of the land uses are highly preferred in that location, and a cell with a 11111 value has few mixed use opportunities. Using an extraction tool the values from the mixed use opportunities raster are extracted based on their location within buffer 1. Cells with a high opportunity for mixed use were extracted next. Only cells that had a high preference (3) for multi family combined with a medium (2) or high (3) preference for any of the other land uses were extracted. The suitability of areas for mixed use development is also influenced by the size or acreage of land under consideration. Areas of less than an acre are likely to be less desirable for this type of development. Therefore it is necessary to locate cells of high mixed use opportunity that are clustered into groups of raster cells of four or more; as each raster cell is approximately one quarter acr e. This is performed in the model ich is illustrated in Figure 4 5

PAGE 106

106 Figure 4 5 The region group process During the region group process the value of t he output raster cells represents a unique value that corresponds to its relationship with its neighboring cells. The process moves through the raster starting in the cell in the top left corner moving to the right and down through subsequent rows. The a ttribute table of the output raster contains a column listing the count of cells with t he same value. Using Figure 4 5 as an example, if the cell sizes had an area of one quarter acre then none of the cells are grouped in a one acre configuration, meaning in the attribute table none had a count of 4 or greater. In this example the highest count is 3, representing an area of only three quarters of an acre. At this point in phase one the output raster (Buffer 1 Reg Group) contains all the cells in buffer 1 that have a high opportunity for mixed use development. The cells also have a value that represents their relationship to neighboring cells so that groups of a certain size can be located. The final stage of phase one involves reclassifying the cells w ith a value of NoData to zero so that all data is retained during the subsequent combine process of the model.

PAGE 107

107 Combine model phase two. The second phase of the model uses the comb ine process described earlier. Fourteen rasters form the data inputs to t he process which represent the different attributes necessary for defining suitable locations for future urban growth. Seven of the fourteen rasters represent different land use suitabilities determined using the LUCIS process described in the literature review. The original suitability rasters were created by Dr. Paul Zwick of the University of Florida as part of a study for the ECFRPC in 2008 2009 concerning future land use change in the central Florida region. The seven rasters represent the suitabili ty for the following land uses : single family residential, multi family residential, commercial, service, retail, institutional, industrial. Each suitability raster contains values ranging from 1 to 9, with 9 representing the highest suitability and 1 t he lowest suitability. In the attribute table of each suitability raster therefore only 9 records or values exists, and each value has a count representing the number of cells with that unique value. In this study cells are selected based on their suitab ility using a tool that queries the attribute table to select cells with particular values, for example select all cells with a value equal to 9, or select all cells with a value of less than 7 but greater than 4. It is not possible therefore to select a fewer number of cells than there is in total for that value. For example, if only 90 cells were needed with a value of 9 but that row in the attribute table (with a value of 9) contained a total of 200 cells it would not be possible to select fewer than 2 00 using this selection method. Therefore it is necessary to reclassify the data so that there is a greater range of value s on which to base selections. A slice tool is used to reclassify

PAGE 108

108 the data into smaller slices, or more values in the attribute table. The seven suitability rasters were reclassified into 1,000 slices of equal area. A raster representing the level of conflict between the three major land use groups of agriculture, conservation and urban is included as an input. This raster was al so created by Dr. Zwick for the ECFRPC study mentioned ea rlier. The cell values of this raster represent the preference for each of the three land use groups, with a value of 333 corresponding to an area that is equally preferred by all uses, and a value of 111 representing an area that has a low preference for all three. A raster of land uses is included as an input so that existing population and employment can allocated according to their land use. The land use values correspond to the generalized TA Z land use categories listed in Appendix I A raster of the TAZs is also included so that each cell can be summarized in the final stage of the allocation process according to its TAZ. A buffer raster is included in the combine process that acts like a mask so that only cells located in that buffer are included in the output combine raster. The mixed use opportunities raster described in phase one is also included as is the region group raster that was a result of phase one. The remaining input raster is one that represents areas that have a potential for redevelopment. This was created by querying the vector parcel data to locate areas that met certain criteria and then exporting the file to a raster format. The selection criteria for redevelopment parcels are as follows: Cells within half mile buffer of future transit stations ;

PAGE 109

109 properties of the same land use sell for in an open market Parcels with buildings that will be more than 50 years old in 2025, i.e. were built prior to 1975 ; and Exclude parcels with the following uses churches, orphanages, hospitals, schools, sanitariums, homes for the aged, and cemeteries. Combine model phase three. The third phase of the LUCIS plus model involves the adding of necessary fields for the allocation proces s. These fields are as fol lows: MFPOP for multi family population SFPOP for single family population COMEMP for commercial employment SEREMP for service employment INDEMP for industrial employment EMPDEN for employment density POPDEN for population density NEW_LU f or new land use code ACRES for total acres The LUCIS plus model is then run five times for each scenario, each time using one of the five buffers as a mask.

PAGE 110

110 Establish ment of existing gross densities To identify areas suitable for future growth the size a nd distribution of existing population and employment must first be determined. GIS vector format p arcel data from the year 2007 was the most current land use data available during the study and forms the basis for establishing the extent and land use dis tributions of the urban for m in the study area The parcel data contains a land use code used by the F DOR which identifies individual land uses across a range 99 different uses. This study compares only five different land use categories; single family residential, multi family residential, commercial, service, and industrial uses. The F DOR codes are aggregated to reflect the five generalized land use categories of this study A ppendix I lists the F DOR codes aggregated to reflect TAZ land use codes. A table containing the TAZ land use codes is joined to the GIS parcel layer and the TAZ land use codes are copied over to a new field within the parcel layer called TAZ_LU. Land th at is not available for development is then removed from the parcel layer processing tools which removes areas from the parcel layer removed from the parce l layer in this way. Roads, rights of way and other undevelopable land uses such as undefined and centrally assessed parcels are removed by selecting them from within the parcel layer and deleting their records from table. The total number of acres within each of the TAZ land use categories is then calculated by summing the area of the individual parcels within each category. Table 4 1 shows the total acreage in the study area by TAZ land use categories. The total number of acres for each land use is the denominator in the

PAGE 111

111 calculation of gross density: Total population or employment in land use X Total acreage of land use X Table 4 1 Total acreage by TAZ land use categories. TAZ Land Use Category Total Acres Single family residential 67,734 M ulti family residential 1,482 Mobile homes residential 29,312 Commercial 3,256 Service 37,276 Industrial 161,044 Vacant residential 69,213 Vacant commercial 5,859 Vacant industrial 1,325 Vacant 29,969 Total acres 406 ,470 The numerator in the gross density calculation is the total population or employment in a particular land use. Population values are calculated by averaging data from the Census Bureau and available TAZ data. Census Bureau estimates of population and dwelling units by type of residence for the year 2007 are summarized according to their residential TAZ land use code; single family multi family or mobile home. These population totals are t hen divided by the total number of acres in each land use category to determine a gross density. Table 4 2 lists Census Bureau population estimates and the gross density for each residential TAZ land use Table 4 2 Census Bureau estimates for 2007 for Lake County population by residential TAZ land use categories TAZ Land Use Category Population Total Acres Gross Density Single family residential 171,335 67,734 2.53 Multi family residential 40,360 1,482 27.24 Mobile homes residential 73,272 29,312 2.51 Total population 284,967 98,528 2.89 Source United States Census Bureau, 2007.

PAGE 112

112 TAZ data is also used to supplement the Census Bureau population estimates. TAZ data for the years 2000, 2012 and 2025 is provided with the CFRPM and the datasets for 2000 and 2012 are compared to calculate an average annual increase in population and employment tot als over that period. The average annual increase is then applied to the 2000 TAZ data to project totals forward to the year 2007. Population and employment totals are summarized by TAZ land use category and gross densities calculated. Table 4 3 lists t he population estimates and gross densities; employment estimates are listed in a subsequent table. Table 4 3 TAZ estimates for 2007 for Lake County population by residential TAZ land use categories TAZ Land Use Category Population Total Acres Gross Density Single family residential 163,994 67,734 2.42 Multi family residential 38,747 1,482 26.15 Mobile homes residential 70,127 29,312 2.39 Total population 284,967 98,528 2.77 Gross density estimates from both the Census and TAZ data are averaged to establish the gross residential densities that will be used to distribute existing residential population across the study area Table 4 4 lists the existing gross residential densities for the study area Table 4 4 Existing g ross residential densiti es for study area TAZ Land Use Category Census TAZ Existing Gross Densities Single family residential 2.53 2.42 2.48 Multi family residential 27.24 26.15 26.69 Mobile homes residential 2.51 2.39 2.45 Gross density 2.89 2.77 2.83 Employment gross densities are calculated in a similar way by averaging two sources of estimates. Employment projections for the years 2005, 2010 and 2025 created by the ECFRPC for a 2008 2009 study of land use modeling techniques are used to establish employment estima tes for the year 2007. The 2007 estimate was

PAGE 113

113 based on the average annual increase in employment between the 2005 and 2010 estimates. The ECFRPC used the REMI Policy Insight model, an econometric input output model, to project employment across 23 categor ies. These employment categories are listed in Appendix J according to the ir TAZ land use category; commercial, service or industrial uses. Table 4 5 lists the gross employment densities using the ECFRPC employment projections. Table 4 5 ECFRPC estimat es for 2007 for Lake County employment by TAZ land use categories TAZ Land Use Category Employment Total Acres Gross Density Commercial 19,657 3,256 6.04 Service 75,681 37,276 2.03 Industrial 23,805 161,044 0.15 Total employment 119,143 201,576 0.59 Gross employment densities are c alculated using the 2007 TAZ data and are listed in Table 4 6 Table 4 6 TAZ estimates for 2007 for Lake County employment by TAZ land use categories TAZ Land Use Category Employment Total Acres Gross Density Commercial 28,286 3,256 8.69 Service 62,069 37,276 1.67 Industrial 22,211 161,044 0.14 Total employment 112,566 201,576 0.56 Gross density estimates from both the ECFRPC and TAZ data are averaged to establish the gross employment densities that will be used to distribute existing employment across the study area. Table 4 7 lists the existing gross employment densities for the s tudy area.

PAGE 114

114 Table 4 7 Existing gross employment densities for study area TAZ Land Use Category ECFRPC TAZ Existing Gross Densities Commercial 6.04 8.69 7.36 Service 2.03 1.67 1.85 Industrial 0.15 0.14 0.14 Gross density 2.89 2.77 0.78 Establish ment of control totals for growth To establish the amount of growth to be allocated into the future scenario s estimates of 2007 population and employment are deducted from forecasts for 2025. Table 4 8 lists the estimates and forecasts used and their ave rages that result in the control totals used for this study. Table 4 8 Population and employment control totals Population Projections / Estimates Total Average Control Total 2025 Population BEBR (med high) 449,050 2025 Population TAZ 463,500 2025 Population Average 456,275 2007 Population Census Bureau 288,580 2007 Population TAZ 272,860 277,224 Total Population Growth 2007 to 2025 179,051 Employment Projections / Estimates Total Average Control Total 2025 Employment ECFRPC 173,138 2025 Employment TAZ 185,580 179,359 2007 Employment ECFRPC 119,143 2007 Employment TAZ 112,566 115,855 Total Employment Growth 2007 to 2025 63,504 Allocation of existing population and employment using gross densities Using the attribute table of the combine raster for each of the buffers, cells are selected based on their existing TAZ land use category. Once cells are selected the field calculator tool is used to create a new value for the popula tion or employment field

PAGE 115

115 by multiplying the count of cells by the appropriate gross density The following script describes the calculation for the allocation of existing single family uses: [COUNT] (2.48/4.211) where [COUNT] represents the numbe r of cells with SF use 2.48 the SF gross density per acre and 4.211 the value by which the per acre density must be divided to reflect the 31 by 31 meter cell size. All cells with a TAZ land use of SF, MF, COM, SER and IND are selected and population or e mployment is allocated to them using their existing gross density. The NEW_LU field is also calculated to show a new land use code for each cell that reflects the TAZ land uses. This is done so that the data can be summarized at the end of the allocation process and also to show where which cells may have changed land uses. Table 4 9 shows the NEW_LU codes. Table 4 9 NEW_LU codes Land Use Category NEW _LU Land Use Category NEW_L U Single family SF 1 Vac. Commercial VACCOM 8 Mobile home MOB 2 Vac. Residential VACRES 10 Multi family MF 3 Agriculture AG 11 Industrial IND 4 Conservation CON 12 Commercial COM 5 Ag Preservation AGPR 13 Service SER 6 Mixed Use MU 14 Vac. Industrial VACIND 7 Vacant VAC 15 The attribute table of each of the five combine rasters is summarized so that the total number of population, employment and acres can be tabulated. Establish ment of guidelines for distribution of growth TOD g uidelines. The two LUCIS plus scenarios are examples of an urban growth pattern that is compacted within existing transportation corridors and clustered near future transit stations and route s to enhance the achievement of long range transportation g oals. The low scenario allocates at a lower densi ty and the high at a

PAGE 116

116 higher density. Draft guidelines created by the FDOT for the design of TODs are used to determine the appropriate mix of land uses, densities and intensities of development within a one quarter mile distance of transit stations. A ra nge of development standards are presented in the guidelines that reflects a variety of different urban contexts that, depending on the type of planned future transit such as light rail or rapid bus, would be most appropriate considering the existing and f uture land uses of the area (FDOT, 2009b, p. 2) (TDP) for 2009 to 2020, the future transit system planned is one that enhances existing fixed route bus transit, and incorporates new bus rapid transit an d commuter rail routes. Appendix K shows the planned transit system and a description of the modes and routes it includes. development, the TOD standard chosen to guide this study is one described in the FDOT design guidelines as T3. Appendix L shows the T3 standards. 4 6 shows an Excel spreadsheet used to calculate the densities for mixed use development within TOD areas for the low LUCIS plus scenario. This scenario uses the low value in the ranges suggested by FDOT for TOD population, employment and dwelling unit densities.

PAGE 117

117 Figure 4 6 Calculation of mixed use densities for the low LUCIS plus scenario. Figure 4 7 shows an Excel spreadsheet used to calculate the mixed use densities for the high LUCIS plus scenario. This scenario uses the median of the medium and high values in the ranges suggested by FDOT for densities in TODs.

PAGE 118

118 Figure 4 7 Calculation of mixed use densities for the high LUCIS plus scenario. Establishing the amount of growth to be allocated into each buffer was performed using an Excel spreadsheet that calculated the amount of growth in population and employment for the period 2007 to 2025. Growth is compacted in both the LUCIS plus scenarios to within a three mile distance of existing and future transit routes. Based on the existing proportions of single, multi family and mobile home populations within that distance, adjustments are made in both scenarios to increase proportions thereby comp acting growth. In 2007 forty percent of the total population lived within a three mile distance to transit and sixty percent outside of that distance. Sixty percent of the total 2007 population was in single family residential, fourteen percent in multi family, and

PAGE 119

119 twenty six percent in mobile homes. Figures 4 8 and 4 9 show Excel spreadsheets used to determine how growth is to be allocated for both the low and high scenarios, respectively. Figure 4 8 Guidelines for allocation of growth for low LUCIS plus scenario

PAGE 120

120 Figure 4 9 Guidelines for allocation of growth for high LUCIS plus scenario Allocation of growth to suitable areas based on established guidelines Order of allocations. In both LUCIS plus scenarios growth is allocated in order by buffer, beginning with buffers 1 and 2 which represent TOD areas, followed by buffer 3 which represents commercial corridors, then buffer 4 which includes remaining areas within a three mile distance of transit, and finally buffer 5 which is the remaind er of the study area.

PAGE 121

121 Using a spreadsheet into which are entered the existing and control totals for population and employment, a total amount of growth to be allocated for each buffer is determined and can be further disaggregated by land use according to the established guidelines. As growth is allocated the amounts are deducted from the control totals to determine how much growth remains to be allocated. Figure 4 10 shows an Excel spreadsheet that illustrates this process. Figure 4 10 Calculation of remaining growth to be allocated

PAGE 122

122 For example, if the total amount of population growth for TOD areas (buffers 1 and 2) is 28,740 then beginning in buffer 1, a selection of the most suitable locations can be made, and based on that area the amount of population that can be allocated can be determined by divi ding that area by the population or employment density. Thus if 1,000 suitable cells are located in buffer 1 for multi family population (i.e. 237.43 acres) then using the TOD density for multi family (54.59 people/acre) a total of 12,963 people can be al located in buffer 1. The remaining 15,777 people (28,740 less 12,963) need to be allocated in buffer 2. allocations so that a running total is kept to guide what remains to be allocated. Once all growth h as been allocated in the buffer the attribute table of the combine raster for that buffer is summarized and the new population and employment totals are inserted into the spreadshee t of control totals (Figure 4 10 ) to determine what remains to be allocate d in subsequent buffers. Selection of suitable areas and allocation of growth. The selection of suitable areas for allocating growth is steered firstly by the established guidelines and secondly by land use suitability analysis. The allocation of growth to buffer 1 will be described in detail as an example of how this process was undertaken for all of the buffers, and then an overview of the selection process for the remaining buffers is provided Appendix M Example allocation t o buffer 1. Using the control totals the amount of population and employment growth to be allocated is determined. As buffers 1 and 2 are TOD areas all population and employment growth is first to be allocated into suitable locations for mixed use redeve lopment and then into vacant land uses (VACRES, VACCOM, VACINST or VAC). In this analysis all mixed use development

PAGE 123

123 will incorporate both residential and commercial/service components. The residential component is multi family population and the employme nt component is divided between commercial and service according to their relative proportion of total employment growth. Twenty five percent of all non industrial employment growth is forecasted to be commer cial and seventy five percent service, so in mi xed use areas the employment growth is distributed according to those percentages. In the attribute table of the buffer 1 combine raster, a selection by attribute is performed to firstly locate cells within redevelopment areas ( RCLREDVBF1_2 > 0 ). From t his selection, cells that are grouped in areas of one acre or more (four or more cells) are located (RREGIONBUFF1 >= 4 ). From this selection, cells that have a medium to high mixed use opportunity are selected (RCINMFRSCONF = 3333 3 OR RCINMFRSONF = 23333 For the selected cells the field calculator tool is used to allocate population and employment at the mixed use densities into the appropriate fields (MFPOP, COMPOP and SERPOP). A NEW_LU code of 14 (MU) is also calculated. Th e count of cells in the selection is entered into the buffer 1 working spreadsheet and the remaining number of cells to be allocated is calculated. The remaining growth for buffer 1 is to be allocated into vacant land uses (VAC, VACRES, VACCOM and VACINS T). This selection starts by selecting all cells that have not already been allocated in the previous step (NEW_LU = 0), then selecting all From this selection all cells that have a medium to high mixed use opportunity are selected, followed by those grouped in areas one acre or larger. Using the field calculator,

PAGE 124

124 population and employment is allocated to the appropriate fields and an NEW_LU code of 14 is given. The count of cells in the s election is entered into the spreadsheet and if any growth remains unallocated it is carried over to buffer 2. Finally the attribute table of the combine raster is summarized and the new population and employment totals are added to the control totals spr eadsheet (Figure 4 10 ) so that an accurate count of remaining growth can be determined. Allocation to remaining buffers. Allocation of growth to the remaining buffers continues in a similar way. Opportunities for mixed use redevelopment exist in buffer s 1, 2 and 3 and are selected first in those buffers, followed by vacant land uses based on their level of land use conflict, followed by the suitability for individual land uses. Mixed use opportunities d o not exist in buffers 4 and 5 so selection in those buffers begins with allocation into DRIs and the remainder according to land use conflict and land use suitability. The method of allocation is the same for both the low and high LUCIS plus scenarios with the only difference between them being their level of compactness (as determined by their allocation guidelines) and their mixed use densities. Appendix M lists the general steps and criteria used for allocation in all buffers. Creation of ZDATA tables. Once all growth has been a llocated, summaries of all the combine raster attribute tables are collated into one spreadsheet. The data is then summarized according to TAZ number and the data separated according to the required attributes of the ZDATA1 and ZDATA2 .dbf files. Appendi x D lists those attributes and highlights in bold type the attributes that are derived from the LUCIS plus method. The dwelling units totals (SFDU and MFDU) are calculated by dividing the SFPOP and MFPOP by the average household size of 2.48 people. With the exception of school

PAGE 125

125 enrolment, t he remaining fields are carried over from the 2025 TAZ data that accompanies the CFRPM. School enrolment is calculated outside of the LUCIS plus method and is done using a combination of GIS and spreadsheet functions. Allocation of school enrolment. School enrolment is calculated and allocated following several steps. Using a GIS vector file of the school area zones, the 2025 population (as allocated in each LUCIS plus scenario) is summarized. Then using data from t he Census Bureau 2007 household estimates the percentage of school aged children is determined according to school type (elementary, middle and high school). These percentages are used to determine the number of school aged children in each school area z one according to school type. Figure 4 11 shows a spreadsheet used to calculate school enrolment for 2025 for the low LUCIS plus scenario. A similar calculation is done for the high LUCIS plus scenario. TAZs where the individual schools are located are selected in the ZDATA2 .dbf file and the enrolment allocated.

PAGE 126

126 Figure 4 11 Calculation of school enrolment for 2025 for the low LUCIS plus scenario Step Three The third and final step of the methodology is the modeling of transportation demand for the three future land use scenarios created in the previous step. The study uses the CFRPM version 4.5 which is run using the Citilabs software CUBE/VOYAGER. Inputs to the modeling process that are unique to each scenario include TAZ data (ZDATA1 and ZDATA2 dbf files) and a transit route system (TROUTE.lin) Other inputs to the CFRPM, such as the highway network and its associated characteristics are not altered from those contained in the 2025 scenario of the CFRPM and remain the same across all scenarios.

PAGE 127

127 FLUAM S cenari o TAZ D ata a nd Transit S ystem The FLUAM scenario uses the 2025 TAZ data files that accompany the CFRPM. This scenario is modeled using the 2025 transit system that accompanies the CFRPM, refer red to in this study as transit system A. Figure 4 12 is a map of transit system A routes and stops This system is comprised of six local bus routes, each operating on 60 minute headways in both peak and non peak hours. Headway indicates the frequency of service. In the CFRPM headway1 refers to peak times and headway2 to non peak times. Table 4 10 lists the modes and headways for each route in transit system s A and B Table 4 10 Transit systems A and B routes, modes, and headways Transit System A Transit System B Route Mode Hea dway1 Headway2 Mode Headway1 Headway2 Lake 1 Local bus 60 min 60 min Local bus 30 min 60 min Lake 2 Local bus 60 60 Local bus 30 60 Lake 3 Local bus 60 60 Local bus 30 60 Lake 4 Local bus 60 60 Local bus 30 60 Lake 5 Local bus 60 60 Local bus 30 60 Lake 6 Local bus 60 60 Local bus 30 60 Lake 7 Local bus 60 60 Lake 8 Local bus 30 60

PAGE 128

128 Figure 4 12 Transit system A routes and stops. Note. Map courtesy of author. LUCIS Plus S cenarios TAZ Data and Transit S ystem The LUCIS plus scenarios use their respective TAZ datasets created in the previous step of this methodology. B oth LUCIS plus scenarios use the same transit system which was created to resemble the system outlined in the Lake County TDP and shown in Appendix K It is referred to in this stud y as transit system B. Figure 4 13 shows a map of transit system B routes and stops. This system is comprised of eight local bus routes, two more than transit system B. Routes 1 through 5 remain unchanged from transit system A, keeping the same route configuration and number of

PAGE 129

129 stops. Route 6 differs from transit system A in that it is extended at its southern end to join Route 1 and contains four additional stops This was done to resemble the bus rapid transit cross county connector route planned in the TDP. Two additional routes were added to this system ; Route 7 which resembles the configuration of the local bus route proposed to connect Altoona and Zellwood, an d Route 8 which resembles the route of the commuter rail line planned from Zellwood to Eustis. The transit system contained in the TDP contains two additional modes than those that were created in transit system B; bus rapid transit for Routes 1, 2 and 6, and commuter rail for Route 8 In transit system B these modes are replaced with local bus service. Due to the time and technical constraints of this study it was not possible to process the model using those modes for the routes indicated. To compensa te for some difference in the level of service the local bus service for these routes were enhanced with additional stops and headway times were decreased C entral Florida Regional Planning Model Model R un for Each S cenario It should be noted that the CFRPM is a regional TDM that encompasses a ten county area comprised of Brevard, Flagler, Lake, Marion, Orange, Osceola, Polk, Seminole, Sumter and Volusia counties. When the CFRPM is run in this study it models the whole region and it is necessary to ext ract the Lake County data from the model results. The CFRPM contains three transportation modeling scenarios, a base scenario that models known data from 2000, and two future scenarios for 2012 and 2025 which rely on forecast data This study uses the year 2025 transportation scenario and models all three future land us e scenarios by altering the TAZ data (ZDATA1 and ZDATA2 .dbf files) and transit route data (TROUTE.lin) used as inputs into the model

PAGE 130

130 Figure 4 13 Transit sys tem B routes and stops. Note. Courtesy of author. The model is first run for the FLUAM scenario as no changes to the model inputs are required since the FLUAM scenario uses the existing TAZ and transit system data of the year 2025 transportation scena rio. The CFRPM produces a number of reports that quantify the results generated by the different steps of the model. This study uses results from the highway assignment report, specifically the total VMT, and the mode split report which reports the number of highway and transit trips according to different modes and trip purposes The CFRPM creates reports in an HTML format. After the

PAGE 131

131 model run is complete t hese reports are opened using a web browser and then copied into a spreadsheet for further analysi s. The model is run a second time for the FLUAM scenario using the same TAZ data (ZDATA1 and ZDATA2) but replacing the transit system A data (TROUTE) file with the transit system B data. This is done so that any effect the different transit system s has on VMT and mode choice can be ascertained. When the model run is complete the highway assignment and mode split reports are copied into a spreadsheet for further analysis. Each LUCIS plus scenario is run two times, both with their respective TAZ data file s created in the previous step of the methodo logy. One model run however uses the transit system A data and the other uses the transit system B data. After each model run the highway assignment and mode split reports are copied into a spreadsheet for fur ther analysis. The data from each of the six model runs is then summarized in to a single spreadsheet for the analysis of this study.

PAGE 132

132 CHAPTER 5 FINDINGS This study compares methods of forecasting future land use change to determine what affect diff erent methods may have on the achievement of long range planning goals. Four interrelated land use and transportation goals were selected for this study: 1. Discourage urban sprawl through a future land use pattern that promotes orderly, compact development; 2. Increase energy conservation as measured in vehicle miles of travel ; 3. Reduce dependency on single occupant vehicles; and 4. Increase transit ridership. By comparing several data outputs from the different forecasting methods and the transportation demand mod el that uses them, the extent to which each planning goal was achieved can be measured. Compact Development The extent to which a land use pattern is compact can be measured according to the number and density of people and employment located within specific zones within a region. In this study the compactness of urban form is measured across two zones; one that includes the area of land within a 3 mile radius of future tr ansit routes, and the other the area outside that distance Figure 5 1 shows t he two zones. Table 5 1 shows the distribution of population and employment across the two zones according to the scenarios created using the different forecasting methods The distribution of population and employment is also shown for TAZ data for the year 2007 for comparative purposes.

PAGE 133

133 Figure 5 1. TAZs within and outside a 3 mile distance of future transit. Note. Map courtesy of author.

PAGE 134

134 Table 5.1 shows that the LUCIS plus high scenario has the most compact land use pattern compared to the oth er two scenarios. The LUCIS plus high scenario has 60% of the study area population and 59% of employment located within a 3 mile distance of future transit routes Th e LUCIS plus high scenario has 10 % more population living in the 3 mile zone than the L UCIS plus low scenario and 11% mo re than the FLUAM scenario. The LUCIS plus high scenario also has 2% more employment in the 3 mile zone than the FLUAM scenario and 5% more than the LUCIS plus low scenario. Table 5 1. Population and employment distributi on FLUAM and LUCIS plus scenarios compared to 2007 TAZ data W ithin 3 miles of transit 2007 TAZ FLUAM LUCIS plus Low LUCIS plus High Single Family 116,559 182,239 95,881 80,647 Multi Family 39,779 46,023 134,287 192,177 Buffers 1 to 4 Population 156,338 228,262 230,168 272,824 % of Total Population 57% 49% 50 % 60% Commercial 20,478 31,865 26,897 30,224 Service 45,688 49,353 63,744 70,502 Industrial 12,669 17,488 5,986 5,990 Buffers 1 to 4 Employment 78,835 98,706 96,627 106,715 % of Total Employment 70% 57% 54% 59% Outside 3 miles of transit 2007 TAZ FLUAM LUCIS plus Low LUCIS plus High Single Family 88,797 193,196 132,507 101,870 Multi Family 31,048 42,042 93,840 83,149 Buffer 5 Population 119,845 235,238 226,347 185,019 % of Total Population 43% 51% 50% 40% Commercial 7,808 17,128 10,213 7,070 Service 16,381 39,615 46,132 39,831 Industrial 9,542 16,288 26,352 26,362 Buffer 5 Employment 33,731 73,031 82,697 73,263 % of Total Employment 30% 43% 46% 41%

PAGE 135

135 The compactness of the land use pattern can also be measured in terms of the density of population and employment in the two zones Table 5 2 shows the average TAZ population and employment densities within the different buffer zones shown in Appendices E through H The average TAZ densities were determined by calculating the population and employment densities for each TAZ within a buffer zone and then calculating the average of those densities. Table 5 2 shows that the highest average population and employment densities within a 3 mile zone of future transit routes that is within buffers 1, 2, 3 and 4, exist in the LUCIS plus high scenario It also has the lowest average population and employment densities outside of that 3 mile zone. Table 5 2. Average TAZ population and employment densities by buffers Population Densities FLUAM Scenario LUCIS plus Low Scenario LUCIS plus High Scenario people per ac people per ac people per ac Buffer 1 & 2 TAZs 4.0 7.6 13.3 Buffer 1,2 & 3 TAZs 4.0 5.2 7.7 Buffer 1,2,3 & 4 TAZs 3.5 4.2 6.0 Buffer 5 TAZs 1.2 0.9 0.8 Employment Densities FLUAM Scenario LUCIS plus Low Scenario LUCIS plus High Scenario jobs per ac jobs per ac jobs per ac Buffer 1 & 2 TAZs 4.0 1.7 5.1 Buffer 1,2 & 3 TAZs 3.0 1.5 2.8 Buffer 1,2,3 & 4 TAZs 2.4 1.4 2.2 Buffer 5 TAZs 0.4 0.3 0.3 Energy Conservation Energy conservation can be measured by determining how many vehicle miles of travel (VMT) have been reduced by a particular land use pattern. VMT is a measure of total traffic on a road and is calculated as the product of the average daily traffic count a nd the length of the road (FDOT, unknown). In this study the three land use scenarios were each modeled using two different transit routes. To ascertain the extent to which

PAGE 136

136 the different transit systems may have reduced VMT the systems are compared acros s the different land use scenarios. Table 5 3 shows that in terms of reducing daily volumes of VMT there was little difference between the transit systems, with all differences being less than two tenth s of a percent. Table 5 3 Daily volume of v ehicle miles traveled in Lake County comparison of transit system s by land use scenario FLUAM LUCIS plus Low LUCIS plus High Transit System VMT VMT VMT Transit System B 13,221,000 12,876,000 12,489,000 Transit System A 13,215,000 12,892,000 12,486,000 Difference between systems 6,000 16,000 3,000 % change 0.05% 0.12% 0.02% As the difference that the two transit systems made to VMT is only very slight the remainder of the results will only display those where transit system B was modeled. Table 5 4 displays a comparison of VMT across all of the scenarios and shows that the L UCIS plus high scenario ge nerated the least amount of VMT. It resulted in 3% fewer VMT than the LUCIS plus low scenario and 5.5% fewer than the FLUAM scenario. Table 5 4 Daily volume of v ehicle miles traveled in Lake County comparison of all scenarios for transit system B Transit System B Land Use Scenario VMT FLUAM scenario 13,221,000 LUCIS plus low scenario 12,876,000 LUCIS plus high scenario 12,489,000 % difference between FLUAM & LUCIS plus low 2.6% % difference between FLUAM & LUCIS plus high 5.5% % difference between LUCIS plus low & LUCIS plus high 3.0% It is possible to illustrate the difference in VMT across the scenarios according to the type of area in which the VMT is generated. Table 5 5 shows the VMT for the three land use scenarios according to area type. Roads in the CFRPM are classified

PAGE 137

137 according to area type and the VMT generated by area type road s is detailed in the model s highway assignment report. The location of roads by area type used in the modeling of transportation demand of all the scenarios is shown in Appendix N Rural areas create approximately 70% of all VMT in all scenarios and CBD area s create the least with around 2% in all scenarios. Residential areas and outlyi ng business districts each account for between 10 12% of VMT across all the scenarios. Table 5 5 Vehicle miles traveled in Lake County by transit s ystem B comparison of LUCIS plus l ow and FLUAM scenarios by area type FLUAM Scenario LUCIS plus Low Scenario LUCIS plus High Scenario Area Type VMT VMT VMT CBD 252,000 268,000 254,000 CBD Fringe 733,000 726,000 744,000 Residential 1,557,000 1,356,000 1,329,000 Outlying Business Districts 1,488,000 1,396,000 1,561,000 Rural 9,192,000 9,131,000 8,600,000 Total 13,221,000 12,876,000 12,489,000 A pair wise comparison of the three scenarios demonstrates how VMT changed across the various area types between the different scenarios. A comparison of the FLUAM and LUCIS plus low scenarios shows that the largest decrease in VMT occurred in residential areas, followed by the outlying business districts and rural areas. A small decrease (1%) occurred in CBD fringe areas and in CBD areas VMT increased by 6.3% Table 5 6. Vehicle miles traveled in Lake County by transit System B comparison of the FLUAM and LUCIS plus low scenarios by area type FLUAM Scenario LUCIS plus Low Scenario % Area Type VMT VMT Difference Change CBD 252,000 268,000 16,000 6.3 CBD Fringe 733,000 726,000 7,000 1.0 Residential 1,557,000 1,356,000 201,000 12.9 Outlying Business Districts 1,488,000 1,396,000 92,000 6.2 Rural 9,192,000 9,131,000 61,000 0.7 Total 13,221,000 12,876,000 345,000 2.6

PAGE 138

138 A comparison of the least and most compact scenarios, i.e. the FLUAM and LUCIS plus high scenarios (Table 5 7), shows that the greatest reduction in VMT of occurred in rural areas, followed by the residential areas I n all other areas VMT increases Table 5 7. Vehicle miles traveled in Lake County by transit System B comparison of the FLUAM and LUCIS plus high scenarios by area type FLUAM Scenario LUCIS plus High Scenario % Area Type VMT VMT Difference Change CBD 252,000 254,000 2,000 0.8 CBD Fringe 733,000 744,000 11,000 1.5 Residential 1,557,000 1,329,000 228,000 14.6 Outlying Business Districts 1,488,000 1,561,000 73,000 4.9 Rural 9,192,000 8,600,000 592,000 6.4 Total 13,221,000 12,489,000 732,000 5.5 A comparison of the two LUCIS plus scenarios shows that the more compact sce nario, i.e. LUCIS plus high, has the greatest reduction in VMT in rural areas followed by the residential and CBD areas. The LUC IS plus low scenario however has 165,000 fewer VMT in outlying business districts than the high scenario. Table 5 8. Vehicle miles traveled in Lake County by transit System B comparison of LUCIS plus low and LUCIS plus high scenarios by area type LUCIS plus Low Scenario LUCIS plus High Scenario % Area Type VMT VMT Difference Change CBD 268,000 254,000 14,000 5.2 CBD Fringe 726,000 744,000 18,000 2.5 Residential 1,356,000 1,329,000 27,000 2.0 Outlying Business Districts 1,396,000 1,561,000 165,000 11.8 Rural 9,131,000 8,600,000 531,000 5.8 Total 12,876,000 12,489,000 387,000 3.0

PAGE 139

139 Single Occupant Vehicle and Transit Trips A reduction in single occupant vehicle (SOV) trips and an increase in transit trips can be measured using outputs from the mode split report generated by the CFRPM. The report shows the daily volume of trips according to trip types, either home based work (HBW) trips or non home based (non HBW) work trips. HBW trips reflect peak hour trips and non HBW off peak hour trips (J.Banet, FDOT, personal communication, January 14, 2010). Table 5 9 show s the daily volumes of trips, including HBW (peak) and non HBW (off peak) trips across all scenarios. In all scenarios the HBW trips account for 16% and non HBW trips 84% of total trips. The LUCIS plus high scenario has the lowest number of total trips of all the scenarios 1.8% fewer than the LUCIS plus low scenario and 7.0% lower than the FLUAM scenario It has 3% fewer total HBW trips than the LUCIS plus low scenario and 11% fewer than the FLUAM scenario. The LUCIS plus high scenario has 2% fewer total non HBW trips than the LUCIS plus low scenario and 6% fewer than the FLUAM scenario. Single occupant ve hicle (SOV) or drive alone trips account for the largest proportion of all trips, making up 81% of HBW trips and 44% of non HBW trips. Shared ride car trips with 2 people account for 13% of HBW trips and 35% of non HBW trips, and shared rides with 3 or mo re people 6% of HBW trips and 20% of non HBW trips. The LUCIS plus high scenario has the fewest SOV trips, 0.8% fewer than the LUCIS plus low scenario and 7% fewer than the FLUAM scenario. Transit trips account for the smallest proportion of trips comprising only 0.004% of total HBW trips and 0.08% of total non HBW trips. The LUCIS plus high scenario has 2% more HBW transit trips than the LUCIS plus low scenario and 6% fewer than the

PAGE 140

140 FLUAM scenario. It also has 3% fewer non HBW trips than the LUCIS plus low scenario and 11% fewer than the FLUAM scenario. Table 5 9 Dai ly volume of home based work and non home based work t rips in Lake County by mode all scenarios HBW trips FLUAM Scenario LUCIS plus Low Scenario LUCIS plus High Scenario Drive alone 232,261 213,288 208,990 Drive Shared ride 2 37,743 35,340 33,399 Drive Shared ride 3+ 18,130 17,226 15,173 Transit 242 212 255 Total HBW Trips 288,376 266,067 257,818 Non HBW trips Drive alone 652,172 615,256 613,123 Drive Shared ride 2 513,772 486,554 480,950 Drive Shared ride 3+ 295,137 289,163 275,565 Transit 33 58 56 Total Non HBW trips 1,461,115 1,391,032 1,369,695 All trips Drive alone 884,433 828,544 822,133 Drive Shared ride 2 551,515 521,894 514,349 Drive Shared ride 3+ 313,267 306,389 290,738 Transit 275 270 311 Total Trips 1,749,491 1,657,099 1,627,513 Summary The most compact urban growth pattern was achieved using the LUCIS plus method. The LUCIS plus high scenario had the most compact urban form with 60% of total population and 59% of total employment existing in TAZs intersected by a 3 mile buffer of future transit routes. Population and employment densities within that area were 6.0 people per acre and 2.2 jobs per acre The least compact scenario is created using the FLUAM method which had 49% of total population and 57% of total employment existing in T AZs intersected by a 3 mile buffer of future transit routes Population densities within that area were lower than the LUCIS plus high scenario,

PAGE 141

141 being 3.5 people per acre however the employment density was slightly higher than the LUCIS plus high scenario, being 2.4 jobs per acre. The greatest reduction in VMT was achieved using the LUCIS plus method. The study modeled two transit systems, one with two additional routes and higher frequency of stops than the other The difference in VMT between both transit systems was very small, being less than two tenths of one percent. The LUCIS plus high scenario generated the fewest VMT, 5.5% fewer than the FLUAM scenario and 2.6% fewer than the LUCIS plus low scenario. W hen the most compact and least compact scenarios are compared, i.e. the LUCIS plus high and FLUAM scenarios, the greatest reduction in VMT occurred in rural areas followed next by residential areas. The VMT in all other areas increased. The greatest poten tial reduction in highway trips was achieved using the LUCIS plus method. The LUCIS plus high scenario generated the fewest total trips, 1.8% fewer than the LUCIS plus low scenario and 7% fewer than the FLUAM scenario. The fewest SOV trips are generated by the LUCIS plus high scenario, 0.8% fewer than the LUCIS plus low and 7% fewer than the FLUAM scenarios. The LUCIS plus high scenario also generated the most transit trips, 13% more than the FLUAM and 15% more than the LUCIS plus scenarios.

PAGE 142

142 CHA PTER 6 DISCUSSION Discussion of Findings and M ethods This study aims to assess the affect methods of forecasting future land use have on achieving long range planning goals. The study uses several long range planning goals that can be measured to determine the affect the method of forecasting land use change has on the achievement of the stated goals. The planning goals the study uses are ones often cited in Smart Growth initiatives goals that aim to combat sprawling development patterns and decrease travel demand by encouraging the use of public transportation to reduce dependency on SOV and conserve energy by redu cing VMT. The literature shows that urban form and travel demand are inextricably linked and that several land use characteristics, such as density, diversity of land uses, and proximity to transit, have the potential to increase the use of public transportation and reduce automobile d ependency (TRB, 2009) This study shows that allocating urban growth near future transit routes using a method of forecasting future land use change based in land use suitability analysis and scenario planning techniques created a more compact urban for m and achieved greater reductions in SOV and VMT than a method that u sed a gravity model to allocate future growth based on historic development trends The following discussion highlights how this was achieved and why the results are relevant for the imp rovement of coordinated planning of land use and transportation. Compact D evelopment The study shows th at the LUCIS plus method creates a scenario with a development pattern that is more compact than a scenario created with the FLUAM method. This was in p art due to the two zon es of measurement the study uses with

PAGE 143

143 growth being measured either inside or outside of a 3 mile buffer of future transit routes. The scenarios created using the LUCIS p lus method specifically target these zone s in allocating urban growth, whereas the FLUAM scenario concentrates growth around existing and proposed activity centers regardless of their proximity to future transit While this difference create s some bias in the study it also serve s to highlight an important point about how the different land use forecasting methods have significant consequences for the coordination of land use and transportation in Lake County The FLUAM uses a gravity model that assigns attractiveness factors to TAZs based on their proximity to adjacen t activity centers and the size of those activity centers. This method emphasizes existing growth patterns by assuming that as activity centers grow larger they will continue to attract growth, which is like ly to be true to some extent. However, t he under emphasis the gravity model places on slower paced areas of growth, such as those adjacent to future transit, is problematic unless the model compensates for this when assigning attractiveness factors. The land use change forecasted us ing the FLUA M appears to show that proximity to future transit does not influence a TAZs attractiveness resulting in a land use pattern that is less supportive of the long range transportation goals the study investigated. The LUCIS plus method uses land use confli ct and suitability analysis to find suitable locations for urban growth that specifically addressed the long range planning goals of the study. In both of the LUCIS plus scenarios proximity to future transit routes determined the extent to which the large st proportion of population growth would be allocated. Land use conflict and suitability analysis used in the method, undertaken at the small scale of one quarter acre areas, is able to accurately determine how many

PAGE 144

144 acres of suitable land are available fo r growth so that different visions of varying densities and intensities of land uses could be created and tested. The varying population and employment densities of both the LUCIS plus scenarios are illustrated in maps shown in Appendices O through R. The result of using the LUCIS plus method is the creation of a land use pattern that is more supportive of the long range transportation goals under investigation than the F LUAM me thod The effectiveness of the LUCIS plus method in quantifying, and represe nting the land use planning goal of encouraging compact development improved the achievement of the long range transportation goals and ultimately c ould enhance the coordination of land use and transportation plans for the Lake County. Increased Transit R i dership The results of this study are inconclusive in determining whether the me thod of forecasting land use has an effect on increasing levels of transit ridership. The method that creates the most compact development pattern, the LUCIS plus high scenario projects only a very small increase in actual transit trips. Two different transit systems are modeled to show how transit improvements in conjunction with compacting growth might increas e transit ridership. H owever little difference results between the two transit systems that show either an increase use of transit or decrease in VMT. This was mostly due to the fact that there was not a large difference be tween the systems modeled. On e of the two systems include s t he visionary transit routes outlined in the Lake County 2020 transit development plan and should have included a commuter rail line connecting north western Orange County to central Lake County and a bus rapid transit (BRT) route cros sing through central Lake County. Due to time and technical constraints of the study however it was not possible to create a transit route that

PAGE 145

145 incorporates those modes Measures to compensate for this deficiency in one of the transit systems s uch as including bus services along the commuter rail and BRT routes with more frequent and numerous stops, however did not result in any significant differences between the two transit systems across any of the scenarios modeled. It might be possible ho wever to speculate what difference more transit trips, and therefore fewer highway trips may have had on reducing VMT. If the visionary transit routes could have been modeled and an increase in transit ridership occurred then a corresponding decrease in h ighway trips would have also occurred. In the study a 122,014 decrease in highway trips occurred between the FLUAM (less compact) and LUCIS plus high (most compact) which corresponded to a decrease of 732,000 VMT. This represents a decrease on average of 7.6 VMT with the reduction of a single highway trip. If transit ridership increased ten fold from the existing route when modeling the visionary transit route, meaning an increase from 311 to 3110 transit trips, then highway trips would decrease by the s ame amount (3110 trips) and a decrease of 23,636 VMT could be achieved. This reduction however only creates a 0.20% decrease in overall VMT for the study area indicating that for transit ridership to have any significant effect on reducing VMT a large inc rease in ridership is necessary. Such an increase in transit ridership may be possible in Lake County however w here in 2000 25% of all workers, or 20,000 people traveled to Orange County to work. Assuming that this number would increase by 2025 then oppo rtunities could exist for significantly increasing transit ridership and decreasing VM T if future land use change was planned in a way that increased accessibility to transit like that done in the LUCIS plus high scenario

PAGE 146

1 46 Evidence from the literature review undertaken by the TRB in 2009 indicates that proximity to transit has a strong influence on transit use as does the employment densities at trip ends (p. 47) The LUCIS plus scenarios modeled in this study used proximity to transit and increased e mployment densities in areas adjacent to transit as visionary transit system been mo deled in conjunction with this type of future land use configuration that larger increase s in transit ridership could have been achieved and a further reduction in VMT observed when forecasting future land use change using the LUCIS plus method Decreased D ependency on SOV This study showed that the method of forecasting land use change did ha ve an effect on decreasing SOV use. The LUCIS plus scenarios both decreased the number of SOV trips compared to the FLUAM scenario, with the LUCIS plus high scenario achieving the greatest reduction in SOV trips overall. It should be noted however that t he decrease in SOV trips was due to the overall decrease in highway trips rather than an increase in shared highway trips. The CFRPM allocates highway trip types (SOV and shared trips) according to set percentages within the model with approximately 81% o f all highway trips being assigned as SOV trips, 13% as two person shared trips and 5% as three or more person shared trips. The reduction in overall highway trips between the LUCIS plus high and FLUAM scenarios did not come as a result of an increase in transit trips, as discussed above. The reduction occurred as a result of fewer trips occurring between TAZs, as the CFRPM only measure s trips that are produced in one zone and attracted to another It is not able to measure intra zonal trips. Therefore a reduction in the number of

PAGE 147

147 highway trips between the two scenarios is a result of more trip productions and attractions bein g located within the same TAZ in the LUCIS plus high scenario than in the FLUAM scenario The CFRPM uses the ZDATA1 table, whic h includes data on residential population to calculate trip productions for each TAZ, and the ZDATA2 table, which includes employment data, to calculate trip attractions for each TAZ. The LUCIS plus high scenario resulted in more population and employmen t being located within the same TAZs because the land use conflict and suitability variables used in the LUCIS plus method allowed specific locations to be found where opportunities for compact growth such as mixed use and infill development exist where po pulation and employment could be more closely clustered together than in the FLUAM scenario The approach to allocating population and employment in FLUAM relies heavily on the assumption that the location of existing development is appropriately located for the achievement of long range planning goals, such as the ones under investigation in this study. It also assumes that the future land use element of the comprehensive plan, a document with a relatively short planning horizon of only ten years, predicts the most appropriate distribution of land uses to achieve long range planning goals. However, as not ed in the literature review land use change typically occurs at a slow pace and policies targeting changes to urban form, such as a more compact development pattern, may take a long time to become a reality Methods of forecasting future land use change t hat use land use variables based on existing development patterns and comprehensive plans may unintentionally continue short term trends at the expense of encouraging long range goals that broader strategic plans seek to achieve. The

PAGE 148

148 following discussion regarding the reduction of VMT between the FLUAM and LUCIS plus scenarios illustrates this point. Energy Conservation R eduction in VMT This study showed that the method of forecasting future land use change had an effect on conserving energy by reducing VMT. The LUCIS plus scenarios both generated fewer VMT than the FLUAM scenario, with the LUCIS plus high scenario generating the fewest VMT of all the scenarios. VMT is calculated as the product of average traffic counts by roadway lengths, so the fewer highway trips generated by the most compact scenario, LUCIS plus high, resulted in the greatest reduction in VMT. The 5.5% reduction in total VMT and 14.6% reduction in residential VMT between the FLUAM and LUCIS plus high scenarios are in line with conclu sions from the TRB report (2009) which suggest that a doubling of residential densities may be associated with a 5 to 12 percent decrease in household VMT (p. 4). Comparison of VMT across the three scenarios by area type illustrates how differences in fore casting methods resulted in different patterns of travel demand. Although Table 5 1 shows that the FLUAM and LUCIS plus low scenarios had similar overall distributions of population and employment between the two zones of measurement (within 3 miles of fu ture transit routes, and outside the 3 mile distance), the number of VMT generated in residential areas is significantly different (Table 5 6). Appendix N shows that the majority of residential roads (69%) are located within 3 miles of future transit ro utes so because the LUCIS plus low scenario generated fewer VMT in this area type it is must be due to more trip productions and attractions occurring with the same TAZs within this area for this scenario. Table 5.2 shows that the LUCIS plus

PAGE 149

149 low scenario has higher population densities within this 3 mile distance of future transit (buffers 1, 2, 3 and 4) that would account for more trip productions in residential areas. Table 5 2 also shows however that the LUCIS plus low scenario has lower employment densities than the FLUAM scenario and Table 5 1 shows it also has less employment allocated to this area The mix of employment in each scenario for this area however makes an important impact o n the number of trips attracted. Although the FLUAM scenario has more overall jobs in residential areas, the LUCIS plus low has a higher percentage of service jobs which attract more trips than commercial and industrial jobs. The ability of the LUCIS plus method to find suitable locations for more service jobs in resi dential areas therefore results in more intra zonal trips and a greater reduction in VMT in residential areas than FLUAM was able to achieve. Another important difference in VMT between the three scenarios can be seen in the levels generated in rural areas. Table 5 6 shows that both the FLUAM and LUCIS plus low scenarios produce similar VMT in rural areas, both more than the LUCIS plus high scenario which had the largest reduction in VMT occur in rural areas (Tables 5 7 and 5 8). The reason fo r this difference lies in how the population and employment are distributed between the two zones of measurement in LUCIS plus high scenario compared with the other two scenarios. Both the FLUAM and LUCIS plus low scenarios allocated population and employ ment into future developments of regional impact (DRIs) located in buffer 5, the area most distant from future transit and which contains the largest percentage of rural roads (see Appendix N ). The LUCIS plus high scenario however does not allocate into D RI s in buffer 5. After allocating population and employment into buffers 1 through 4 the remainder of unallocated population and

PAGE 150

150 employment was too low for the projected growth in DRIs in buffer 5 to be allocated in the LUCIS plus high scenario. The LUC IS plus method was able to allocate higher levels of population and employment into suitable locations in a more compact development pattern than the FLUAM method. T he FLUAM scenario assumes that DRI s in buffer 5 will be developed ture land use plan dictates this development pattern and as a result this method was unable to forecast a land use pattern that reduced VMT in the areas where the greatest proportion is created, Similarly the LUCIS plus low scen failed to significantly reduce VMT in rural areas when compared to the more compact LUCIS plus high scenario. Regional Accessibility and the C hallenges of DRIs The challenges that DRIs in distant rural locations creates for the coordination of land use and transportation are illustrated by this study. Ewing and Cervero (2001) noted in their met a analysis of literature reviews on the impacts of the built environment on VMT that overall regi onal accessibility is the main driver of vehicle travel demand. This study co ncurs with this finding, as the more compact development pattern created in the LUCIS plus high scenario produced residential areas that had greater accessibility to employment o pportunities which resulted in fewer highway trips and greater reduction in VMT by 5.5% compared with the FLUAM scenario. The FLUAM and LUCIS plus low scenarios both contained regionally isolated DRIs in rural areas that resulted in more VMT in these area s. If significant proportions of future population growth are permitted to occur in regionally isolated areas in Lake County the population

PAGE 151

151 and employment densities needed adjacent to transit routes would not be achievable for a significant increase in tr ansit ridership and reduction in SOV and VMT. Proximity to T ra nsit and Employment D ensities The literature review undertaken by the TRB (2009) reported that proximity to transit and employment densities at trip ends has a stronger influence on transit us e than urban design features such as walkability and land use factors such as mixed uses and increased residential densities. Although this study was unable to show an increase in transit ridership the LUCIS plus method was able to allocate greater propor tions of population and employment in closer proximity to future transit than did the FLUAM method and the result was a reduction in VMT due mostly to residential and employment land uses in closer proximity to each other. If a transit route that truly r eflected the planned future transit system for Lake County was modeled it is likely greater reductions in VMT would have occurred as higher employment densities proximal to transit routes was achieved using the LUCIS plus method. Criticisms of the FSM an d the CFRPM Some criticisms of the FSM cited in the literature review are evident in this study, as the CFRPM follows the same four step process of modeling transportation demand. The policy insensitivity of the FSM is apparent in this study in the way th at the CFRPM is unable to capture the potential impacts that TOD has on intra zonal trips. This is largely a result of another criticism of the FSM; that it relies on zonal level data. If TAZs are t oo large in areas where micro level change is being atte mpted, such as in areas targeted by policies encouraging transit use and TOD then important intra zonal activity may not be captured by the FSM model. For example, while the FSM shows that fewer trips may occur in TOD areas because more productions and at tractions are occurring

PAGE 152

152 with the same TAZ, it is also possible that a higher level intra zonal travel is also occurring for the same reason but the FSM is unable to account for it. Complex Land U se Models vs. Deterministic Approaches to Forecasting Land Use C hange The literature review showed that a recent trend has existed in land use modeling that favors more complex disaggregated simulations models over simpler deterministic and rule based approaches to forecasting future land use change. While the FL UAM uses a gravity model and not a disaggregated simulation model, it is based in economic theories about land use that are mathematically calculated giving the model an appearance of being more scientific than perhaps a deterministic approach such as the LUCIS plus method. A criticism of deterministic approaches is that they are too subjective and not grounded in theory, however as Handy (2008) points out complex and more scientific approaches to forecasting land use change and transportation demand also incorporate subjective measures often in ways that are less apparent to the non techni cal user of the forecasts. The scientific validity often afforded to complex forecasting methods sometimes result s in them being seen more as predictions rather than simply as possible futures which all forecasts inherently are When a forecast is treated as inevitable because it was scientifically derived the planning process suffers because equally possible forecasts may be overlooked. The criticism that determin istic approaches to forecasting land use change are too subjective to be useful belies one of the greater strengths of such approaches their ability to envision different futures based on a variety of different inputs and viewpoints. While the LUCIS pl us method used in this study focused on creating future land use patterns that supported transportation objectives, it could easily have been used to

PAGE 153

153 focus on agricultural preservation or environmental conservation, or hazard mitigation, or a combination o f all of these viewpoints. The strength of methods such as LUCIS plus lies not in that they forecast futures that will happen but instead on futures that could happen allowing a wide variety of possible alternative to be weighed and considered in the plan ning process. For the successful coordination of land use and transportation the ability of a land use forecasting method to envision a future rather than predict one is particularly useful, as the results of this study demonstrate. Study L imitations A s noted earlier this study was not able to model the visionary transit route constraints of the study. Due to this limitation the study was not able to comment about whether the compact development created by the LUCIS plus method had any impact on the transportation goal of increasing transit ridership. An opportunity for further research exists if a TROUTE.lin file for the CFRPM reflecting the visionary transit route was a what effect the transit route makes to the achievement of the transportation goals Another limitation of the study involves the suitability rasters used in both the land use c onflict and suitability analysis used in the combine raster upon which the LUCIS plus scenarios were created. As noted earlier the conflict and suitability rasters were originally created for a different project involving the central Florida region. The visionary transit route for Lake County was not used in the creation of any of these suitability rasters. GIS vector data for regional major roadways was included in the creation of the suitability rasters however, and the visionary transit routes, for t he most part, follow these same roadways. Nonetheless, had the visionary transit routes data

PAGE 154

154 been included in the suitability models that deal t with p roximity to transportation, it would have likely been weighted more highly than major roadways which woul d have resulted in areas adjacent to transit receiving a higher suitability than they did in the absence of this data. As a result it is likely that more areas that were favorable to TOD would have existed than the suitability rasters used in this study w ere able to show. An opportunity exists for further research into the effect proximity to future transit routes has on the transit ridership by testing new suitability rasters that include this data Both the FLUAM and LUCIS plus scenarios use population e stimates that were based on 2000 Census data, or other Census estimates which also used the 2000 data. Considering the age of this data, many assumptions were made about how population has actually changed since that time, which must be considered a study limitation, although a largely unavoidable one. An opportunity for future research exist once the 2010 Census data becomes available to do a study similar to this one using the latest available data and to compare the results of the two studies to see ho w they are the same or differ ent

PAGE 155

155 CHAPTER 7 CONCLUSION range transportation plan contains a public transit system that in the next 15 to 25 years is to include a commuter rail line that links eastern Lake County to western Orange County. It also includes a cross county bus rapid transit line moving through central Lake County connecting west and east. Apart from increasing transit ridership within Lake County the long range plan is to capture an increasing number of choice riders, those traveling to work, particularly those traveling to Orange County, where in 2000 one in four Lake County res idents were employed Research into the impact of urban form on transit use shows that proximity to trans it is an important factor in attracting transit users. An important question Lake County planners must therefore ask is whether the future urban for m their comprehensive plans prescribe is one that will support the transit system planned in the next two decades. Considering the slow pace at which land uses change an answer to this question is highly pertinent to long range plans presently under devel opment. Anticipating the urban form needed to support these long range transportation goals and beginning to encourage development in appropriate locati ons makes good sense I ncremental and disjointed land use change however has made the task of forecas ting future land use patterns that support public transportation difficult in Florida. H istory and research into the relationship between land use and transportation show that as urban development patterns disperse, and land uses become more separated, th e need for mobility increases. In Florida particularly and the United States in general where personal mobility is highly reliant up on the automobile, a dispersion of urban growth most often leads to increased investment in road infrastructure as a remed y for

PAGE 156

156 the increased and dispersed travel demand forecasted Over time regional trips grow in length as distances between activity centers increase and higher levels of VMT occur in areas that are more sparsely populated ; a pattern of development which is not conducive to the support of a public transit system Such a scenario is demonstrated in this study of Lake County which has shown that the largest volume of VMT occurs daily in the more distant rural areas of the county where population densities are the lowest. The aim of this study is to show how methods of forecasting future land use change can impact the achievement of long range planning goals. It put forth the hypothesis that a forecasting method based in land use suitability analysis and scenario planning techniques could improve the achievement of mutually supportive land use and transportation goals, more so than a traditional forecasting met hod based on historical development trends a nd a gravity model The contention is that land use suitability analysis used as a forecasting tool can precisely locate areas for future growth that meet specific planning guidelines designed to enhance the co ordination of long range land use and transportation goals. Various scenarios can be created that meet the planning guidelines to varying degrees and each can be tested by modeling their transportation demand to determine which scenario best reflects th e achievement of all the desired goals. The results of this study show that the LUCIS plus method was better able to reflect the long range land use and transportation goals that this study investigated, more so than the FLUAM method th at has been histori cally used in forecast ing mand. The value of this research is two fold. Firstly, it highlights the difficulty of coordinating long range land use and transportation plans when future land use

PAGE 157

157 forecasting methods are based on short term planning policies such as comprehensive plans. Because future land use elements of comprehensive plans typically forecast land use for a ten year period the risk of replicating short term development patterns is very likely when forecasting me thods use this land use data as a main indicator of future growth. The degree of land use change needed in places like Lake County to support a public transit system will take place over a much longer period of time than the ten years address ed in local c omprehensive plans. The cumulative effects of incremental land use change on transportation systems may not be adequately anticipated when future land use forecasting methods are based on short term policies, making it difficult to plan the land uses needed for longer range investments such as public transit. The LUCIS plus method offers a positive alternative to addressing this problem since it anticipates future land use change based on suitability guidelines of how future land use should be configured to meet desired planning goals, rather than following a short term future land use map. A major strength of the LUCIS plus method is its ability to both allocate land uses at a s mall scale for in depth analysis and aggre gate land uses to a macro scale so that change needed to support long range goals can be regionally envisioned. This study has also been valuable in highlighting the importance of regional accessibility in the co ordination of long range land use and transportation plans, and in demonstrating how methods of forecasting future land use change can either aid or abet this process. Large scale regionally isolated land developments have a considerable impact on a reg adjacent to future transit systems and the levels of ridership needed to reduce VMT. In

PAGE 158

158 this study two of the three land use scenarios modeled each contained two DRIs located more than 3 miles from future transit routes. Both scenarios had higher vehicle trips and VMT than the more compact scenario. Despite the fact that both of the DRIs land uses, their regional isolation negated any reductions in VMT that were gained from those design measures. As a result of the LUCIS plus method used in this study, the range goal of public tra nsit was made apparent. As an analytical tool the LUCIS plus method has much potential in estimating the regional impacts of DRIs and smaller incremental changes on future travel demand. A continuing goal of the LUCIS and LUCIS plus research currently b eing undertaken at the University of Florida is the amalgamation of all the LUCIS tools into a unified model able to be run at a variety of scales, where data inputs and model parameters are modified by user friendly GUIs These enhancements will enable n on expert GIS planners to utilize the LUCIS plus method to create scenarios for comparative analysis and further modeling within other planning frameworks such as transportation and hazard mitigation planning. The relative ease and flexibility that the LUCIS plus method brings to the assessment of regional land use plans offers many potential benefits to regional and state planning decision makers, particularly with regard to the approval of comprehensive plan amendments. If a state wide and regional LU CIS models were available then the cumulative impacts of individual plan amendments could be more easily envisioned, providing valuable input into the approval process. In light of the

PAGE 159

159 current state of affairs in Florida with regard to the over allocation of land uses in many local comprehensive plans this kind of analytical input would be very helpful in guiding decision makers According to Interim Report 2010 107 of the Florida Senate Committee on Community Affairs (2009) there is controversy surround The required to meet projected future population growth. Due to an uneven appli cation of rule by the Department of Community Affairs large amounts of land have been assigned to future land uses far in excessive of what projected population growth require s (p. 8) and have resulted in urban sprawl in many countie s. Changing the land use designation of many of these inappropriately located developments would be difficult due to an interference of private property rights and their existence increases the difficulty of approving more appropriately located developme nts. Report recommendations for resolving these issues include a clearer articulation of the as it relates to the goals for planning growth. S pecifically the se goals are the discouragement of urban sprawl and the efficient use of infrastructure spending, the prevention of the fragmentation of the environment; and the promotion of coordinated plans among adjacent local governments (pp. 8 9) The use of the LUCIS plus method to translate interrelated planning goals into quantifiabl e a nd comparable regional scenarios was demonstrated in this study and could well be their approval process for comprehensive plan amendments.

PAGE 160

160 Prior to the commencement o f this study in January 2009, the LSMPO began work on updating their LRTP for the period 2015 to 2035 As part of the update process consideration was given to the methods used to forecast future land use change. After a comparison of the FLUAM and the LU CIS plus methods it was decided in May 2009 that a future land use scenario derived using the LUCIS plus method would provide the socio economic data for modeling the future transportation demand needed for the ir 2035 LRTP update. This recognition by the LSMPO of the merits of the LUCIS plus method helps raise awareness among other regional planning organizations and state planning agencies that analytical tools now exist that can improve the coordination of long range planning initiatives at loc al, regional and state levels.

PAGE 161

161 APPENDIX A LAND USE MODELS REVI EWED BY AUTHOR AND MODEL TYPE Model Name Model Type Miller et al. (1998) EPA (2000) Chang (2006) Zhao & Chung (2006) DRAM/EMPAL/ITLUP METROPILUS Spatial Interaction Y Y Y Y HLFM 11+ Spatial Interaction N N N Y LILT Spatial Interaction N N Y Y LUTRIM Spatial Interaction N N N Y DELTA Spatial Input/Output N Y N Y MEPLAN Spatial Input/Output Y Y N Y TRANUS Spatial Input/Output Y N N Y CUFM: CUF 1/CUF 2 Rule Based N Y N Y SAM/LAM/SAM IM Rule Based N Y N Y SLAM Rule Based N N N Y ULAM Rule Based N N N Y UPLAN Rule Based N Y N Y WHAT IF? Rule Based N Y N Y METROSIM Random Utility / Discrete Choice Y Y Y Y INDEX Other N Y N Y LUCAS Other N Y N Y Markov Model of Residential Vacancy Other N Y N Y SMART PLACES Other N Y N Y IRPUD Micro Simulation N Y N Y MASTER Micro Simulation N N N Y NBER/HUDS Micro Simulation N N N Y UrbanSim Micro Simulation Y Y N Y Herbert Stevens Mathematical (Linear) Programming N N Y Y POLIS Mathematical (Linear) Programming N N Y Y TOPAZ/TOPMET Mathematical (Linear) Programming N N Y Y SLEUTH Cellular Automation N Y N Y MUSSA Bid Rent Y N Y N

PAGE 162

162 APPENDIX B STRENGTHS AND LIMITA TIONS OF LAND USE MODEL TYPES BY AUTHOR Author Miller et al. (1998) Model Type Strength Limitation Spatial Interaction history of validation Spatial Input/Output o include chain of demand (land buildings activity floor space, reliance use of aggregate input/outputs restricts zone size Rule Based N/A N/A Random Utility / Discrete Choice economic reliance on equilibrium Micro Simulation N/A N/A Linear Programming N/A N/A Cellular Automation N/A N/A Bid Rent uses traffic zone level and finer resulting in forecasts with good face validity that must be run by time step Author EPA (2000) Model Type Strength Limitation Spatial Interaction constraints or other influences (to account for local knowledge) no mechanism for simulating land market variables may lead to underestimation of infrastructure investments Spatial Input/Output analysis of different kinds of policies problematic depending on level of observed data used in base year Rule Based alternative future development scenarios basis to examine interrelated factors such as transportation, fiscal and planning policies that affect land use change Random Utility / Discrete Choice recognition of how market forces shape and change land use None stated Micro Simulation world processes dynamically in time and space None stated Mathematical (Linear) Programming N/A N/A Cellular Automation temporal booms and busts policies and economic impacts on land use change Bid Rent N/A N/A

PAGE 163

163 Author Chang (2006) Model Type Strength Limitation Spatial Interaction diverse transport components aggregates trends w/out social or income disaggregation Spatial Input/Output N/A N/A Rule Based N/A N/A Random Utility / Discrete Choice high degree of behavioral validity of individual's decision making process responses between transport and land use (endogeno us locational accessibility subject to transport system one directional instead of mutually determined). Micro Simulation N/A N/A Linear Programming models able to produce endogenously determined transport costs; impedance generated by mutual adjustment between land use and transport s little consideration to unique decision making process of an individual Cellular Automation N/A N/A Bid Rent represents unique characteristics of locations by using describes the behavior of decision making process of consumers between land use and transport Author Zhao & Chung (2006) Model Type Strength Limitation Spatial Interaction employment distributed as a function of location attractiveness and travel cost land market and prices not considered detail limited Spatial Input/Output estate and labor markets are modeling process changes in one or more inputs Rule Based range scenario testing economic theories model complex economic and market processes in detail Random Utility / Discrete Choice explicitly with land use policy and land use change None stated Micro Simulation which are aggregated to form the to behavioral responses to changes in transport system or land use policies because it models the decision making process of individuals None stated Linear Programming standable & computationally easier than other optimization techniques responses to changes in transport system or land model Cellular Automation representing interactions between a location and its immediate surrounds behavior Bid Rent N/A N/A

PAGE 164

164 APPENDIX C DATA DICTIONARY FOR LUCIS PLUS SCENARIOS Data Description Type Source Population & household estimates Lake County 2007 Tabular Census Bureau Traffic analysis zone data Lake County 2000 and 2012 Tabular and GIS vector FDOT Traffic analysis zone data Lake County 2007 Tabular and GIS vector Thompson projected using FDOT 2000 & 2012 TAZ data Average of medium and high population projections Lake County 2025 Tabular BEBR Employment projections Lake County 2005, 2010 and 2025 Tabular ECFRPC created using REMI Policy Insight Model Parcel data Lake County 2007 GIS vector Florida Geographic Data Library (FGDL) Hydrography Lake County lakes and ponds GIS vector FGDL Florida Managed Lands Lake County conservation areas 2009 GIS vector FGDL Developments of regional impact Tabular and GIS v ector ECFRPC LUCIS agriculture, conservation and urban suitabilities Lake County created in 2008 2009 GIS raster Dr. P. Zwick UF LUCIS land use conflict Lake County created in 2008 2009 GIS raster Dr. P. Zwick UF Visionary transit lines and stations for 2035 Lake County GIS vector ECFRPC School attendance zones Lake County 2008 2009 GIS vector FGDL

PAGE 165

165 APPENDIX D ZDATA1 AND ZDATA2 DE SCRIPTIONS ZDATA1 Attributes Description TAZ Zone number SFDU Single family dwelling units SF_PCTVNP % of single family dwelling units that are vacant or occupied by non permanent residents SF_PCTVAC % of single family dwelling units that are vacant SFPOP Single family population SF_0AUTO Single family dwelling units with 0 automobiles SF_1AUTO Single family dwelling units with 1 automobiles SF_2 AUTO Single family dwelling units with 2 automobiles MFDU Multi family dwelling units M F_PCTVNP % of multi family dwelling units that are vacant or occupied by non permanent residents M F_PCTVAC % of multi family dwelling units that are vacant MFPOP Multi family population M F_0AUTO Multi family dwelling units with 0 automobiles M F_1AUTO Multi family dwelling units with 1 automobiles MF_2 AUTO Multi family dwelling units with 2 automobiles HMDU Hotel/motel dwelling units HMOCC % of Hotel/motel dwelling units occupied HMPOP Hotel/motel population ZDATA 2 Attributes Description TAZ Zone number INDEMP Industrial employment COMEMP Commercial (retail) employment SEREMP Service employment TOTEMP Total employment SCHENRL School enrolment Note. Adapted from p. 8.

PAGE 166

166 APPENDIX E ALLOCATION BUFFERS 1 AND 2 Note: Map courtesy of author.

PAGE 167

167 APPENDIX F ALLOCATION BUFFER 3 Note. Map courtesy of author.

PAGE 168

168 APPENDIX G ALLOCATION BUFFER 4 Note. Map courtesy of author.

PAGE 169

169 APPENDIX H ALLOCATION BUFFER 5 Note. Map courtesy of author.

PAGE 170

170 APPENDIX I AGGREGATION OF F DOR LAND USE CODES IN THE PAR CEL DATA TO TAZ LAND USE CODES Land Uses Land Uses F DOR TAZ F DOR TAZ Vacant residential VACRES Grazing land IND Single family SF Poultry, bees, fish, rabbits IND Mobile homes MF Dairies, feed lots IND Multi family MF Orchards, groves, citrus IND Condominia MF Ornamentals, misc. agriculture IND Cooperatives MF Mining, petroleum, gas lands IND Multi family < 10 units MF Subsurface rights IND Vacant commercial VACCOM Sewage disposal, borrow pits IND Stores one story COM Retirement homes SER Mixed use (store/office) COM Boarding homes SER Department stores COM One story non prfssnl offices SER Supermarkets COM Multi story non prfssnl offices SER Regional shopping malls COM Professional service buildings SER Community shopping centers COM Airports, marinas, terminals SER Restaurants, cafeterias COM Financial institutions SER Drive in restaurants COM Insurance company offices SER Repair service shops COM Camps SER Service stations COM Golf courses SER Auto repair, service, sales COM Hotels, motels SER Parking lots, mobile home sales COM Churches SER Florist, greenhouses COM Private/public schools, colleges SER Drive in theaters, open stadiums COM Private/public hospitals SER Encl. theaters, auditoriums COM Homes for aged SER Night clubs, bars, lounges COM Mortuaries, cemeteries SER Tourist attractions COM Clubs, lodges, union halls SER Race horse, auto, dog tracks COM Sanitariums, convalescent SER Vacant Industrial VACIND Cultural organizations SER W/sale manufacture/processing IND Military SER Light and Heavy Manufacturing IND Forest, park, recreation areas SER Lumber yards, sawmills IND Other county/state/federal SER Fruit, veg and meat packing IND Gov. owned and leased SER Canneries, distilleries, wineries IND Utilities SER Other food processing IND Outdoor recreational SER Mineral processing IND Acreage not zoned for agric. VAC W/houses, distribution centres IND Vacant Institutional VACCOM Industrial storage (fuel, equip) IND Undefined No use Improved agriculture IND Rights of way, streets, roads No use Cropland IND Rivers, lakes, submerged land No use Timberland IND Centrally Assessed No use

PAGE 171

171 APPENDIX J AGGREGATION OF ECFRP C EMPLOYMENT CATEGOR IES TO TAZ LAND USE CODES ECFRPC Employment Category TAZ Land Use Code Wholesale trade COM Retail trade COM Forestry, fishing, other IND Mining IND Construction IND Manufacturing IND Farm IND Utilities SER Transportation, warehousing SER Information SER Finance, insurance SER Real estate, rental, leasing SER Professional, technical services SER Management of companies SER Administration, waste services SER Educational services SER Health care, social assistance SER Arts, recreation, entertainment SER Accommodation, food services SER Other services (excluding government) SER State and local governments SER Federal civilian SER Federal military SER

PAGE 172

172 APPENDIX K LAKE COUNTY 2009 2020 TRANSIT DEVELOP MENT PLAN MAP AND DESCRIPTION OF MODES AND ROUTES Associates, 2008, p. 9 26 and p. 9 29.

PAGE 173

173 APPENDIX L TOD DESIGN STANDARDS USED AS STUDY GUIDEL INES Source. Florida Department of Transportation, 2009b, p. 9.

PAGE 174

174 APPENDIX M GENERAL SELECTION CRITERIA F OR ALLOCATION INTO B UFFERS 1 TO 5 FOR LOW AND HIGH LUC IS PLUS SCENARIOS Buffer Selection 1 Selection 2 Selection 3 Selection 4 Selection 5 1 Redev areas with med to high MU opptnty in areas >= 1acre @ MU densities VAC, VACRES, VACCOM or VACINST with med to high MU opptnty in areas >= 1 acre @ MU densities 2 Redev areas with med to high MU opptnty in areas >= 1acre @ MU densities Infill opptnty & incr empl density VACCOM and VACRES to SER @ MU density; exist SER @ MU density; MOB to SER @ MU density 3 Redev areas with med to high MU opptnty in areas >= 1acre @ MU densities Opptnty to incr res density in redev areas remaining redev (areas < 1 ac) select SF,MF & VACRES allocate @ trend MF density Opptnty to incr empl density in redev areas remaining redev areas (< 1 ac) select SER, IND, VACCOM incr SER to MU density; change VACCOM & IND to SER @ MU density; SF opptnty select VACRES and using conflict & sliced SFSUIT allocate to most suitable (conf = 113, 112, 123, 213 or 223 and highest SFSUIT ) Emp opptnty outside redev select high urban pref (conflict = 113, 112, 123, 213, 223) allocate VAC, VACCOM, VACINST to 75% SER and 25% COM trend densities 4 DRIs select DRI>0; allocate based on conflict (med high urb pref) and suitability @ DRI densities Outside of DRIs MF & SF to VACRES @ trend density; SER to VACRES @ trend density (excessive amounts of VACRES in buff 4); COM to VACCOM and IND to VACIND @ trend density 5 DRIs select DRI>0; allocate based on conflict (avoid CON conf) and suitability @ DRI densities Outside of DRIs MF & SF to VACRES @ trend density; SER & COM to VACCOM and IND to VACIND @ trend density any shortfalls allocate to VAC

PAGE 175

175 APPENDIX N STUDY AREA ROADS USE D IN CFRPM BY AREA T YPE Note. Map courtesy of author

PAGE 176

176 APPENDIX O LUCIS PLUS LOW SCENARIO PO PULATION DENSITY Note. Map courtesy of author.

PAGE 177

177 APPENDIX P LUCIS PLUS HIGH SCENARIO P OPULATION DENSITY Note. Map courtesy of author.

PAGE 178

178 APPENDIX Q LUCIS PLUS LOW SCENARIO EM PLOYMENT DENSITY Note. Map courtesy of author.

PAGE 179

179 APPENDIX R LUCIS PLUS HIGH SCENARIO E MPLOYMENT DENSITY

PAGE 180

180 LIST OF REFERENCES Bartholomew, K. (2005). Integrating land use issues into transportation planning: Scenario planning. Retrieved March 19, 2010, from http://faculty.arch.utah.edu/bartholomew/SP_SummaryRpt_Web.pdf Bartholomew, K., & Ewing, R. (2009). Land use transportation scenarios and future vehicle travel and land consumption: A meta analysis. Journal of the American Planning Association, 75 (1), 13 2 7. Bento, A. M., Cropper, M. L., Mobarak, A. M., & Vinha, K. (2005). The effects of urban spatial structure on travel demand in the United S tates. Review of Economics & Statistics, 87 (3), 466 478. Borning, A., Waddell, P. & Forster, R. (2006). UrbanSim: Us ing simulation to inform public deliberation and decision making. Retrieved November 25, 2008, from http://www.cs.washington.edu/homes/borning/papers/borning urbansim case study 2006.pdf Carbon Dioxide Information Analysis Center. (unknown). Top 20 emitting countries by total fossil fuel CO2 emissions for 2006. Retrieved February 10, 2010, from http://cdiac.esd.ornl.gov/trends/emis/tre_tp20.html Chang, J. S. (2006). Models of the rela tionship between transport and l and use: A review. Transport Reviews, 26 (3), 325 350. Collins, M.G., Steiner F.R., & Rushman, M.R. (2001). Land u se suitability analysis in the United S tates: Historical development and promising technological achievements. Environmental Management, 28 (5), 611 621. Dalton, L. C. (2001). Thinking about tomorrow: Bringing the futur e to the forefront of planning. Journal of the American Planning Association, 67 (4), 397. Deka, D. (2004). Social and environmental justice issues in urban transportation. In S. Hanson, & G. Guiliano (Eds.), The geography of urban transportation (3rd ed., pp. 332 355) New York: The Guildford Press. Environmental Systems Research Institute. (2010). ArcGIS desktop 9.3 help. Retrieved March 27, 2010, from http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?TopicName=welcome Florida Department of Transportation. (unknown). Florida highway mileage reports glossary. Retrieved April 1, 2010, from http://www.dot.state.fl.us/planning/statistics/mileage rpts/glossary.shtm Florida Department of Transportation. (2005). FSUTMS powered by CUBE/VOYAGER data dictionary. Retrieved March 25, 2010, from http://www.fsutmsonline.net/images/uploads/mtf files/datadictionary.pdf

PAGE 181

181 Florida Department of Transportation. (2009 a ). MPO program management handbook Chapter 4: Long range transportatio n plan. Retrieved February 8, 2010, from http://www.dot.state.fl.us/planning/policy/metrosupport/mpohandbook/ch4.pdf Florida Department of Transportation. (2009b). Tr ansit oriented development design guidelines. Retrieved March 27, 2010, from http://www.floridatod.com/docs/Products/TODGuide041409.pdf Gannett Fleming Inc. (unknown). CFRPMv45 Cube Voyager conversion technical memorandum 1 model conversion procedural guide Orland o FL: Florida Department of Transportation District 5. Handy, S. (2005). Smart growth and the transportation land use connection: What does the research tell us? International Regional Science Review, 28 (2), 146 167. Handy, S. (2008). Regional transportation planning in the US: An examination of changes in technical aspects of the planning process in response to changing goals. Transport Policy, 15 (2), 113 126 Hanson, S. (2004). The Context of Urban Travel: Concepts and recent trends. In S. Hanson, & G. Guiliano (Eds.), The geography of urban transportation (3rd ed., pp. 3 29). New York: The Guildford Press. Iacono, M., Levinson, D., & El Geneidy, A. (2008). Models of transportation and land use change: A guide to the territory. Journal of Planning Literature, 22 (4), 323 340. Isserman, A. (1985). Dare to plan: An essay on the role of the future in planning practice and education. Town Planning Review, 56 483 491. Johnston, R. A. (2004). The urban transportation planning process. In S. Hanson, & G. Guiliano (Eds.), The geography of urban transportation ( 3rd ed., pp. 115 140). New York: The Guildford Press. Gertler, A. W. (2009). Application of a scenario based modeling system to evaluate the air quality impacts of future growth. Atmospheric Environment, 43 (5), 1021 1028. Kelly, E. D. (1994). The transportation land use link. Journal of Planning Literature, 9 (2), 128 145. Kenworthy, J. R. (2003). Transport energy use and greenhouse gases in urban passenger transport systems: A study of 84 global cities. Paper presented at the International third c onference of the regional government network for sustainable d ev elopment, Notre Dame University, Fremantle, Western Australia. Retrieved February 10, 2010, from http://cst.uwinnipeg.ca/documents/Transport_Greenhouse.pdf

PAGE 182

182 Knaap, G. (unknown). A requiem for smart growth. Retrieved March 17, 2010, from http://www.smartgrowth.umd.edu/research/pdf/Knaap_Requiem_022305.pdf Koomen, E., Stillwell, J., Bakema, Aldrik, Scholten, & Henk J. (Eds.). (2007). Modelling land use change: Progress and applications The Netherlands: Springer. Lake County Department of Growth Management. (2009). Lake County, Florida C omprehensive p lan as amended th rough ordinance #2009 32. Retrieved January 26, 2010, from http://www.lakecountyfl.gov/pdfs/2025/Comp_Plan_as%20of_Ordinance_2009 32.pdf Lake Sumter Metropolitan Pl anning Organization. (2006). Lake Sumter 2025 long range transportation plan prepared for: Lake Sumter metropolitan planning organization. Retrieved January 26, 2010, from http://www.lakesumtermpo.com/pdfs/lrtp/lake_sumter_lrtp_complete_052307.pdf McHarg, I. L. (1992). Design with nature (25 anniversary ed.). New York: J. Wiley. McNally, M. G. (2007a). The activity based approach. In K.J. Button & D. A. Hensher (Ed s .), Handbook of transport modelling (2nd ed., pp. 53 69). Amsterdam: Elsevier Science. McNally, M. G. (2007 b ). The four step model. In K.J. Button & D.A. Hensher, (Ed s .), Handbook of transport modelling (2nd ed., pp. 35 52). Amsterdam: E lsevier Science. Meyer, M. D., & Miller, E. J. (2001). Urban transportation planning : A decision oriented approach New York: McGraw Hill. Myers, D., & Kitsuse, A. (2000). Constructing the future in planning: A survey of theories and tools. Journal of Pla nning Education and Research, 19 (3), 221 231. Miller, E. J., Kriger, D. S. & Hunt, J. D. (1998). Integrated urban models for simulation of transit and land use policies. Retrieved October 10, 2008, from http://www.nap.edu/openbook.php?record_id=9435&page=R1 Muller, P. O. (2004). Transportation and urban form: Stages i n the spatial evolution of the A merican metropolis. In S. Hanson, & G. Giuliano (Eds.), The geography of urban t ransportation (3rd ed., pp. 59 85). New York: The Guildford Press. Smith, S. K. (2005). Florida population growth: Past, present and future. Retrieved January 30, 2010, from http://www.bebr.ufl.edu/system/files/FloridaPop2005_0.pdf

PAGE 183

183 State of Florida Senate Committee on Community Affairs. (2009). Interim report 2010 107 Population need as a criteria for changes to a local government's future land use map. Retrieved March 9, 2 010, from http://www.fl counties.com/Docs/Legislative%20Division/ Growth%20Management/Population_N eed_as_a_Criteria_for_Changes_to_a_Local_Government_s_Future_Land_Use_ Map.pdf Transportation Research Board. (2009). Special Report 298: Driving and the built environment the effects of compact development on motorized tra vel, energy use, and CO 2 emissions Washington DC: The National Academies. Transportation Research Board. (2010). About TRB. Retrieved March 9, 2010, from http://www.trb.org/AboutTR B/Public/AboutTRB.aspx United Nations Economic Commission for Europe. (2008). Pas senger vehicles (number) at 31 D ecember by vehicle category, country and year. Retrieved March 9, 2010, from http://w3.unece.org/pxweb/Dialog/varval.asp?ma=ZZZ_TRvehiclePsgFlit_r&ti=Pas seng er+Vehicles+%28number%29+at+31+December+by+Vehicle+Category%2C +Country+and+Year&path=../DATABASE/Stat/40 TRTRANS/02 TRRoadFleet/&lang=1 United States Census Bureau. (2000). Profile of selected housing characteristics: 2000 Census 2000 summary file 3 L ake County, Florida Retrieved February 2, 2010, from http://factfinder.census.gov/servlet/QTTable?_bm=y& geo_id=05000US12069& qr_name=DEC_2000_SF3_U_DP4& ds_name=DEC_2000_SF3_U& _lang=en& _sse=on United States Census Bureau. (2000). Profile of selected housing characteristics: 2000 Census 2000 summary fil e 3 Orange County, Florida Retrieved February 2, 2010, from http://factfinder.ce nsus.gov/servlet/QTTable?_bm=y& geo_id=05000US12095& qr_name=DEC_2000_SF3_U_DP4& ds_name=DEC_2000_SF3_U& _lang=en& redoLog=false& _sse=on United States Census Bureau. (2000). Profile of selected housing characteristics: 2000 Census 2000 summary file 3 Osceola County, Florida Retrieved February 2, 2010, from http://factfinder.census .gov/servlet/QTTable?_bm=y& geo_id=05000US12097& qr_name=DEC_2000_SF3_U_DP4& ds_name=DEC_2000_SF3_U& _lang=en& redoLog=false& _sse=on

PAGE 184

184 Un ited States Census Bureau. (2007 ). Selecte d housing characteristics: 2005 200 7 American communi ty survey 3 year estim ates Lake County, Florida. Retrieved October 2 1, 2009 from http://factfinder.census.gov/servlet/ACSSAFFFacts?_event=&geo_id=05000US12 069&_geoContext=01000US|04000US12|05000US12069&_street=&_county=Lak e+County&_cityTown=&_state=&_zip=&_lang=en&_sse=on&A ctiveGeoDiv=&_use EV=&pctxt=fph&pgsl=050&_submenuId=factsheet_1&ds_name=DEC_2000_SAF F&_ci_nbr=null&qr_name=null®=null%3Anull&_keyword=&_industry= United States Census Bureau. (2008). Journey to work and place of work Census 2000 data 2000 County to c ounty worker flow files. Retrieved January 30, 2010, from http://www.census.gov/population/www/cen2000/commuting/index.html#FL United States Census Bureau. (2008). Sele cted housing characteristics: 2008 2008 American community survey 1 year estimates Lake County, Florida. Retrieved February 2, 1010, from http://factfinder.census.gov/servlet/ADPTable?_bm=y& geo_id=05000US12069& qr_name=ACS_2008_1YR_G00_DP4& context=adp& ds_name=& tree_id=308& _lang=en& redoLog=false& format= United S tates Census Bureau. (2008). Selected housing characteristics: 2008 2008 American community survey 1 year estimates Orange County, Florida. Retrieved February 2, 1010, from http://factfinder.census.gov/servlet /ADPTable?_bm=y& context=adp& qr_name=ACS_2008_1YR_G00_DP4& ds_name=ACS_2008_1YR_G00_& tree_id=308& redoLog=true& _caller=geoselect& geo_id=05000US12095& format=& _lang=en United States Census Bureau. (2008). Selected housing characteristics: 2008 2008 American community survey 1 year estimates Osceola County, Florida. Retrieved February 2, 1010, from http://factfinder.census.gov/servlet/ADPTable?_bm=y& context=adp& qr_name=ACS_2008_1YR_G00_DP4& ds_name=ACS_2008_1YR_G00_& tree_id=308& redoLog=true& _caller=geoselect& geo_id=05000US120 97& format=& _lang=en United States Environmental Protection Agency, Office of Research and Development. (2000). Projecting land use change: A summary of models for assessing the effects of community growth and change on land use patterns (Report No. EPA/6 00/R 00/098). Cincinnati, OH: U.S. EPA. United States Environmental Protection Agency. (2009). Human related sources and sinks of carbon dioxide. Retrieved February 10, 2010, from http://www.epa.gov/climatechange/emissions/co2_human.html

PAGE 185

185 United States Government Accountability Office. (2007). Crude oil: Uncertainty about future oil supply makes it important to develop a strategy for addressing a peak and decline in oil production (Report No. GA 07 283). Washington DC: United States Government Accountability Office. University of Florida Bureau of Economic and Business Research, & University of Florida Population Program. (2009). Florida population studies bulletin 153 Florida county population projections, 2008 2035 Gainesville, FL: Bureau of Economic and Business Research, College of Business Administration, University of Florida. Wachs, M. (2001). Forecasting versus envisioning: A new window on the future. Journal of the American Planning Association, 67 (4), 367 72. Wachs, M. (2004). Reflections on the planning process. In S. Hanson, & G. Guiliano (Eds.), The geography of urban transportation (3rd ed., pp. 141 162). New York: The Guildford Press. Waddell, P., & Ulfa rsson, G. F. (2004). Introduction to urban simulation: Design and development of operational models. In P. Stopher, K. Button, K. Haynes & D. Hensher (Eds.), Handbook of transport volume 5: Transport geography and spatial systems (pp. 203 236) Pergamon Pre ss. Weitz, J. (1999). From quiet revolution to smart growth: State growth management programs, 1960 to 1999. Journal of Planning Literature, 14 (2), 266 337. Wilbur Smith Associates. (2008). Lake County transit development plan 2008 major update. Retrieved February 4, 2010, from http://www.lakesumtermpo.com/pdfs/tdp/Lake_County_2020_TDP_adopted0808.p df Zegras, C., Sussman, J., & Conklin, C. (2004). Scenario pl anning for strategic regional transportation planning. Journal of Urban Planning and Development, 130 (1), 2 13. Zhao, F., & Chung, S. (2006). A study of alternative land use forecasting models (Final Report No. BD015 10). Tallahassee, FL: Florida Departmen t of Transportation.

PAGE 186

186 BIOGRAPHICAL SKETCH Liz Thompson was born in 1963 in Camp Hill, Pennsylvania. Her family emigrated to Perth, Wester n Australia in 1971 where she lived until returning to live in the United States in 2003 Liz is currently p ursuing a Master of Arts degree in Urban and Regional Planning at the University of Florida, in Gainesville. She ha s a Bachelor of Arts degree in geography and h istory from the University of Western Australia. As a graduate student at the University of F lorida, Liz worked as a research assistant for Dr. Paul Zwick assisting in the continuing development of the Land Use Conflict Identification Strategy (LUCIS), work that she will continue after her graduation in summer 2010. Liz began working for the Shim berg Centre for Housing Studies in April 2010 where her proficiency in geographic information systems is assisting in the development of the Florida Neighborhood Data Initiative (FNDI) which aims to support affordable housing and community revitalization. Liz met Kevin Thompson in 1994 during a five year working holiday that began in 1992 and traversed Japan, the U nited States Taiwan, and New Zealand. In 1996 they they married and later that year their beautiful daughter Amelie Helen was born. They retu rned to live in the USA in 2003, moving to Gainesville Florida in 2007 where they continue to reside.