<%BANNER%>

Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.

Permanent Link: http://ufdc.ufl.edu/UFE0042633/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.
Physical Description: Book
Language: english
Creator: Wang, Ruoniu
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

Statement of Responsibility: by Ruoniu Wang.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2010.
Local: Adviser: Steiner, Ruth L.
Local: Co-adviser: Blanco, Andre.
Electronic Access: INACCESSIBLE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042633:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042633/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.
Physical Description: Book
Language: english
Creator: Wang, Ruoniu
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

Statement of Responsibility: by Ruoniu Wang.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2010.
Local: Adviser: Steiner, Ruth L.
Local: Co-adviser: Blanco, Andre.
Electronic Access: INACCESSIBLE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042633:00001


This item has the following downloads:


Full Text

PAGE 1

1 MEASURING URBAN FORM AND EXAMINING ITS IMPACT ON TRAFFIC CONGESTION IN FLORIDA By RUONIU WANG 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 Ruoniu Wang

PAGE 3

3 To my family and f riends

PAGE 4

4 ACKNOWLEDGMENTS First and foremost, I would like to thank my family and friend s Most importantly, I owe my deepest gratitude to my parents and grandparents for all of their love, support, and encouragement. Thank you for always being there. I would also like to thank my beautiful girlfriend Zhou Dan for all of her love, support, and incredible patience and understanding. I am looking forward to exploring the future together with her no matter where it will be. This thesis would not have been possible without my Chair Dr. Ruth Steiner. Having been in a peculiar pre dicament at the beginning of th is thesis endeavor, Ruth took me under her wing and guided me in the right direction. It is her who helped me gain a better understanding of the bigger picture. I would also like to present my profound gratitud e to my Co Chair Dr. Andres Blanco, who provided me infinite instruction, wisdom, and encouragement throughout this process. I t is an honor for me to be their student. I am very grateful to Abdulnaser Arafat and Russell Provost. I would never have been abl e to complete this without their assistance on the key issues of my thesis. Additionally, I am indebted to my thesis advisory board members Nathaniel Wingfield, Jeffrey Schmucker, Laura Erdely, Maksim Shamaltsuyev, Robe rt Narvaez, and Gareth Hanley, wh o have supported me in a number of ways. Finally, I would like to thank all the people from Department of Urban and Regional Planning who have supported me along the way. You have truly made my experience here enjoyable and unforgettable.

PAGE 5

5 TABLE OF CONTEN TS page LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ..................... 9 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 11 Overview ................................ ................................ ................................ ................. 11 Traffic Congestion and the Built Environment ................................ ......................... 11 Study Area ................................ ................................ ................................ .............. 16 Conceptual Model ................................ ................................ ................................ ... 17 Objectives ................................ ................................ ................................ ............... 18 Organization ................................ ................................ ................................ ........... 18 2 LITERATURE REV IEW ................................ ................................ .......................... 20 Urban Form ................................ ................................ ................................ ............. 20 Urban Form at Different Scales ................................ ................................ ........ 21 Dimensions of Sub metropolitan Urban Form ................................ .................. 23 Multi dimensional Approach of Measuring Urban Form ................................ ... 27 Traffic Congestion ................................ ................................ ................................ ... 29 Measures of Traffic Congestion ................................ ................................ ........ 30 Understanding Traffic Congestion ................................ ................................ .... 32 Research on Urban Form and Traffic Congestion ................................ .................. 34 Overview ................................ ................................ ................................ .......... 34 Models Measuring Urban Form and Congestion ................................ .............. 35 Summary ................................ ................................ ................................ ................ 39 3 METHODOLOGY ................................ ................................ ................................ ... 41 Overview of Study Design ................................ ................................ ....................... 41 Study Area Selection ................................ ................................ .............................. 41 Classification of Land Type ................................ ................................ ..................... 42 Establishing the Analysis Areal Units ................................ ................................ ...... 43 Size of Areal Unit ................................ ................................ .............................. 44 Location of Areal Unit ................................ ................................ ....................... 44 Operationalizing Variables ................................ ................................ ...................... 45 De pendant Variables ................................ ................................ ........................ 45 Independent Variables ................................ ................................ ..................... 47 Quantifying Independent Variables ................................ ................................ ......... 49 Statistical Method ................................ ................................ ................................ ... 50

PAGE 6

6 Summary ................................ ................................ ................................ ................ 51 4 FINDINGS AND RESULTS ................................ ................................ ..................... 58 Areal Unit Effect on Urban Form Outcomes ................................ ............................ 58 Description of Data Characteristics ................................ ................................ ......... 60 Urban Form Outcomes ................................ ................................ ..................... 60 Congestion Outcomes ................................ ................................ ...................... 64 Correlations among Indices ................................ ................................ .................... 66 Correlations among Urban Form Indices ................................ .......................... 66 Correlations among Congestion Indices ................................ ........................... 68 Factor Analysis of Urban Form Indices ................................ ................................ ... 69 Ranking Areas by Measures ................................ ................................ ................... 71 Ranking on Urban Form Factors ................................ ................................ ...... 71 Ranking on Congestion Indices ................................ ................................ ........ 73 Correlation Analysis between Congestion and Urban Form ................................ ... 74 Summary ................................ ................................ ................................ ................ 78 5 DISCUSSIONS ................................ ................................ ................................ ....... 88 Discussion of Findings and Results ................................ ................................ ........ 88 Limitations of this Study ................................ ................................ .......................... 90 Opportunities of Future Research ................................ ................................ ........... 94 6 CONCLUSIONS ................................ ................................ ................................ ..... 97 APPENDIX A CLASSIFICATION OF LAND TYPES ................................ ................................ ... 100 B P ROCESS OF QUANTIFYING URBAN FORM ................................ .................... 104 C URBAN FORM OUTCOMES BY SCENARIO AND MEAN VALUES .................... 112 D TRAFFIC CONGESTION OUTCOMES ................................ ................................ 127 LIST OF REFERENCES ................................ ................................ ............................. 129 BIOGRAPH ICAL SKETCH ................................ ................................ .......................... 133

PAGE 7

7 LIST OF TABLES Table page 4 1 Ratios of the maximum to the minimum values calculated in four scenarios ...... 80 4 2 Descriptive statistics of urban form indices ................................ ......................... 81 4 3 Descriptive statistics of traffic congestion indices ................................ ............... 81 4 4 Bivariate correlation between urban form indices ................................ ............... 82 4 5 Bivariate correlation between congesti on indices ................................ ............... 83 4 6 Rotated component matrix describing five urban form factors ............................ 83 4 7 UA rankings on urban form factors ................................ ................................ ..... 83 4 8 UA rankings on traffic congestion ................................ ................................ ....... 84 4 9 Pearson correlation coefficients: urban form factors and congestion indices ..... 84 A 1 Land use codes to category of land types lookup table ................................ .... 100 C 1 Urban form outcomes Base Scenario ................................ ............................ 112 C 2 Urban form outcomes West Scenario ................................ ............................ 114 C 3 Urban form outcomes South Scenario ................................ ........................... 117 C 4 Urban form outcomes Southwest Scenario ................................ ................... 119 C 5 Mean values for urban form out comes ................................ ............................. 122 C 6 Composite z scores of urban form factors ................................ ........................ 125 D 1 Traffic congestion outcomes ................................ ................................ ............. 127

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Conceptual model ................................ ................................ ............................... 19 3 1 Map of study areas ................................ ................................ ............................. 52 3 2 Ma p of county population ................................ ................................ ................... 53 3 3 Change of a real u nit s ystem for f our s cenarios in Miami Dade County. ............. 54 3 4 Dimension s of u rban form. ................................ ................................ ................. 56 4 1 Maps of urban form factors. ................................ ................................ ................ 85

PAGE 9

9 Abstract of 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 MEASURING URBAN FORM AND EXAMINING ITS IMPACT ON TRAFFIC CONGESTION IN FLORIDA By Ruoniu Wang December 2010 Chair: Ruth L. Steiner Cochair: Andres G. Blanco Major: Urban and Regional Planning Traffic congestion is an increasing problem in the United States. This is especially true in the State of Florida which has been experiencing fast population and job growth. Among many causes, urban form has long been treated as a contributing factor of co ngestion because land use patterns and street connectivity affect transportation decision and the resultant travel behavior. However, c onventional wisdom presents con tradictory opinions about the relationship between urban form and traffic congestion. Its relation is further complicated by many empirical studies differing in study area and scale, research design, data sources, statistical examinations, and methods used to measure urban form and congestion. In this regard the study was conducted to: (1) a da pt the multi dimensional approach to measure urban form at sub metropolitan level; (2) m ake a valuable contribution towards the understanding of urban sprawl in the context of Florida; and (3) p rovide further insight of the impacts that the built environme nt poses on traffic congestion.

PAGE 10

10 These objectives were achieved through exploring the relationship between urban form and traffic congestion by using data from 45 county level urban areas as of 2007 in the State of Florida, measuring 12 indices of urban for m and 3 measures of traffic congestion, grouping urban form indices into 5 distinct factors, and conducting bivariate analyses between variables. The results of the study show that divergent relationships exist between urban form factors and congestion measures. Size/density/mixed use factor and continuity factor are significantly positively correlated to congestion measures; whereas housing proximity factor is significantly negative ly correlated to all three congestion measures. Only the relationship between housing proximity factor and congestion supports the In addition, correlation and factor analyses prove some urban form indices ar e inter correlated and they can be grouped into smaller number of factors. The findings suggest that sprawl is a relative term that depends on which urban form dimension and which kind of land use is addressed Based on these findings it is recommended tha t planners and policy makers should consider directing urban form patterns as an alternative approach to handling traffic congestion

PAGE 11

11 CHAPTER 1 INTRODUCTION Overview It is essential for planners public officials, and the private sector to understand the impacts the built environment has on travel. Conventional wisdom presents con tradictory opinions about the relationship between urban form and t raffic congestion. On one hand, more compact development characterized by higher density and mixed use development has been linked to less car travel and therefore, lower levels of traffic congestion ( Downs, 1992; Gillham, 2002 ) On the other hand, develop ment with high density and concentration is believed to be positively associated with localized congestion (Downs, 2004) Both opinions are supported by many studies that used different data sources, ad hoc research designs, and different study areas. This paper details the use of Geographic Information Systems (GIS) and uses a multi dimensional approach to quantify urban form and traffic congestion in urban areas of forty five counties that are experiencing traffic congestion on freeways or major arterials in the State of Florida. These results enable more precise and comprehensive quantifications of urban form and traffic congestion at the sub metropolitan level. Bivariate correlation is then sought between urban form variables and congestion variables. Th e result of this analysis provides insight on which dimension of built environment is statistically correlated with traffic congestion in Florida. Traffic Congestion and the Built Environment Traffic congestion has become a major local and regional concer n for millions of U.S. metropolitan area residents (Downs, 2004). Nationwide trends show that congestion affects more roads, causes longer delays, and lasts longer than in the past.

PAGE 12

12 Between 1982 and 2007, yearly delay for the average peak period traveler i n urbanized areas of the country has risen from 14 hours to 36 hours. This figure means for that traveler the annual wasted fuel has risen from 9 gallons to 24 gallons and the annual monetary cost has risen from an inflation adjusted $290 to $760 during th e same period. In aggregation of 439 urban areas nationwide, it means that in 2007 traffic congestion consumed 2.8 billion gallons of wasted fuel and 4.2 billion hours of extra time, representing a monetary cost of $87.2 billion, up from a cost of $14.9 bi llion in 1982 (Schrank & Lomax, 2009). Moreover, recent studies show that congestion levels have risen in areas of all sizes, indicating that congestion has been extending to smaller cities. In Florida, traffic congestion reflects national traffic patterns This argument is supported by the study of Blanco, Steiner, Peng, Shmaltsuyev, and Wang (2010) which estimation methods to calculate congestion metrics and costs. Blanco et al. estimated that congestion costs totaled within the range of $6.3 billion to $7 billion throughout the state in 2007. In addition, the overall findings can be summarized as follows: (1) although higher in urban areas, congestion happens in both urban and rural areas; (2) in urban areas, both freeways and major arterials suffer from congestion; (3) coastal areas and Central Florida have higher levels of congestion than other areas; and (4) levels of traffic congestion have been rising in every county f or both urban and rural areas between 2003 and 2006, only decreasing in 2007 due to economic recession (Blanco et al., 2010).

PAGE 13

13 It is imperative to reduce traffic congestion because existing literature on estimating the cost of congestion only presents part of the problem. There are other important social costs, such as delays in shipping and unreliability, which are hard to estimate but account for significant total losses of time (Downs, 2004). Moreover, reaction in allocating excessive resources to mitigate congestion and the urban sprawl caused by congestion add much more to the total costs, yet none of existing literature is able to quantify them. Besides the fact that congestion gives rise to economic inefficiency in terms o f loss of time and fuel and indirect costs of these distortions, it also causes environmental and social problems such as air pollution and psychological stress. Urban form, among other varying factors, is associated with traffic congestion. Urban form can pattern of land uses and their densities as well as the spatial design of transport and communication infrastructure (Anderson, Kanaroglou, & Miller 1996, p. 9 ). It addresses the physical characteristics of the built humanly made, arranged, or maintained (Bartuska 2007, p.5). When seeking the relationship between the built environm ent and traffic congestion, previous studies neighborhood level, measures of urban form usually contain design category and street layout (Cervero & Kockelman, 1997; Ste ad & Marshall, 2001); whereas at larger scales, design category and street layout are not considered and thus urban form and land use are often conceptualized in the same way (Tsai, 2005). For the purpose of this research urban form will be limited to the spatial configuration of land uses.

PAGE 14

14 Traffic congestion is essentially a comparison of vehicle miles traveled (demand or the use of the transportation system) with lane miles (supply or capacity of the system). From this perspective urban form affects traff ic congestion in both conditions On the supply side, street network determines the number of lane miles in a given area. On the demand side, different types of urban form will cau se variation in travel behavior such as travel frequency, length, and mode choice which in turn influences the level of congestion F o r instance, sprawl like urban form characterized by low density, segregated land uses, and poor street connectivity leads to more vehicle mile s travel ed (VMT) and towards this end results in h igher level s of cong estion than compact development Nevertheless, higher travel frequency is likely to happen in compact urban areas than in sprawling areas, resulting in higher level s of congestion (Boarnet & Crane, 2001 ) Another important reason, as addressed by Ewing, Pendall, and Chen (2002), is the conflicting forces presented by the distribution of jobs and housing. Sprawl like urban form with dispersal of jobs and housing allows residents to live close to their workplaces, w hich, if dominant, would ease traffic congestion. However, levels of traffic congestion may rise if jobs housing imbalance and increased VMT per capita resulting from dispersal of jobs and housing is overriding in the system (Ewing et al., 2002). As a by p roduct of post Industrial Revolution planning solutions, urban sprawl is characterized as low density, auto dependent development, and high segregation of land uses in suburban and its outskirts areas. Sprawl has long been criticized as it is believed to b e the contributing factor to many environmental, social, and economic problems. For its negative impact on transportation, sprawl requires high car

PAGE 15

15 dependence and leads to inefficient street layouts. These disadvantages in transportation result from the fa ct that traditional transportation planners seldom considered the influence of urban form on travel patterns when they built infrastructure to accommodate the current and project ed demand (Boarnet & Crane, 2001 ). The separation of urban form from transport ation projects was even formalized in traditional planning process. Responding to this, a number of planning strategies have been proposed to limit automobile use by means of reshaping urban form, among which New Urbanism is arguably the most compelling a nd influential planning idea. New Urbanism has been employed since the late 1980s to combat urban sprawl. One of the earliest and best known works of this type is a new urbanist town in Seaside, Florida, developed by the Miami team of town planners Andres Duany and Elizabeth Plater Zyberk. The design uses and getting people out of t Boarnet & Crane 2001 p.5). The integration of land use and transportation planning is commonly accepted in transportation policies throughout the states, since many planners believe that fundamental change in urban vercome traffic congestion and a ir quality (Boarnet & Crane 2001 p.9). However, whether those planning policies can really work and whether transportation problems can be solved by changing urban form remain in question. To test it, a comprehensive and ca refully designed research should be conducted to explore the relationship between different dimensions of urban form and travelling

PAGE 16

16 indicators in this case those traffic congestion metrics. To achieve this goal, the link needs to be developed though empi rical studies that could accurately quantify those variables. This study presents such an attempt. Study Area The State of Florida has been undergoing fast population growth in the past six decades. According to a report presented by Floridians for a Susta inable Population 2.771 million to 18.538 million, at a growth rate of about 267,000 a year. Population growth was particularly profound in the seven years from 2000 to 2007 as Florida grew by 16.9% during that time period In this same time frame domestic and international immigration accounted for 87.5% of the growth compared to only 12.5% of natural increase ( Floridians for a Sustainable Population, http://www.flsuspop.org/ ). This national recession, the trend of fast population growth in Florida is believed to continue with the fact that baby boomers will peak in their retiring age soon and many of them will Research (2009) projects an average population growth of 317,000 a year between 2010 and 2020. Findings from million people between 2005 and 2060. Fast population growth in Flori da results in profound changes in the built efficient and livable compact urban form. However, residential preference for low

PAGE 17

17 density lifestyles is prevailing statewid e ( Audirac et al., 1990 ) Should government authorities insist on their plan of compact urban form; or should they to some extent allow sprawl like development? This study intends to provide research insight of this issue by linking urban form to one of th e consequences of urban growth traffic congestion. Conceptual Model This study intends to contribute further investigation into how travel patterns correspond to the built environment, by exploring how urban form impacts traffic congestion. It adapts the methodology and the evaluation process of similar research. Urban areas in forty five counties in the State of Florida are examined by using several indicators to measure both urban form and traffic congestion for the year 2007. Urban form is characterize d into seven dimensions: (1) population size, (2) density, (3) continuity, (4) concentration, (5) centrality, (6) proximity, and (7) mixed use. Four dimensions density, concentration, centrality, and proximity, are further developed into two indices by c onsidering two major land uses that cause traffic congestion in peak hours: housing unit and job. In addition, two indices developed in mixed use dimension focus on measuring the interaction of these two land uses: mix of job to housing and mix of housing to job. In terms of traffic congestion, three measures are used : (1) roadway congestion index (RCI), (2) delay per capita, and (3) travel time index (TTI). Other factors, including socio economic conditions, public interventions, and transportation system concerns, are increasingly recognized as interven ing variables in the relationship between urban form and travel characteristics. However, these factors add much complication to the analysis and to measure their influences on the system has not been well explored (Stead & Marshall, 2001). This study focuses on addressing

PAGE 18

18 the link between urban form and traffic congestion by using statistical correlation metho d and by analyzing causal relationship based on system knowledge (such as theory and existing literature). Therefore, this study does not ta ke into account these factors ( s ee Figure 1 1). Objectives Objectives of this research are: 1. Adapt the multi dimensio nal approach to measure urban form at sub metropolitan level; 2. Make a valuable contribution towards the understanding of urban sprawl in the context of Florida; and 3. Provide further insight of the impacts that the built environment poses on traffic congestio n. Organization Six chapters are included in this document The first chapter addresses the justification and value of this research. The second chapter reviews literature related to issues of urban form and traffic congestion and their relationships. The third chapter presents the methodology that is used in this study. The fourth chapter details the research findings. The fifth chapter contains discussions based on the findings. Limitation of this research and recommendations for future research are also discussed in this section. Finally, chapter six draws conclusions for this study.

PAGE 19

19 Figure 1 1. Conceptual model

PAGE 20

20 CHAPTER 2 LITERATURE REVIEW This chapter details the review of literature pertaining to issues of urban form and traffic congestion and their relationships, providing necessary context for the arguments stated in this research study. The p rimary focus of this chapter is on the measur e s of urban form and traffic congestion. The first section starts with the consideration of urban form at different scales. Next characteristics of urban form at sub metropolitan scale are discussed, followed by the review of multi dimensional approach in me asuring urban form. The second section begins with a brief look into the cost of traffic congestion, followed by a discussion of how congestion is measured. Causes of congestion and its solutions are also discussed in this section. The third section descri bes the connection between urban form and traffic congestion, including comparisons of integrated land use transportation models and their results. Finally, a summary is provided at the end of this chapter. Urban Form An understanding of the physical chara cteristics that make up built environments is essential in understanding the relationship between urban form, traffic congestion, transportation modes and other planning outcomes Urban form refers to the spatial configuration of land uses and human activi ties at a certain point in time (Anderson, Kanaroglou, & Miller, 1996; Tsai, 2005). Previous studies present longstanding attempts to interpret urban form; the findings, however, are inconclusive. For example, Cervero and Kockelman (1997) develop three di mensions of the built environment density, diversity, and design (a.k.a. the 3Ds) when examining travel demand. Tsai (2005), on the other hand, classifies urban form based on three categories: (1) density

PAGE 21

21 (the degree of activity intensity), (2) diversity (the land use interactions), and (3) spatial structure pattern (overall land use shape). This classification can be complemented by other two categories design (when focusing on micro level) and transport spatial structure (when focusing on transportati on networks) (Tsai, 2005). Such difference of categorizing the built environment implies that measuring urban form is relevant to the various scales as well as disciplines involved, which is described in the following section. Urban Form at Different Scale s Existing literature has examined different characteristics of urban form at a variety of geographic scales, ranging from local to regional levels. At different scales, the environments are interpreted differently. When studying the relationship between u rban form and energy consumption, Owen (1986) groups structural variables into five scales: regional, sub regional, individual settlement, neighborhood, and building. This typology is adapted and further developed by Stead and Marshall (2001) as they exami ne the relationship between urban form and travel patterns. In their interpretation, urban form (Stead & Marshall 2001, p.114). At the local level, it concerns mix of land uses and the spatial distribution of development. At the neighborhood level, street layout is included in the analysis; yet land use type is not considered in that it is almost homogeneous at this level (Stead & Marshall, 2001). Similar to Stead and M indices into three levels: metropolitan area, city, and neighborhood. He goes on to

PAGE 22

22 certain levels, such as the jobs housing balance variable 1 (such as density) 2 may carry different meanings at different levels and may affect Knaap, Song, Ewing, and Clifton (20 05) develop a new approach of measuring urban form by adding a disciplinary dimension into the classification. In their study, five categories are identified: Metropolitan structure which primarily deals with nationwide census data employed by planners and economists using GIS technology to examine the effects of various dimensions of urban form on urban performance measures. Sub metropolitan structure which is often used by transportation planners and engineers using four step transportation model and focusing on transportation analysis zone (TAZ) 3 to examine transportation behavior. Community design which deals with urban form measures at local level by planners and policy makers using more disaggregated GIS data to explore transportation and other physical activity behaviors. Urban design which is the most disaggregate approach to examine effects of urban form and variation in physical activity on urban phenomena using primary data sources collected through field observation or interviews. Landsca pe ecology which focuses primarily on areas not developed for urban use and examines effects on ecological outcomes. p.8), they do provide a comprehensive and meaningful framework for understanding 1 2 Tsai (2005) addressed that the meaning of density at the metropolitan level was less clear than at the neighborhood level; Knaap, Song, Ewing, & Clifton (2005) pointed out that densities could be computed for many different features at local level because more urban attributes data were available at this level as opposed to regional level. 3 Tr affic Analysis Zone (TAZ) refers to the transportation planning analysis unit with homogeneous socioeconomic characteristics, which form the basis for the analysis of travel movements within, into, and out of metropolitan area (Meyer & Miller, 2001).

PAGE 23

23 urban form. Traditionally, studies on urban form focused on local levels, such as neighborhood and community (Kulash, Anglin, & Marks, 1990 ; Friedman, Gordon, & Peers, 1994; Cervero & Gorham, 1995; NcNally & Kulkarni, 1997). However, this tendency has been offset by increasing number of recent studies focusing on regional level, such as urbanized area and metropolitan area (Galster et al., 20 01; Ewing et al., 2002; Cutsinger, Galster, Wolman, Hanson, & Towns, 2005; Jaret, Ghadge, Reid, & Adelman, 2009). As addressed by Knaap et al., the reasons of such change are threefold: (1) rising concerns over social and environmental issues at regional l evel; (2) broad application of GIS technology in analyzing large scale, complex spatial patterns; and (3) the availability and increasing quality of spatially referenced data. As later described in the methodology part of this paper, this study follows thi s tendency by measuring urban form for U.S. Census defined urban areas (UAs) 4 bounded by county boundaries. In this regard this study examines urban form at sub metropolitan level even though census data and GIS technology are used which correspond to the metropolitan structure category. Dimensions of Sub metropolitan Urban Form Measures of urban form at the sub metropolitan level have mostly been conducted by social scientists primarily for the purpose of quantifying sprawl/compactness and examining the ef fects of urban form on transportation behavior (Tsai, 2005; Knaap et al., 2005). A considerable number of quantitative indices have been proposed by 4 The an urbanized area or an urban cluster (UC), which consists of: Core census block groups or blocks that have a population density of at least 1,000 people per square mile; and Surrounding census blocks that have an overall density of at least 500 people per square mile.

PAGE 24

24 previous studies. These indices can be classified into five categories: size, density, distribution, shape, and diversity. Size. As one dimension of urban form, the size of settlement has long been considered as an important indicator of sprawl (Hess et al., 2001; Tsai, 2005). A great many studies have been done to examine settlement size and its consequences, in particular, travel patterns (Stead & Marshall, 2001). Among these studies, urban population size is most widely used as an index of sprawl (Gordon et al., 1989a; Banister, 1996). An alternative of using area size measure, rather than population size mea sure, is proposed by Hess et al. (2001), as they find out that these two methods are statistically correlated and the former one is more commonly used to measure landscape characteristics. Although statistically sound, this approach is questioned by Tsai ( 2005) in that theoretically land area is not independent from density another important index of sprawl. Therefore, this study decides to utilize population size as an index of urban form. Density. Density is the most commonly used measure of urban form in existing studies (Tsai, 2005). Although density related information are mostly derived from the census data at census tract or TAZ level, a variety of indices have been proposed to measure density. Knaap et al. (2005) list measures of density in previo us studies, including population density (number of persons per acre), household density (number of households per acre), employment density (number of jobs per acre), housing density (number of housing units per acre), and total person density (number of residents plus jobs per acre).

PAGE 25

25 Contrary to the majority of studies that simply expressed density as a ratio (Galster et al., 2001), Malpezzi (1999) ranks population density by census tract and developed specific density indices according to the ranking, in cluding minimum density, maximum density, weighted median density, percentile density, and others. In addition, different measures of variation in population density were proposed by Malpezzi (1999), including: Coefficient of variation: measure of density variation from the mean; Density entropy: measure of proportionate density variation; and Density Gini: measure of difference in density from uniform distribution. (Knaap et al., 2005) Although different measures of density are proven to be highly correlated (Tsai, 2005), Galster et al. (2001) present theoretical arguments which show that some indices are better than others. First, using number of housing units is better than using numbe r (Galster et al. 2001, p.688). Second, residential density is a superior indicator over nonresidential densities, which can easily be affected by economic agglomeration a nd g overnmental regulations. Third developable land area is a better denominator than total land area in calculating density, since undevelopable land, such as water body, may generate misleading results (Galster et al., 2001). Distribution. This measure distributed (Tsai 2005, p.143). Among a variety of indices that have been developed to characterize distribution, three of them are widely used (Galster et al., 2001; Tsai, 2005). The first and simplest approach is to create a grid system on top of the land use

PAGE 26

26 layer across the study area, calculate density for each grid, and measure the proportion of high density grids withi (Galster et al. 2001, p.700). The second one, also known as Gini coefficient, measures riation (standard deviation divided by the mean) of the density of known as Delta index, can be interpreted as the share of land use (e.g., employees or housing units) that would need to shift to achieve a uniform distribution across the study area (Massy & Denton, 1988). Tsai (2005) found that Delta index is superior to others areas in calculating 05, p.143). Shape. There are several sub classifications of this dimension of urban form. Tsai (2005) uses two measures to characterize metropolitan shape: centrality and continuity. Centrality examines whether a study area is developed in monocentric form or polycentric form; whereas continuity measures both the size of discontinuous developments and their distance from the center of study area. Different measures in this category are found in other studies. For example, one index oblong ratio consider s city form as an oblong ellipse and computes the ratio of the shortest and longest diameter (Knaap et al., 2005). Another one, called contiguity, is computed & Guo 20 01, p.7). Diversity. This category can also be called mixed use since it concerns the mix of land uses and activities. Previous studies have developed many diversity indices, such

PAGE 27

27 as jobs housing ratio (Ewing, DeAnna, & Li, 1996), share of single family to multi family housing, and employment diversity (Messinger & Ewing, 1996; Boarnet & Sarmiento, 1998). In addition, there are several ways of measuring diversity. A common way is to use ratios and proportions, which are relatively simple to calculate and ar e easy to interpret. Besides, job housing Gini coefficient, which is calculated using similar equation as Gini coefficient in distribution category, measures the difference of job housing ratio from even distribution (Knaap et al., 2005). Among these measu res, exposure index has been widely used recently to measure job housing balance (Galster et al., 2001; Cutsinger et al., 2005). This index is well explained by Massey and Denton (1988) and is interpreted as the extent to which one particular land use (e.g ., housing units) is exposed to another land use (e.g., nonresidential or employees). Multi dimensional Approach of Measuring Urban Form Existing literature presents the trend of measuring metropolitan urban form by using a multi dimensional approach. Two sets of composite indices have been developed and widely cited. The first one was proposed by Galster et al. (2001) using U.S. Census blo ck data in 1990. In general, Galster et al. develop three steps to measure urban form. They start by classifying land in Urbanized Area (UA) into three land types: developed, developable, and undevelopable. Then a grid system composed by one quarter square mile cell is superimposed over the area for the purpose of areal unit analysis. Finally, they define and measure eight dimensions of metropolitan structure, including: density, continuity, concentration, clustering, centrality, nuclearity, mixed uses, and proximity (Galster et al., 2001). These eight dimensions are then converted into standardized Z scores and are summed up as a sprawl index. High Z scores indicate low levels of sprawl.

PAGE 28

28 Another set of composite index is developed by Ewing et al. (2002) usi ng three primary sources of data: the Census of Population and Housing, the Annual Housing Survey, and the Census Transportation Planning Package. Ewing et al. construct a hierarchical structure to compute the overall sprawl index. They develop four subcat egories of ur ban form density, mix, centerness and streets; then multiple indicators are used to measure each of the four subcategories (Ewing et al., 2002). approach is superio s, in that: 1. Ewing et al. use more data sources to enrich the measures of urban form dimensions; and 2. by introducing a hierarchica l structure Ewing et al. avoid the problem of multi method is cr iticized by Jaret et al. (2009) who argue that three of those four subcategories are empirically inter correlated 5 prove to be an important i mprovement. First, they combine are derived from 14 land use indices: Density (Housing Unit Density of Deve lopab le Land and Job Density of Developable Land) Continuity (Micro Continuity and Macro Continuity) Concentration (Housing Unit Concentration and Jobs Concentration) Centrality (Housing Centrality and Job Centrality) 5 Jaret et al. (2009) find that density, mix, and streets are inter correlated and can be reduced to one dimension, while centerness is inde pendent.

PAGE 29

29 Proximity (Housing Proximity, Job Proximity and Jobs to Housing Proximity) Mixed use (Jobs to Housing and Housing to Jobs Mixed use) Nuclearity. Second, instead of using different cell size to compute urban form indices as Galster et al. did, all indices are calculated on the base of uniform 1 sq uare mile cell grid. Third, they adopt a new metropolitan spatial area called Extended Urban Area defined urbanized area, as well as each additional outlying square mile cell comprising the metropolitan statist ical area that has 60 or more dwelling units and from which at least 30% of its workers commute to the employing additional data sources to the analysis, including 1990 CTPP as used by to classify land types. More importantly, Cutsinger et al. goes on by reducing 14 conceptually distinct indices into 7 distinct factors using factor analysis. This approach is promising since it effectively handles the problem of inter correlation among urban form dimensions. Also, fewer urban form variables simplify further analyses when linking urban form with its causes or consequences. A recent study by Jaret et al. (2009) reduced urban form dimensions into two categories: density mixed land use (used by Galster et al.) and centers (used by Ewing et al.). However, whether this simplified classification of metropolitan urban form is applicable requires further stud y. Traffic Congestion Traffic congestion is a common problem in the U.S., especially in most large urban areas. It is considered as one of the most important policy problems in surveys of

PAGE 30

30 elected officials and public opin ion polls (Boarnet & Crane, 2001 ; S arzynski, Wolman, Galster, & Hanson, 2006). Congestion is expensive. Sc hrank and Lomax (2009) estimat e that congestion cost, including delay cost and extra fuel cost, totals $87.2 billion in 439 urban areas in the U.S. in 2007. Down to individual level, th is figure represents 36 hours extra traveling and 24 gallons of extra fuel spent by average peak traveler in those selected urban areas in 2007 (Schrank & Lomax, 2009). Congestion is also getting worse. Previous studies have shown increased travel delay an d congestion costs in regions of all sizes since 1982 (Downs, 2004; Schrank & Lomax, 2009). This trend suggests that the congestion problem is not likely to be solved in the near future (Sarzynski et al., 2006). Measures of Traffic Congestion Existing tran sport planning literature presents a diverse, yet inconsistent, set of congestion indicators. Sarzynski et al. (2006) summarizes them into two basic concepts a verage d aily t raffic (ADT)/lane and c ommute time. The first one measures the average number of daily vehicles per roadway lane. Data of freeway ADT/lane are System (HPMS) in all urbanized areas since 1989 report year. ADT/lane data for lower arterials and collectors) can be attained through the state department of transportation. Commute time measures the average home to work travel time and the data can be obtained from U.S. Census Bureau, with data availability dating back to 1980. By compa rison, ADT/lane directly estimates congestion in the sense that it evaluates operational performance of roadways; whereas Commute time is an indirect indicator of congestion level in that it is influenced by both travel distance

PAGE 31

31 and speed, where only lower speed on a given road directly suggests higher level of congestion (Sarzynski et al., 2006). Evolving from these two basic indicators, two widely adopted indices roadway congestion index (RCI) and travel time index (TTI), are developed by the Texas Tran sportation Institute (Schrank & Lomax, 2009). As a modified version of ADT/lane, RCI refers to a ratio of existing ADT/lane on major roadways to a threshold of the starting point of congested conditions (e.g., 5,000 ADT/lane for major arterials and 14,000 ADT/lane for freeways). An important advantage of RCI over ADT/lane is that RCI takes into account different roadway types at one time by using VMT of each roadway type as a weighting factor. Unlike RCI which is a measure of average daily traffic condition s, the TTI converts ADT/lane to an estimate of the average speed of travel during peak hours (in the study conducted by the Texas Transportation Institute, peak hour travel is 50% of daily travel) and compares this with speeds under free flow conditions (e .g. 35 mph for arterials and 60 mph for freeways) (Sarzynski et al., 2006; Schrank & Lomax, 2009). When linking land use and traffic congestion, Sarzynski et al. (2006) employs three measures of traffic congestion: Commute time: the average one way travel time to work (in minutes; averaged across all modes) as reported by the US decennial census (US Census Bureau, 2004). ADT/lane: the average daily traffic per freeway lane (in vehicles per freeway lane per day) as reported by the Federal Highway Administration (Federal Highway Administration, 2001). Delay per capita: the annual peak hour highway congestion delay per traveler (in hours per year per person) as computed by the Texas Transport Institute (Schrank & Lomax, 2004). (2006, p.611)

PAGE 32

32 Each cong estion indicator has its own limitation. For example, Commute time computes travel time across different travel modes; ADT/lane and Commute time are averaged numbers across time and space; and delay per capita is limited in peak hours and only for major ar terials and freeways (Sarzynski et al., 2006). Given these limitations, a multi dimensional approach of measuring traffic congestion is needed. Understanding Traffic Congestion Previous studies on causes of traffic congestion can be grouped into two classe s the inadequacy of the supply of roads and travel modes, and excessiveness of travel demand. On the supply side, road building and other transport investments have not kept pace with the increase of automobiles on the streets (Downs, 2004; Sarzynski et al., 2006). Besides the fact of rise in private vehicle ownership, other empirically evidenced explanations include induced travel 6 and triple convergence 7 (Downs, 1992; Hills 1996 ). In addition, public transit is not believed to be the cure for all metropolitan areas since many urban densities in the U.S. are too low to support its operation (Downs, 2004). On the demand side, population and job growth is considered to be the most direct factor that contributes to traffic congestion (Downs, 2004; Shra nk & Lomax, 2009). However, other studies have shown little relationship between city size and commute time (Gordon, Kumar, & Richardson, 1989a; Gordon & Richardson, 1994). Another possible explanation of congestion addressed by previous studies is the ris e of household incomes, which increases travel demand by encouraging more overall travel 6 As defined by Hills (1996), induced travel refers to increases in VMT due to roadway improvements. 7 According to Downs (1992), triple convergence happens when roadway improvements result in increased peak period travel, due to shifts from other routes, times and modes.

PAGE 33

33 and by resulting in the increase of private vehicles per household (Crane, 1996; Gillham, 2002; Downs, 2004; Sarzynski et al., 2006). Responding to the causes of traff ic congestion, policy remedies are sought on both supply and dem and sides (Boarnet & Crane, 2001 ; Downs, 2004). Supply expansion, such as road building and transit improvements, is a common strategy advocated by transportation planners and implemented by t he U.S. government authorities at all levels. However, this strategy is increasingly criticized by scholars and government officials as they point out that the effect of supply expansion is often offset by fiscal constraints, latent demand 8 and excessive increase in travel demand (Tay lor, 1995; Boarnet & Crane, 2001 ). On the demand side of solutions, transportation demand management (TDM), such as dedicated lanes for car poolers, is supported by transportation planners. In addition, congestion pricing is a dvocated by many urban economists despite criticism of its feasibility and effe ctiveness (Boarnet & Crane, 2001 ). Another widely adopted strategy of congestion mitigation involves regulating land use patterns and better integrating land use and transportat ion system. This strategy is based on the belief that in many urban areas, traffic congestion stems from inefficient use of land which is associated with rising VMT and commute time. Downs (2004) lists congestion mitigation policies related to land uses, i ncluding: increasing minimum residential densities, transit oriented development (TOD), limiting growth in local communities, improving the jobs/housing balance, and concentrating jobs in a few suburban clusters. Regardless of various policies, existing li terature presents an unclear and complex relationship between traffic congestion and physical environment (Boarnet 8

PAGE 34

34 & Crane, 2001 ). To better understand impacts of the urban form on traffic congestion, the following section provides a detailed review of thi s issue. Research on Urban Form and Traffic Congestion Overview There is much literature exploring the relationship between urban form and travel patterns (Gordon, Kumar, & Richardson, 1989a; NcNally & Kulkarni, 1997; Stead & Marshall; 2001); however, stud ies of urban form and traffic congestion, especially those using multi dimensional approach and controlling for confounding factors, ar e limited (Boarnet & Crane, 2001 ; Sarzynski et al., 2006). T his can be explained by the fact that travel behavior is cons idered to be the bridging factor between urban form and traffic congestion (Boarnet & Crane, 2001 ). Conventional wisdom presents a contradictory view in understanding the link between urban form and congestion. On one hand, urban sprawl is believed to be p rominent cause of traffic congestion, due to separated and segregated land uses and the subsequent need for longer and more frequent vehicle trips (Downs, 1992; Gillham, 2002). On the other hand, congestion is alleged to be positively related to the dense and concentrated development due to trip convergence both spatially and temporally (Downs, 2004). This paradox is evidenced by many studies consistently showing that congestion has been getting worse over the past two decades both in the areas that experie nced fast urban sprawl and those with large population and/or job growth (Wheaton, 1998; Schrank and Lomax, 2009). Although congestion is found to be growing worse, previous studies show that commute times have remained roughly constant during the study pe riod (Boarnet & Crane, 2001 ). One reason for this, as explained by Gordon and Richardson (1991), is that both jobs and residents chose to relocate to less congested urban or suburban

PAGE 35

35 areas, which results in more frequent and longer trips. This explanation means that (Boarnet & Crane 2001 p.26). This is statistically proven by a later study conducted by Ewing et al. (2002), which shows that the overall sprawl index does not have a marginal relationship to either aver age commute time or traffic delay per capita when population size and other socio demographic variables are controlled. As one of the pioneer studies that attempted to address the link between rch unfolds the important fact that these two variables are interchangeably, causally related. That is, urban form will be a contributing factor of congestion; and congestion can also lead to the change of urban form Acknowledging the complexity of causality between these two variables, this study, however, tend s to consider urban form as the cause and congestion as the consequence. Models Measuring Urban Form and Congestion Given the interaction of urban form and traffic congestion as addressed abov e, previous studies have developed different models based upon various methodological concerns to explore the relationship between them. The results turn out to be quite diversified due to variation of geographical and temporal selections, comparative data used, and statistical methods involved. Izraeli and McCarthy (1985) first brought commute time as an indicator to seek the relationship between urban form and congestion. They use cross sectional data during year 1975 1977 from Annual Housing Survey for 6 1 metropolitan statistical areas. In terms of statistical method, they use linear regression and correlation between independent variables (commute behaviors) and dependant variables (population size and density). Control variables in their study include i ncome and education levels,

PAGE 36

36 housing age and public transport usage. They find that commute time, which relate s to traffic congestion, is significant ly positive ly influence d by population size and density. Like Izraeli and McCarthy, Gordon, Kumar, and Richa rdson (1989b) explore the relationship between commuting time as an indicator of congestion and urban size and density. Conversely, Gordon et al. (1989b) look at larger data samples from 82 metropolitan statistical areas. In addition, they use U.S. Geologi in year 1980 and develop three densities (residential densities, industrial densities, and commercial densities) for the independent variables. In terms of dependant variable, commuting time is grouped by traffic modes (automobile and public transit). By running regressions, they find that at statistically significant levels, residential and commercial densities are positively associated to commute time, while reversed relationship is observed between industrial densities and commute ti me. Malpezzi (1999) studies the relationship between predicted median population density and average commute time. He uses 1990 U.S. Census data and includes all U.S. metropolitan statistical areas. In addition, by incorporating a transit variable bus se at miles per capita from 1980 Census, an instrumental variable is constructed in his model for the sake of controlling the influence of transit supply on commute time. Contrary to the findings from Izraeli and McCarthy (1985) and Gordon et al. (1989b), Mal pezzi finds a strong but negative relationship between predicted density and average commute time. Ewing et al. (2002) first introduces multi dimensional variables for both urban form and traffic congestion. Two cross sectional datasets (1990 and 2000) are used in the study for 83 metropolitan statistical areas. In terms of urban form, they select the

PAGE 37

37 primary factor for each of four principal pre defined components including density, land use mix, degree of centering, and street accessibility Then an over all sprawl index is computed by combining those four components. On congestion side, in addition to commute time, Ewing et al. include congestion delay per capita, which conceptually is superior to commute time as a direct measure of congestion. In their m odel, four factors are considered as control variables: population size, income, percentage of working population, and household size. A correlation study is chosen as the statistical method to explore the statistically significant associations between dep endant and independent variables. As a result, they find that density factor is inversely related to VMT per capita and traffic fatality rate, and the centers factor is inversely related to delay per capita and traffic fatality rate, both at statistically significant levels. Findings from Ewing et al. (2002) indicate that the more highly compact development equals less congestion. etropolitan statistical areas which usually derive from aggregations of county level data Gordon et al. first use data on two cross (Gordon et al. 2004, p.9). Through ordinary least squares regression 9 test, they find that population density shows a statistically negative significant relationship with commute time. 9 Ordinary least square regressio n, which is based on the least squares method of finding regression parameters, is one kind of ordinary linear regression. Detailed explanation is available at Statistics.com: http: //www.statistics.com/resources/glossary/o/olsregr.php

PAGE 38

38 Sarzynski et al. (2006) repr esent the most comprehensive study on the link between urban form and traffic congestion to date. A major innovation of their study is that they acknowledge the issue of time lag between traffic congestion and transport network response. By considering lan d use patterns in both 1990 and 2000 and congestion conditions only in 2000, they first incorporate a time lag in their model. In the conceptual model, they develop three congestion indicators commute time, ADT/lane, and delay per capita, which are corre lated with seven land use measures: density, continuity, concentration, centrality, proximity, mixed use, and nuclearity. In addition, three measures of transport network (roadway provision, public transport provision, and rail) are included in the model a s explanatory variables and the model is controlled for other influencing factors such as population growth rate, and changes in income and household size between 1990 and 2000. In terms of statistical method, they start by using preliminary bivariate anal ysis to examine the relationship between land use factors in 1990 and congestion measures in 2000. Then three multiple regression analyses are done to further explain the statistical relationship between land use and traffic congestion. In results, they fi nd that at statistically significant levels density and continuity factors are positively associated with ADT/lane and delay per capita; housing centrality is also positively associated with delay per capita; while housing job proximity is negatively assoc iated with commute time. Taken together, existing literature presents an inconclusive relationship between urban form and traffic congestion. Density is the most commonly used measure of urban form when examining its relation to congestion, which is usuall y measured by commute time. However, less conclusive results are shown in previous studies. For

PAGE 39

39 example, density is found to be positively associated with commute time in studies by Izraeli and McCarthy (1985) and Gordon et al. (1989), while the opposite r elation is addressed by Malpezzi (1999) and Gordon et al. (2004). In terms of density to VMT per capita and centrality to delay per capita, Ewing et al. (2002) find that both pairs are negatively correlated as opposed to the positive relationship found by Sarzynski et al. (2006). The conflicting results may have risen for two main reasons. First, due to the coexistence of two contradictory forces as addressed earlier (higher degree of sprawl, more congestion versus higher degree of compact development, more congestion), findings vary as one force overrides the other according to the year selected and number of areas studied. Second, the differences of conceptual model in previous studies are likely to generate contradictory results. For instance, urban form is measured in different ways, so is congestion; control variables differ in the number and type as explored by different studies; and whether land use and congestion data are obtained in the same year or in different years accounts for time lag issue. Sum mary Traffic congestion is an increasing problem in the United States and, if not properly addressed, may pose detrimental effects on the form and function of our built environments. This is especially true in the State of Florida which has been expe riencing fast population and job growth. Among many causes, urban form has long been treated as a contributing factor of congestion because land use patterns and street connectivity affect travel behavior. However, the influence of urban form on traffic co ngestion is not clearly understood due to the fact that contradictory views exist in theory. Its relation is further complicated by many empirical studies differing in study

PAGE 40

40 area and scale, research design, data sources, statistical examinations, and metho ds used to measure urban form and congestion. A review of existing literature suggests that it is imperative to use a multi dimensional approach to measure both urban form and congestion. The purpose of this research is to link urban form and congestion. T o achieve this goal, both variables need to be quantified. The following chapter details the methodology that is used in this research.

PAGE 41

41 CHAPTER 3 METHODOLOGY Overview of Study Design This chapter provides a detailed description of the methodology that is used in this research and the reasoning for it. Overall, th e study used integrated land use transportation model to examine the relationship between urban form and traffic congestion in urban areas (UA) of 45 counties across Florida Both urban form and co ngestion were measured using multi dimensional approach. After urban form variables were calculated, 12 indices were grouped into 5 factors using factor analysis; then correlation was sought between urban form factors and congestion measures. As f or the da ta, even though it would have been better to have at least two observations to look for the relationship between congestion and urban form (Stead & Marshall, 2001), the study is cross sectional for year 2007 due to inconsistent study samples in previous ye ars with regard to traffic congestion measures. Study Area Selection Based on the findings of Blanco et al. (2010), UAs in 45 counties across Florida underwent area wide congestion in year 2007 and thus, they were selected as study population in this resea rch ( s ee Figure 3 1). The United States Census defined UA has long been selected to address the connection between urban form and travel patterns by previous studies focusing on sub metropolitan scale. As described in the literature review part of this pap er, the use of EUA in like studies tends to be more theoretically sound than UA because EUA includes development in urban fringe and outskirt which also contribute to the regional congestion. This study, however, only considered UA when measuring urban for m because of consistency concerns in two aspects. First,

PAGE 42

42 urban areas in 45 counties of this study differ considerably in size and population distribution. Smaller size UAs have sharper decline of population density from the center to peripheries than large r size UAs. I n this regard, u sing the same criteria to define EUA may generate erroneous results due to inconsistency of area size. Second, study area should be the same for both urban form and congestion. The fact that congestion measures cannot be calcul ated using EUA boundary limits the use of EUA for urban form. Classification of Land Type As pointed out by Cutsinger et al. (2005), measures of urban form should be based on the net land that is developed or have the potential to be developed in the UA. T herefore, areas such as water bodies and wetlands should not be included when measuring dimensions of urban form. In this regard Cutsinger et al. grouped urban land into three types developed land, developable land, and undevelopable land, by utilizing d ncept of classifying land types But instead of using nation wide data sources, this study uses state level Florida Par cel Data by County (FPDC) obtained from Florida Geographic Data Library (FGDL). Compared to NLCDB, FPDC contains land information that is the most accurate and in detail available for the state and is up to date ( in contrast, the most updated dataset avail able for NLCDB is in year 2001). Preferably, dataset of FPDC in 2007 should have been used in order to be consistent with other datasets. This study, however, used 2009 dataset because it only affected one dimension of urban form

PAGE 43

43 (continuity) and the influ ence to statistical result was negligible 10 Ninety nine land use categories identified by FPDC were grouped into three land types. Appendix A presents the specific grouping list. The basic process used for grouping land types is outlined below : Downloaded FPDC from FGDL site, http://www.fgdl.org/metadataexplorer/explorer.jsp Used the Clip tool in ArcGIS to clip parcels of each county by using the U.S. Census defined UA boundaries. Created a new field in the distribution table of clipped parcel shapefile. Used Select By Attributes tool to select land uses that belong to each land type. Populated a unique code in the newly created field for each selected land type. Establishing the An alysis Areal Units In this study, dimensions of urban form were calculated through using three datasets: (1) locations of employed residents, (2) job locations, and (3) locations of housing unit. The first two provide information at the census block level and were obtained from the U.S. Census. The third one was derived from FPDC and therefore, provides information at parcel level. In order to unify the calculation of each dimension of urban form, spatial data need s to be aggregated at a certain level. Towa rds this end, the areal units (zonal objects) were generated and superimposed on top of other layers. The characteristics of the predefined areal unit influence aggregate values of the urban form measurement. Previous studies have shown that statistical re sults will differ in variation of the spatial units used for aggregation (Openshaw, 1984). Therefore, the size 10 The change of urban form is slow and in UA most of the land has been developed. Therefore, area of land that has converted from developable land type to developed land type within two years is subtle compared to the total area of land in an UA.

PAGE 44

44 and location of areal unit need to be carefully selected to prevent statistical bias known as Modifiable Areal Unit Problem (MAUP) (Guo & Bhat, 2 004; Kwan & Weber, 2008). Size of Areal Unit Curtsinger e t al. (2005) used one square mile cell as the basic areal unit to calculate land use indices. A recent study conducted by Arafat, Steiner, Zwick, and Srinivasan (2010) found that the change in entrop y 11 value is minimal for neighborhood sizes with the square cell size starting from 6.25 square miles (2.5*2.5 mile). Lower value s of entropy indicate more stable urban form measurements Therefore, in this study the 6.25 square miles cell size was applied. Location of Areal Unit To further avoid MAUP, urban form indicators were calculated four times in this study based on different spatial location of areal unit. The first round, named Base Scenario, established areal unit network by overlapping the center of one areal unit with t he geographic center of Florida and then expanding areal units across the state. Are al unit grids for the other three scenarios West Scenario, South Scenario, and Southwest Scenario, are generated by shifting areal units in Base Scenario 1.25 miles to th e west, south and southwest, respectively ( s ee Figure 3 2). Mean values of the four scenarios were used for further analyses. The GIS steps taken to generate and shift areal units are summarized below: Downloaded Boundary of the State of Florida polygon l ayer from Florida Coastal Everglades site, http://fce.lternet.edu/data/GIS/?layer=state Generate center point the State by using Feature to Point tool. 11 The entropy is an index that is calculated at neighborhood level to capture the land use mix (Arafat et al., 2010).

PAGE 45

45 Created areal unit grid that covers the e ntire state boundary with the cell size 2.5*2.5 mile by using Create Fishnet tool. Used Feature to Polygon tool to convert grid into polygon. Used Feature to Point tool to generate centroid of each cell. Created areal unit polygon layer of Base Scenario by moving polygon layer to Moved areal unit grid of Base Scenario to generate areal unit grids of other three scenarios. Operationalizing Variables Dependant Variables In this study, t raffic c ongestio n was operationized into three indicators: roadway congestion index (RCI), delay per capita, and travel time index (TTI). Measurement of these three indicators was conducted using one of the two sets of disaggregate methodology, called Disaggregate Roadway developed as a model in Microsoft Excel Software in the research conducted by Blanco et al. ( 2010). This method of estimating congestion at lower levels of aggregation was an adaption of Texas Transportation called rcitop30, which was obtained from FDOT Roadway Characteristics Inventory for the year 2007, including historical roadway and tr affic data on traffic densities, number of lanes, and lane because the data for this study contains the most complete data that is available. Regardless of the difference in datasets, t his study follows Blanco et al. s approach and only considers freeways and major arterial streets in the calculation process. D etailed explanation and measurement of each traffic congestion indicator are provided below.

PAGE 46

46 See Blanco et al. (2010) for a detailed description of how these three measures are calculated. The first indicator RCI, measures the density of traffic across the urban area by combining VMT and lane miles of roadway into a ratio. The a mount of travel on freeways and major arterials are presented by this ratio. It is then divided by a value (14,000 for freeway and 5,000 for arterials) that represents the start of congestion in the system. VMT also serves as a weighting factor in the calculation process in order to count in the effect of different amount of regional vehicle t ravel on different road types ( s ee Eq.3 indicates an undesirable level of area wide congestion of the index value is greater than or equ 4). As a primary indicator of traffic macroscopic level (Schrank & Lomax 2009, p.A 4). (Eq. 3 1) The second indicator delay per capita, accumulates total annual hours of delay per employed resident on both types of facilities during the peak periods of working days. It was calculated through dividing annual person delay by the number of employed residents 12 in the region. Annual per son delay is the sum of both recurring 12 Theoretically, total number of travelers during peak period (including both home work trav el and non commuting travel) should be used in the calculation process because Annual Person Delay was calculated by considering vehicle movements of all purposes. This study used number of employed residents in UA to replace total number of travelers duri ng peak period because of several reasons: (1) the latter dataset is not available for the study area to current knowledge; (2) people tend to minimize non commuting travel during peak period, which is evidenced by vehicle occupancy in Florida that is lowe r in peak period than in rest of the days (vehicle occupancy data are available from National Household Travel Survey site: http://nhts.ornl.gov/index.shtml ); (3) only commuters were considered in Urban Mobility Report 2009 to calculate Delay per Capita (Schrank & Lomax, 2009) ; and (4) testing

PAGE 47

47 and incident delay, multiplying by the average vehicle occupancy and by 250 working days per year. The third indicator TTI, computes the ratio of travel time during peak periods, including both recurring and incide nt conditions, to the travel time at free flow conditions. It is a measure of average amount of time that takes during peak period compared to non rush hours. Free flow conditions are represented by speeds of 60 mph on freeways and 35 mph on major arterial s. Speeds less than that are an indication of congestion and result in the TTI being greater than 1. Independent Variables The establishment of urban form dimensions in this study was developed by Galster et al. (2001) and Cutsinger et al. (2005). Due to d ata constraints and differences in conceptual framework, a couple of urban form dimensions addressed in th o se two studies were excluded in this research. For instance, this study does not take into account nuclearity 13 dimension because it involves measurem ent of urban form at the regional level. Also, clustering 14 dimension is excluded because it requires information on nonresidential uses and because previous studies do not specify how it is measured. In addition, population size is a newly added dimension of urban form compared to those developed by Galster et al. and Cutsinger et al. It was added because Stead and Marshall (2001) found that population size contributed to the variance of travel patterns results of Delay per Capita using employed residents in UA for three counties are very close to those in Urban Mobili ty Report 2009. 13 14 has been tightly bunched to minimize the (Galster et al. 2001, p.691).

PAGE 48

48 and because of the fact that UAs at the county level a cross Florida vary considerably in size. To this end, seven dimensions of urban form were finalized in this study: population size, density, continuity, concentration, centrality, proximity, and mixed use. Detailed definition and illustration of each indic ator is provided below. 1. Population size The number of employed residents living in UAs per county across the State. 2. Density in an intensive manner relative to land area capable of Cutsinger et al. 2005, p.238) ( s ee Figure 3 3, A). Two indices are measured: Housing Unit Density the average number of housing units per square mile of developable land in the UA. Job Density the average number of jobs per square mi le of developable land in the UA. (Cutsinger et al. 2005, p.238) 3. Concentration Cutsinger et al. 2005, p.239) ( s ee Figure 3 3, B). They are measured as: Housing Unit Concentration the percentage of housing units that would need to move in order to produce an even distribution of housing units within all cells across the UA. Job Concentration the percentage of jobs that would need to move in order t o produce an even distribution of jobs within all cells across the UA. (Cutsinger et al. 2005, p.239) 4. Mixed use Cutsinger et al. 2005, p.240) ( s ee Figure 3 3, C). The Mixed use ind ices are based on exposure (P*) 15 indices: Mix of Job to Housing the average number of jobs in the same cell as a housing unit. Mix of Housing to Job the average number of housing units in the same cell as a job. (Cutsinger et al. 2005, p.240) 5. Continuit y Cutsinger et al. 2005, p.238) ( s ee Figure 3 mile units within the UA in which 50% or more of the land that is or could be developed 15 use (e.g., housing units)

PAGE 49

49 Continuity the calculation process. 6. Cent rality Cutsinger et al. 2005, p.239) ( s ee Figure 3 3, E). The core of the UA per county is defined as the location of county government building. It is measured as: Housing Unit Centrality the ratio of the average distance to county government building of centroids of the cells comprising the UA to the weighted distance to county government building of a housing unit within the UA. Job Centrality the ratio of the average distance to county government building of centroids of the cells comprising the UA to the weighted distance to county government building of a job within the UA. (Cutsinger et al. 2005, p.239) 7. Proximity ing units or jobs are close to each other Cutsinger et al. 2005, p.239) ( s ee Figure 3 3, F). Two proximity indices are measured 16 : Housing Unit Proximity the ratio of the average distance among centroid s of cells in the UA to the weighted average distance among housing units in the UA. Job Proximity the ratio of the average distance among centroids of cells in the UA to the weighted average distance among jobs in the UA. (Cutsinger et al. 2005, p.239 2 40) Quantifying Independent Variables Appendix B presents an example of step by step procedures on how to calculate each dimension of urban form for the Base Scenario of a chosen county using ArcGIS and Microsoft Excel Software. To measure the centrality a nd proximity dimensions of 17 Analysis that cannot be conveniently accomplished with out of the box ArcGIS 16 Another proximity index Housing Unit Job Proximity was developed by Cutsinger et al. (2005). However, it is excluded from this stud y because of time constraint. 17 a free software available in SpatialEcology.com site: http://www.spatialecology.com/htools/download.php

PAGE 50

50 ( http://www.spatialecol ogy.com/htools/overview.php) used to calculate the direct distance between different land use locations. In addition, data of housing units, rather than employed residents, were chosen to calculate centrality and proximit y. The justifications for this decision are twofold. First, existing literature indicates that housing unit better represents physical condition than employed residents in quantifying urban form (Cutsinger et al., 2005). Second, datasets used for calculati ng these two dimensions of urban form should be consistent with those of other dimensions in this study. In order to solve the problem of different aggregate levels of two datasets (dataset of housing units is at parcel level; whereas dataset of jobs is at census block level) and eventually to make the indices of centrality and proximity more comparable, dataset of housing units was aggregated from parce l level to census block level ( s ee Step 18 in Appendix B for detailed aggregation method). Since informat ion of jobs and employed residents are aggregated at census block level, erroneous results may be computed if one census block crosses multiple cells. This problem was solved by allocating number of jobs and employed residents into each cell based on the p ercentage of census block a rea that falls into each cell ( s ee Step 5 and 9 in Appendix B for detailed description). Statistical Method After measures of urban form and congestion were calculated, descriptive statistical analysis was applied to examine patt erns of variables. Bivariate correlation analyses using SPSS software were conducted as the main tool to further examine the relationship between variables. This statistical method was selected because it could sufficiently answer the research question by measuring the direction and magnitude of relation between two variables. Correlation analysis was first applied amongst urban

PAGE 51

51 form indices and congestion measures, respectively, followed by factor analysis to reduce 12 urban form indices into smaller numbe r of factors. In addition, rankings of urban form factors and congestion measures are provided to understand sprawl patterns. Finally, another round of correlation analysis was applied between urban form factors and congestion measures. It is important to note that whether conducting factor analysis for urban form or not does not affect the results of correlation between urban form and congestion 18 Rather, factor analysis was done in order to answer the question posed in literature part of this paper. Also, the results of factor analysis in this study characteristics of sprawl in general. Summary The methodology described in this chapter helps quantify traffic congestion and patterns of urban form and analyze their relationship under the study population. Methods of measuring urban form and congestion used in this study are integration and adaptati on of like methods developed by Blanco et al. (2010), Cutsinger et al. (2005), and Sarzynski et al. (2006). However, changes were made in methodology and data sources, as summarized below: 1. Urbanized area (UA), instead of Extended Urban Area (EUA), was sele cted in measuring urban form and traffic congestion. 2. Florida parcel data from FGDL, instead of National Land Cover Data Base from USGS, were used to classify land types. 18 A test was done to compare the cor relation results with and without doing factor analysis. It turned out that both correlation analyses between urban form and congestion yielded the same results the significant correlation pairs between factor and congestion are the same as those between indices and congestion.

PAGE 52

52 3. To minimize Modifiable Areal Unit Problem (MAUP), four scenarios with different locati ons of areal unit system were conducted. 4. 2.5*2.5 mile areal unit size, instead of 1*1 mile, was used to measure dimensions of urban form. 5. A more complete dataset employed to measure traffic congestion. Figure 3 1. Map of study areas

PAGE 53

53 Figure 3 2 Map of county population

PAGE 54

54 Figure 3 3 Change of a real u nit s ystem for f our s cenarios in Miami Dade County. A) Base Scenario. B) West Scenario. C) South Scenario. D) Southwest Scenario.

PAGE 55

55 Figure 3 3 Continued

PAGE 56

56 A B C Figure 3 4 Dimension s of u rban form. A) Density. B) Concentration C) Mixed use D) Continuity. E) Centrality. F) Proximity. Adapted from Galster et al. (2001).

PAGE 57

57 D E F Figure 3 4 Continued

PAGE 58

58 CHAPTER 4 FINDINGS AND RESULTS This chapter provides results of the study based upon the methodology described in the previous chapter. Corresponding to each step in the analytical process, this chapter is divided into six sections. The first section intends to interpret th e difference in urban form outcomes generated by four scenarios. The second presents findings and statistical descriptions of both urban form and traffic congestion outcomes. The third tries to address the correlation for urban form indices and for congest ion measures. The fourth focuses on the findings of factor analysis on urban form. The fifth interprets rankings of both urban form and congestion measures across the study population. Finally, the last section explains the correlation between urban form f actors and congestion measures. Since all the results and findings in the following section of this chapter is based upon analysis of the previous one, the discussion of the findings and results, such as comparison of results with existing literature and i mplications of the study, are incorporated into this chapter to validate results and provide justifications for following analyses. Areal Unit Effect on Urban Form Outcomes Urban form outcomes of the 45 county level UAs for four scenarios Base, South, We st, and Sou thwest, are listed in Appendix C ( Table C 1 to Table C 4 ). Of the 12 indices, 9 are influenced by the change of areal unit location. In order to better understand the impact of areal unit location on urban form outcomes, for each of the 9 urban form indices in 45 county level UAs, the minimum and maximum values calculated in four scenarios are selected and combined into a

PAGE 59

59 ratio ( s ee Table 4 1). For all 9 urban form indices, high ratio value indicates strong influence of areal unit location on urb an form outcomes. As indicated by Table 4 1, the impact is substantial, which confirms the necessity of conducting the scenario calculation process in this study. In addition, some interesting patterns can be observed from Table 4 1. First, on average, thr ee indices housing unit concentration, housing to job mixed use, and job to housing mixed use are more influenced by areal unit location than other indices. The average ratios for these three indices are 1.33, 1.37, and 1.43, respectively; whereas aver age ratios for other six indices are within the range of 1.05 to 1.17. Second, variation in areal unit location tends to exert a stronger influence on UAs with smaller area size than those with larger area size. For instance, the Hendry UA has the highest values of housing unit concentration ratio, and ratios of both mixed use indices; the Bradford UA has the highest values of continuity ratio and job centrality ratio; and the Desoto UA has the highest value of housing unit centrality ratio. All three UAs h ave small UA area sizes. This pattern can be explained by the fact that UAs with smaller area size have relatively larger areal units, which, by changing their locations, can influence more on statistical results than UAs with larger area size. Previous st udies, such as Cutsinger et al. (2005) and Sarzynski et al. (2006) did not consider the issue of areal unit location since they considered large metropolitan EUAs and used smaller areal unit size than in this study. This study extends the study population into small UAs. Therefore, scenario calculation was developed to offset the impact of areal unit location on urban

PAGE 60

60 form outcomes. The mean value of four scenarios for each urban form outcome is computed and used for the rest of the analysis ( s ee Table C 5 ) Description of Data Characteristics Since no previous study has addressed as many urban form indices and their relationship to traffic congestion in the context of Florida as this study, and since urban form indices in this study are adaptations of previ ous studies, it is necessary to discuss the statistical attributes of the indices for both urban form and traffic congestion. Appendix C lists traffic congestion outcomes for the 45 county level UAs. Descriptive statistics for 12 indices of urban form and 3 measures of traffic congestion are displayed in Table 4 2 and Table 4 3. Statistics description and outcome validation are provided in the following section. In addition, urban form results in this study are compared with those calculated by Cutsinger et al. (2005). Urban Form Outcomes For the study population of 45 UAs, the mean population size is 156,832, which means that on average, nearly 157,000 employed residents live in county level UAs across the state. There is substantial variation in population size. For example, Bradford County has the minimum value of 3,335 employed residents living in the UA across the study population, whereas the Miami Dade UA has the maximum value of 956,389 for this index. This index is validated by comparing it to the fi gures generated by U.S. Census Bureau LED OnTheMap Version 4 (av ailable at http://lehdmap4.did.census.gov/themap4/ ). For exam ple, the Census reports that for the Miami Dade UA in 2007 there are 958,153 employed residents, which is fairly close to the figure computed in this study.

PAGE 61

61 The mean density of housing units per square mile of developable land is 617, whereas the comparable figure for jobs is 627. This pair of figures reflects that, on average, the typical urban household has more than one worker. However, the difference betwe en housing unit density and job density is relatively low compared to the nationwide figures 698 for housing unit density and 782 for job density computed by Cutsinger et al. (2005). This reflects a relatively high proportion of retired people living i n Florida. For example, in the Citrus County UA the housing unit density is 404, whereas job density is only 177, which is the minimum value of this index. The maximum value, in the Miami UA, is 2269. The Bradford County UA has the minimum value of housing unit density of 245, whereas the maximum value of this index, 1605, occurs in the Pinellas UA. The mean value for housing unit concentration on developable land is 0.252, which indicates that about 25% of housing units would need to be redistributed in or der to produce an even distribution across areal unit of developable land in each UA. The comparable index for jobs is 0.45. A mean value of job concentration higher than housing unit concentration reflects the fact that housing units are more evenly distr ibuted than jobs across the UAs. Mean values of these two indices in this study are considerably lower than those of Cutsinger et al. (2005) 0.486 for housing unit concentration and 0.626 for job concentration because in their sample, only large metropo litan EUAs are included. This comparison indicates that larger size areas tend to be more concentrated for both housing units and jobs than smaller size areas, which can be explained using the theory of agglomeration economies. The Highlands UA

PAGE 62

62 has the min imum value of housing concentration on developable land (0.11), whereas the maximum value of this index, 0.48, occurs in Bay UA. Minimum and maximum values for job concentration on developable land are 0.22 (Taylor UA) and 0.64 (Putnam UA) respectively. Fo r the mixed use indices of jobs to housing and vice versa, the mean values are 1677.131 and 2380.470 respectively. These figures imply that on average about 1677 jobs are located in the same areal unit as a typical housing unit and that about 2380 housing units are located in the same areal unit as a typical job. The minimum and maximum jobs to housing mixed use values are 460.17 (Citrus UA) and 5417.18 (Miami Dade UA), respectively. Comparable figures of housing to jobs mixed use values are 699.08 (Taylor UA) and 5628.01 (Broward UA), respectively. The substantial variations of mixed use indices in this study are consistent with those findings by Cutsinger et al. (2005), expanding in urban areas of all sizes (Cutsinger et al. 2005, p.244). For the 45 county level UAs the mean value for continuity is 0.658, which indicates that, on average, about 66% of areal units of land within each UA have half or more of the land available for d evelopment already developed. This figure is more reasonable comparing to that of Cutsinger et al. (0.346) as they over estimated the area of developable land by using the NLCDB dataset (2005). However, the figure in this study tends to be overestimated sin ce some undevelopable land and developable, yet not developed, land are considered to be developed. For instance, the municipal, county, state, and federal owned land

PAGE 63

63 uses (other than military, forests, parks, recreational areas, colleges, and hospitals) a re classified into the developed land type, when in fact they consist of developable yet not developed land (e.g., land on holdings) and undevelopable land (e.g., water bodies owned by governments). Although over estimated, this figure is still believed t o be meaningful because it varies significantly across UAs and correlates significantly with other variables, as shown later in this part. The Highlands UA has the minimum value for the continuity index (0.32), whereas the Seminole UA and the Monroe UA hav e the maximum value for this index (0.92). The mean housing centrality and job centrality values are 1.245 and 1.911 respectively. These statistics show the average degree of centralization of ich is considered le vel UAs. Both values are centrality), because: 1. this study focuses on UAs, whereas Cutsinger et al. con sider EUA; and 2. the dominance of counties in Florida coastal areas that typically have more centralized developed patterns. After b eing standardized by governmental building, the housing centrality value indicates that the average areal unit in a UA is about 25% farther from the county center than the average housing unit; and the job centrality value indicates that the average areal unit in a UA is about 91% farther from the coun ty center than the average job. Higher values of job centrality than housing centrality in most UAs reflect the fact that

PAGE 64

64 for housing centrality and job centrality occur in the Sumter UA (0.83 and 0.82, respectively). And both maximum values for housing centrality and job centrality occur in the Desoto UA (1.99 and 6.91, respectively). The Desoto UA is substantially higher in these two indices because, in this county, UAs are sep arated into several areas and because a relatively high portion of housing building. The housing proximity mean is 1.326, indicating that the average pair of areal units across the s tudy population is about 1.3 times further apart than the average pair of houses. The job proximity mean of 1.540 indicates that the average pair of areal units is about 1.5 times further apart than the average pair of jobs. These figures confirm the findi ngs of concentration indices. First, generally speaking, jobs are more clustered than housing units across UAs, as indicated by the fact that the job proximity mean value is higher than the housing proximity mean value. Second, larger areas tend to be more clustered for both jobs and housing units than smaller areas, as indicated by the fact that the mean values in this study are lower than those of Cutsinger et al. (2005). As indicated in this section, there might be strong correlation among these 12 urban form indices. In this regard, following a description of congestion outcomes a section intends to summarize these 12 indices and try to group them into fewer independent factors. Congestion Outcomes The mean RCI value across the 45 county level UAs is 1.2 93, which means that on average the congestion level on both freeways and major arterials in

PAGE 65

65 2007 is about 1.3 times more than the defined threshold level of congestion. Due to different datasets being used, the mean value of RCI in this study is lower com pared to that of Blanco et al. (2010) (1.330). Since the dataset used in this study is more complete than that used by Blanco et al., it can be inferred that reduce the mean val ue of RCI. There are five UAs (Desoto UA, Hendry UA, Sumter UA, Taylor UA, and Indian River) with RCI values less than 1.0, which indicates that they do not suffer area wide congestion in 2007. The minimum RCI value, 0.56, occurs in the Taylor UA, whereas the maximum RCI value, 1.83, occurs in the Clay UA. The mean value for Delay per Capita is 33.116, meaning on average each employed resident spends approximately 33 extra hours commuting during peak it is found that substantial variations exist across the study population. The Taylor UA has the minimum value of this congestion index (0.2), whereas the Walton UA has the maximum value (131.7). Delay per capita in Walton UA is high mainly due to the fac t that it has the second lowest number of employed residents across the study population. The mean TTI value is 1.311, meaning on average the estimated travel time during peak periods is about 1.3 times more than the travel time during free flow conditions This value is very close to the one computed by Blanco et al. (for the same year, the mean value of TTI is 1.321). In this study, the minimum and

PAGE 66

66 maximum TTI values occur in the Taylor UA (1.00) and the Broward UA (1.67), respectively. Correlat ions among Indices Correlat ion s among Urban Form Indices A Bivariate Pearson correlation analysis was conducted between 12 indices of urban form using SPSS (see Table 4 4). The correlation matrix reveals several noteworthy patterns as detailed below. Since most of the urban form dimensions in this study are concep tualized and measured following the same approach developed by Cutsinger et al. (2005), comparisons of findings of urban form correlations in these two studies are also provided in this section. First and foremost, it is no surprise that all the dual indic es (housing unit and job) developed for the same conceptual dimensions of urban form (density, concentration, mixed use, centrality, and proximity) are highly positively correlated with one another. This pattern indicates that housing and employment land u ses share common characteristics in different dimensions of urban form. This contradicts the finding in Cutsinger et al. (2005), as they found no correlation between concentration indices and between centrality indices. This difference could result from th e variation of size in study area between the two studies. Because Cutsinger et al. considered only large metropolitan areas, high concentration and centrality of housing units does not necessarily follow high concentration and centrality of jobs in large metropolitan areas where transportation facilities are usually well developed. However, this is not the case in smaller urban areas.

PAGE 67

67 Second, all indices of four urban form dimensions size, density, continuity, and mixed use, are highly positively correla ted with one another. This pattern is consistent with the finding in Cutsinger et al. (2005), who explained that high population density results in intensified development of land and high competition for scarce land between different land uses. The correl ation between population size and density can be explained using economic theory that people t end to live together for econom ies of agglomeration Third, all indices of centrality and proximity are highly positively correlated. This pattern could simply be explained by the fact that highly centralized urban areas are often associated with close proximity between housing units and between jobs. A similar pattern was observed by Cutsinger et al. (2005). However, the indices are less significantly correlated i n their study than in this one. Again, variation of size between two studies can explain the difference. Last but not the least, neither of the concentration indices is correlated with any of the other indices. This pattern suggests that concentration in t his study is a measure of independent dimension of urban form. This pattern, however, contradicts the results in Cutsinger et al. (2005) as they found out that both concentration indices are significantly negatively correlated with mixed use, density and c ontinuity dimensions, and are significantly positively correlated with centrality, concentration, and proximity dimensions. A possible explanation of the difference could be that larger areal unit sizes are used in this study, which could lead to lower acc uracy in measuring concentration indices, especially in smaller urban areas. Another possibility could be the leapfrog pattern of urban areas in

PAGE 68

68 this study. Because urban area at the county level is employed as the basic statistical unit there are cases w hen two or more urban areas are in the same county or a single urban area is included in multiple counties Two urban form dimensions centrality and proximity which measure the distances between land uses can be influenced by this issue. This methodolo gical issue may not generate problems among those indices because these measures essentially reflect the location of urban areas. However, the leapfrog pattern of urban areas would affect the correlation results between these two dimensions and concentrati on. In summary, results of correlation analysis in this study further support the idea of grouping those 12 indices of urban form. This finding is consistent with Cutsinger et al. (2005). Correlat ion s among Congestion Indices As shown in Table 4 5, all thr ee congestion indices are significantly positively correlated. The highest magnitude of the correlation appears in the pair of roadway congestion index and travel time index. High correlation between these indices is not surprising because they are all cal culated using a series of steps within one study. Essentially, they are strongly associated with roadway VMTs. Correlations between delay per capita and the other two indices are lower compared to the correlation between RCI and TTI because additional data number of employed residents are used for calculating delay per capita. Even though congestion indices are highly correlated, this study will not group them into a single factor. The reasons for this are twofold. First, there are only three congestion indices and they are conceptually distinct from one another. Second,

PAGE 69

69 as shown later in the section describing correlation between urban form and congestion, the levels of correlation significance between an urban form factor and congestion indices are dif ferent, implying that these three congestion indices differ in association with some of the urban form factors. Thus, leaving three congestion indices will more clearly address the relationship between urban form and congestion than grouping them into a si ngle factor. Factor Analysis of Urban Form Indices uses a principle components analysis of 12 indices using varimax rotation in SPSS. Five factors were developed. The results of four criteria 19 used to determine the number of factors are: Three out of five of the factors have eigenvalues greater than one. The two factors are retained because the solution conforms more closely to other criteria. Scree plot shows that all five factors are in the sharply de scending portion. These five factors cumulatively explain 91% of the variation in the original 12 indices, which exceed the 70% threshold. A comparison of reproduced correlations to the original observed correlations (the smaller the percentage of residuals greater than 0.05, the better to be selected as a solution) shows that in the five factor solution there are only 10 residuals greater than 0.05, whereas these numbers in three and four factor solutions are 15 and 13, respectively. After the num ber of factors was determined, these factors were labeled based upon urban form indices that are most closely related to each factor (in rotated component matrix, indices with index loading greater than 0.6 are 19 See page 247 248 of Cutsinger et al. (2005) for detailed explanation of the four criteria used to determine the number of factors.

PAGE 70

70 selected): 1) size/density/mixed use, 2) cent rality/job proximity, 3) concentration, 4) housing proximity, and 5) continuity. The index loadings and indices selected for each component are presented in Table 4 6. Size/density/mixed use factor: includes population size, both of the density indices and both of the mixed use indices; after rotation, this component accounts for 37% of the total variance in the original urban form indices. Centrality/job proximity factor: includes both of the centrality indices and job proximity index; accounts for 22% of the total variance. Concentration factor: includes both of the concentration indices; accounts for 14% of the total variance. Housing proximity factor: only includes the housing proximity index; accounts for 10% of the total variance. Continuity factor: on ly includes the continuity index; accounts for 8% of the total variance. There are several interesting patterns revealed in the way in which the 12 indices are grouped in the five factors. First, the fact that the job proximity index loads onto the same fa ctor as both centrality indices indicates that urban areas with highly centralized patterns evince high job proximity, but not necessarily housing proximity. Second, except that the proximity measures for both housing and jobs load onto separate factors, t he housing and job measures for density, mixed use, centrality, and concentration all load onto the same factors. This pattern indicates that housing units and jobs are distributed similarly throughout urban areas in the study population. Although this fin ding is confirmed by high bivariate correlations among housing and job measures for each urban form dimension, it contradicts the finding of Cutsinger et al. (2005), implying that in metropolitan areas, the spatial distribution of housing units and jobs ar e different from that in smaller urban areas.

PAGE 71

71 This factor analysis proves that though all dimensions of urban form are distinct in theory, some are not in practice. In addition, the difference in the way in which urban form indices load onto different fact ors of these two studies demonstrates that characteristics of urban form patterns differ in the selection of study population. Two findings together confirm the conclusion of Cutsinger et al. (2005), suggesting that a comprehensive understanding of multipl e dimensions of urban form and their interrelationships is a must before exploring the causes or consequences. Ranking Areas by Measures Ranking on Urban Form Factors In order to compare urban form factors and conduct correlation analysis between urban form and congestion, five factors of each of the 45 county level UAs are calculated using z scores of all 12 indices and their corresponding index loadings. Table C 6 pre sents the composite z scores of each urban form factor. Based on composite z scores, Table 4 7 ranks the top five and bottom five UAs for each of the five urban form factors. Higher values (ranks) in all factors indicate a lesser degree of sprawl like deve lopment. The patterns in Table 4 7 confirm two major findings of Cutsinger et al. (2005). First, some UAs always rank low across several factors. For example, the Taylor UA ranks last in concentration factor and third last in the size/density/mixed use fac tor, reflecting a small urban area characterized by scattered locations of both housing units and jobs. A similar pattern is shown in high ranks as well. For example, the Desoto UA ranks first in both centrality/job proximity and housing proximity factors. The Broward UA ranks second in the

PAGE 72

72 size/density/mixed use factor and fourth in the continuity factor. This pattern indicates that some UAs share commonalities across several sprawl dimensions even though those dimensions are conceptually and practically d ifferent. The second and more significant pattern is that quite a few UAs rank high in some factors but low in others. For example, the Broward UA ranks second in size/density/mixed use factor and fourth in continuity factor but forty third in concentratio n factor, providing a combined urban form image of a built up urban area characterized by centerlessness, high density and mix of housing units and employment. The Bay UA ranks 44th in housing proximity and 41st in continuity factor yet ranks first in con centration factor. The fact that urban areas in Bay County are divided by a significant number of water bodies can explain scattered distribution of residential land use and the low value in continuity factor. On the other hand, both housing units and jobs in the Bay UA are clustered near water bodies, resulting in high value of concentration factor and indicating compact urban development pattern. Five other UAs, including Taylor UA, Bradford UA, Seminole UA, Pinellas UA, and Desoto UA, also reflect the s ame pattern, which supports the idea that sprawl is a relative term that depends upon which urban form dimension and which kind of land use is addressed. The r anking of some factors can be explained through spatial analysis. Figure 4 1 presents maps of adj usted z scores 20 for each of the five factors. Darker color in each map indicates lower value of the factor and therefore, more sprawl like development Two urban form factors concentration and continuity, 20 A djusted z scores are used in figure 4 1 to ensure the minimum value of each factor is zero.

PAGE 73

73 reveal no visible p attern of spatial distributio n ( s ee Figure 4 1 C and E). Notably the size/density/mixed use factor (see Figure 4 1 A) appears to follow the spatial distribution of county population (see Figure 3 2). The larger the county population (darker color in the map), the higher the z score value of the size/density/mixed use factor (lighter color in the map), which indicates not surprisingly that this factor is strongly related to county population. A clearer spatial pattern is observed in the map of centrality/job proximity factor (see Figure 4 1 B) M ost county UAs with low z scores such as Hernando County UA, Monroe County UA, Volusia County UA, Walton County UA, and Okaloosa County UA, concentrate in coastal areas This can be explained by the fact that most development in coas tal areas concentrate along the coast, rather than concentrating around the county center. One exception of this pattern is Hendry County which has a low z score but is located inland. T his is not surprising because this county has two UAs located a signi ficant distance away from each other (see Figure 3 1). As a result the geographical separation of UAs in many counties also results in low value for continuity factor (see Figure 4 1 E), such as Highlands County UA, Charlotte County UA, Bay County UA, and Pasco County UA. T his indicates that the centrality and proximity dimensions of urban form in the model are sensitive to the number and spatial location of UAs in each county. Ranking on Congestion Indices Table 4 8 ranks the top five and bottom five UAs for each of the three congestion indices. Higher ranks of all indices in this table are associated with more congestion. One pattern, as shown in this table, is that some UAs appear to

PAGE 74

74 be consistent performe rs across all three congestion indices in the ranking. For instance, the Taylor UA ranks last in all three indices. Both the Desoto UA and the Hendry UA rank in the bottom four for all three indices. A similar pattern is witnessed in high ranks. For exampl e, the Clay UA and the Okaloosa UA rank in the top five in all three indices. This pattern is consistent with the finding in the research conducted by Sarzynski et al. (2006), shown in Table 3 of their study. This pattern further proves that, although thre e indices differ conceptually, they statistically correlate with each other. Correlation Analysis between Congestion and Urban Form A Bivariate correlation analysis was conducted between each urban form factor and each congestion measure using SPSS softwar e. As explained in the introduction chapter, it is believed that sprawl like urban form patterns can be positively associated with traffic congestion if more car travel is generated. Compact development, on the other hand, can also result in more VMTs and higher levels of localized congestion. Since urban form factors in this study are standardized with higher values indicating less sprawl, negative correlations are expected between urban form factors and congestion measures if the first explanation dominat es the system; positive correlations are expected if the second explanation holds true. The correlation results, however, reveal divergent relationships between urban form factors and congestion measures (see Table 4 9). To begin with, the size/density/mix ed use factor is positively correlated to all three outcomes. The strength of correlation coefficient is medium between size/density/mixed use factor and RCI, and is large between size/density/mixed use factor and TTI, both

PAGE 75

75 at statistically highly signific ant levels. This pattern suggests that UAs with larger population size, denser, and more mixed land uses tend to have higher roadway congestion indices and higher travel time indices. To this end, it is clear that the localized congestion generated by larg e numbers of people taking trips within a confined area leads to increased levels of area wide congestion measures. The lack of significant correlation with delay per capita suggests that the effect of this urban form factor on congestion outcomes is towar ds the system at large, not so much to the individual traveler. Neither the centrality/job proximity factor nor the concentration factor is significantly correlated with any of the congestion measures. This finding is not surprising because essentially, th ese two factors reflect how residential and working land uses are clustered. The lack of correlations indicates that most trips do not happen within the same type of land use. In cases where people hold two or more jobs simultaneously, there is a possibili ty that they will travel from one workplace to an other. T h e reality is that people seldom travel from one workplace to another, especially in rural counties. Thus, the degree to which the same type of land use is clustered does not have much to do with tra vel behaviors or the consequent levels of congestions. In addition, the lack of correlation between centrality and congestion measures suggests that travel behaviors are not associated with the location of the county center. It can be the fact that most tr ips do not take place in the county center; or, it can be speculated that some UAs might be centerless, or they might have multiple centers, which implies that to

PAGE 76

76 define the county government building as the UA center does not reflect the real situation. T he housing proximity factor is negatively correlated with all three congestion measures, all at statistically significant levels. This pattern indicates that UAs with dispersal of housing units tend to have higher levels of area wide congestion and longer individual delay. Interestingly, this factor is the only one To interpret this pattern, it can be speculated that UAs with compact development of residential land use would have a concentration of services A puzzling result of the correlation analysis is that the continuity factor is significantly positively correlated with all three congestion measures. The correlation is moderately significant to RCI and delay per capita, and is highly significant to TTI. What appears to be happening is that UAs with more continuous land development an indication of less sprawl tend to have higher levels of congestion. This pattern does not make sense at all if only considering that leapfrog development should generate longer trips However, when considering the fact that the UAs in this study vary in size, it can be speculated that the degree to which land is developed should be lower in smaller UAs than in larger UAs. Therefore, in this study more continuous land devel opment indicates larger size UAs where levels of congestion are higher. This pattern also confirms the finding that the continuity dimension of urban form is significantly positively correlated with population size, density, and mixed use dimensions.

PAGE 77

77 One l ast noteworthy correlation pattern is that, as expected, almost all three congestion measures have close correlation results with each of the urban form factors. This pattern can be explained by the fact that all three congestion measures are significantly correlated. However, one exception is that different magnitude and significance of correlation exist between delay per capita and the other two congestion measures to size/density/mixed use factor. This exception supported the decision to leave congestion measures separate when conducting correlation analysis with urban form factors. A comparison of the correlation results in this study to the ones in Sarzynski form and traf fic congestion. Both studies find that density and continuity dimensions of urban form are significantly positively correlated to congestion measures. Besides, this study uncovers the significant correlation of mixed use and housing proximity dimensions of urban form to congestion measures, while Sarzynski et al. found no statistically significant correlation. A simple comparison of the correlation results between these two studies may not be sufficient to draw any conclusions without considering the method ological differences, such as size of areal unit, data, study population, scale, and dimensions of urban form. However, at least to this point it can be speculated that congestion measures are not likely to respond to fixed characteristics of urban form; r ather, some dimensions of urban form may have higher impact on travel patterns in one study population than in another.

PAGE 78

78 Summary In summary, the results and findings of this study are diverse and promising. First of all, the descriptive analysis of areal un it effect reveals that some urban form indices are more influenced by different locations of areal units than others. Also, urban form indices tend to be influenced by the combined effects of areal unit location and variation of UA area size. These two pat terns confirm the necessity of conducting scenario analysis to minimize the impact of areal unit effect. Second, by using the mean values of the four scenario outcomes, interpretation of each urban form index was made, suggesting that all indices are mean ingful and reasonable to describe the conceptually distinct urban form characteristics. Descriptive analysis was also made on the congestion side. Both analyses indicate that strong correlation exist amongst urban form and congestion outcomes. This specul ation was proven in the third step when correlation analyses were conducted for urban form indices and congestion measures. In terms of urban form, it is found that all the indices developed for the same urban form dimension are significantly correlated. F urthermore, population size, density, continuity and mixed use dimensions of land use are highly correlated; so are centrality and proximity dimensions. Only the concentration dimension appeared to be not correlated with any of the other dimensions. In ter ms of congestion measures, all three indices are highly correlated. Following the correlation analyses, a factor analysis for urban form indices was conducted in the fourth step trying to group them into smaller number of

PAGE 79

79 factors. It turned out that five f actors were generated, demonstrating that conceptually distinct urban form indices are practically reducible. In addition, a ranking analysis of urban form factors yields interesting results. It is found that some UAs tend to be consistent in the ranking a cross several factors. More importantly, there are quite a few cases where UAs rank high in some factors but low in others, indicating that sprawl is a relative term that depends on which urban form dimension and which kind of land use is address ed. Further spatial an alyse s of urban form factors reveal that the size/density/mixed use factor is spatially related to county population; counties with low values of centrality/job proximity and housing proximity factors concentrate in coastal areas and/or have geographically separated UAs. Finally, the correlation analysis between urban form and congestion reveals that divergent relationships exist between the variables. Among all the five factors, three show statistically significant correlation with conge stion measures. Only the housing proximity factor is negatively correlated with all three phenomenon. The following chapter summarizes the discussion s that are incorporated in this chapter. In addition, limitations of this study and recommendations for future research are also presented in the following chapter.

PAGE 80

80 Table 4 1. Ratios of the maximum to the minimum values calculated in four scenarios County name Housing unit concen tration ratio Job concentration ratio Job/ housing mixed use ratio Housing/job mixed use ratio Continuity ratio Housing unit centrality ratio Job centrality ratio Housing unit proximity ratio Job proximity ratio Alachua 1.13 1.15 1.20 1.30 1.16 1.07 1.07 1.07 1.07 Bay 1.02 1.02 1.11 1.06 1.06 1.01 1.01 1.01 1.01 Bradford 1.50 1.20 2.26 2.46 1.88 1.18 1.50 1.03 1.03 Brevard 1.12 1.07 1.08 1.26 1.04 1.02 1.02 1.03 1.03 Broward 1.26 1.11 1.03 1.07 1.32 1.04 1.04 1.04 1.04 Charlotte 1.20 1.09 1.20 1.30 1.31 1.06 1.06 1.02 1.02 Citrus 1.08 1.13 1.21 1.37 1.14 1.05 1.05 1.03 1.03 Clay 1.40 1.08 1.07 1.45 1.14 1.06 1.21 1.04 1.04 Collier 1.35 1.08 1.17 1.23 1.11 1.04 1.04 1.06 1.06 Desoto 1.63 1.41 2.92 3.05 1.29 1.20 1.20 1.16 1.16 Duval 1.06 1.04 1.05 1.02 1.07 1.02 1.02 1.02 1.02 Escambia 1.39 1.05 1.79 2.86 1.05 1.03 1.03 1.04 1.04 Flagler 1.63 1.13 1.56 1.40 1.22 1.11 1.11 1.13 1.13 Hendry 4.19 1.23 5.08 3.51 1.13 1.14 1.14 1.01 1.01 Hernando 1.12 1.08 1.13 1.27 1.39 1.05 1.05 1.06 1.06 Highlands 1.35 1.46 1.15 1.39 1.52 1.11 1.11 1.03 1.03 Hillsborough 1.02 1.02 1.12 1.18 1.05 1.05 1.42 1.02 1.05 Indian River 1.42 1.09 1.25 1.34 1.05 1.03 1.03 1.03 1.03 Lake 1.27 1.29 1.10 1.29 1.16 1.04 1.04 1.02 1.01 Lee 1.06 1.07 1.02 1.34 1.13 1.08 1.08 1.07 1.07 Leon 1.05 1.17 1.13 1.07 1.06 1.02 1.02 1.02 1.02 Manatee 1.07 1.03 1.16 1.18 1.03 1.03 1.03 1.03 1.03 Marion 1.15 1.04 1.07 1.10 1.15 1.03 1.03 1.02 1.02 Martin 2.40 1.28 1.15 1.10 1.16 1.11 1.11 1.08 1.08 Miami Dade 1.61 1.13 1.12 1.03 1.15 1.04 1.04 1.05 1.05 Monroe 1.06 1.04 1.29 1.73 1.09 1.18 1.08 1.04 1.04 Nassau 1.26 1.17 2.03 1.62 1.29 1.04 1.04 1.02 1.02 Okaloosa 1.31 1.05 1.73 1.88 1.09 1.08 1.08 1.04 1.04 Okeechobee 1.91 1.68 1.17 1.57 1.33 1.08 1.08 1.05 1.05 Orange 1.09 1.11 1.06 1.18 1.08 1.04 1.04 1.04 1.04 Osceola 1.18 1.09 1.18 1.23 1.06 1.08 1.08 1.06 1.06 Palm Beach 1.02 1.03 1.05 1.06 1.07 1.03 1.03 1.02 1.02 Pasco 1.03 1.03 1.09 1.03 1.16 1.04 1.04 1.03 1.03 Pinellas 1.25 1.02 1.04 1.10 1.06 1.06 1.06 1.27 1.27 Polk 1.04 1.03 1.10 1.26 1.11 1.02 1.02 1.02 1.02 Putnam 1.28 1.21 1.63 1.50 1.18 1.12 1.12 1.06 1.06 Santa Rosa 1.15 1.16 1.32 1.39 1.23 1.06 1.06 1.02 1.02

PAGE 81

81 Table 4 1. Continued County name Housing unit concentration ratio Job concentration ratio Job/ housing mixed use ratio Housing/job mixed use ratio Continuity ratio Housing unit centrality ratio Job centrality ratio Housing unit proximity ratio Job proximity ratio Sarasota 1.12 1.06 1.11 1.15 1.15 1.08 1.08 1.01 1.01 Seminole 1.06 1.13 1.07 1.07 1.06 1.03 1.03 1.04 1.04 St. Johns 1.09 1.15 1.18 1.44 1.07 1.03 1.03 1.01 1.01 St. Lucie 1.05 1.07 1.05 1.32 1.17 1.06 1.06 1.04 1.04 Sumter 1.11 1.20 1.33 1.18 1.18 1.04 1.04 1.11 1.11 Taylor 1.75 1.84 1.50 1.34 1.17 1.10 1.10 1.05 1.05 Volusia 1.06 1.07 1.03 1.12 1.10 1.07 1.00 1.01 1.01 Walton 1.42 1.10 1.74 1.77 1.05 1.01 1.01 1.05 1.05 Average 1.33 1.15 1.37 1.43 1.17 1.06 1.08 1.05 1.05 Note: High value of the ratio indicates strong influence of areal unit location on urban form outcomes ; the highest value of each index is shown with Table 4 2. Descriptive statistics of urban form indices Name N Minimum Maximum Mean Standard deviation Population size 45 3335 956389 156832.556 212666.78458 Housing unit density 45 244.79 1604.80 617.337 287.01274 Job density 45 176.87 2268.62 627.988 454.90807 Housing unit concentration 45 .11 .48 0.252 .07688 Job concentration 45 .22 .64 0.454 .08708 Mix of job/housing 45 460.17 5417.18 1677.131 1209.77895 Mix of housing/job 45 699.08 5628.01 2380.470 1425.70447 Continuity 45 .32 .92 0.658 .15411 Housing unit centrality 45 .83 1.99 1.245 .21827 Job centrality 45 .82 6.91 1.911 1.19759 Housing unit proximity 45 .95 2.10 1.326 .26018 Job proximity 45 .94 3.33 1.540 .44392 Mean value of four scenarios for each county level UA. Table 4 3. Descriptive statistics of traffic congestion indices Name N Minimum Maximum Mean Standard deviation RCI 45 .56 1.83 1.29 3 .26102 Delay per c apita 45 .20 131.70 33.11 6 21.90520 TTI 45 1.00 1. 67 1. 311 14325

PAGE 82

82 Table 4 4. Bivariate correlation between urban form indices Housing unit density Job density Housing unit concentr ation Job concentr ation Ratio of job/housi ng Ratio of housing/j ob Continuit y Housing unit centrality Job centrality Housing unit proximity Job prox i mity Settlement size .751 ** .859 ** .029 .156 .866 ** .775 ** .363 .059 .187 .103 .196 Housing unit density .882 ** .120 .311 .812 ** .739 ** .476 ** .074 .267 .147 .299 Job density .096 .240 .937 ** .857 ** .482 ** .167 .066 .028 .064 Housing unit concentration .557 ** .121 .170 .118 .064 .124 .109 .040 Job concentration .103 .061 .206 .025 .209 .075 .245 Ratio of job/housing .964 ** .442 ** .177 .135 .007 .095 Ratio of housing/job .403 ** .181 .119 .035 .076 Continuity .079 .273 .135 .263 Housing unit centrality .685 ** .555 ** .721 ** Job centrality .295 .822 ** Housing unit proximity .702 ** Note: Correlation is significant at the 0.05 level (2 tailed) ; ** Correlation is significant at the 0.01 level (2 tailed)

PAGE 83

83 Table 4 5. Bivariate correlation between congestion indices Delay per Capita TTI RCI Pearson Correlation .702 ** .927 ** Sig. (2 tailed) .000 .000 N 45 45 Delay per Capita Pearson Correlation .644 ** Sig. (2 tailed) .000 N 45 **. Correlation is significant at the 0.01 level (2 tailed). Table 4 6. Rotated component matrix describing five urban form factors Component Urban form index Size/density /mixed use factor Centrality/ job proximity factor Concen tration factor Housing proximity factor Continuity factor Population size .914 .085 .036 .031 .008 Housing unit density .865 .188 .217 .047 .147 Job density .953 .066 .150 .008 .163 Housing unit concentration .063 .167 .883 .263 .087 Job concentration .133 .230 .869 .275 .031 Mix of job/housing .976 .010 .068 .055 .132 Mix of housing/job .926 .042 .196 .009 .139 Continuity .343 .149 .106 .028 .917 Housing unit centrality .151 .836 .045 .301 .003 Job centrality .123 .941 .005 .109 .117 Housing unit proximity .050 .418 .005 .881 .026 Job proximity .118 .878 .124 .352 .076 Note: Extraction method: Principal component analysis ; r otation method: Varimax with Kaiser Normalization ; i ndices that have high index loadings are with Table 4 7. UA rankings on urban form factors Ranking Size/density/ mixed use factor Centrality/job proximity factor Concen tration factor Housing proximity factor Continuity factor 1 Miami Dade Desoto Bay Desoto Seminole 2 Broward Bradford Leon Nassau Monroe 3 Pinellas Putnam Escambia Flagler Walton 4 Hillsborough Leon St. Lucie Escambia Broward 5 Alachua Martin Sumter Taylor Okaloosa

PAGE 84

84 41 Okeechobee Santa Rosa Pinellas Santa Rosa Bay 42 Santa Rosa Seminole Lake Pasco Desoto 43 Taylor Hendry Broward Hendry Charlotte 44 Citrus Hernando Highlands Bay Osceola 45 Bradford Sumter Taylor Monroe Highlands Note: High er ranks indicate less er sprawl Table 4 8. UA rankings on traffic congestion Ranking RCI Delay per capita TTI 1 Clay Walton Broward 2 Okaloosa Monroe Miami 3 Walton Broward Clay 4 Pasco Okaloosa Okaloosa 5 Pinellas Clay Hillsborough 41 Sumter Indian R iver Sumter 42 Hendry Desoto Flagler 43 Indian R iver Hendry Hendry 44 Desoto Nassau Desoto 45 Taylor Taylor Taylor Note: High ranks indicate high level of congestion. Table 4 9. Pearson correlation coefficients: urban form factors and congestion indices Urban form factor RCI Delay per capita TTI Size/density/mixed use .435 ** .183 .590 ** Centrality/job proximity .280 .247 .277 Concentration .143 .116 .041 Housing proximity .434 ** .397 ** .392 ** Continuity .361 .339 .416 ** Note: Correlation is significant at the 0.05 level (2 tailed); ** Correlation is significant at the 0.01 level (2 tailed).

PAGE 85

85 Figure 4 1. Maps of urban form factors. A) Size/density/mixed use. B) Centrality/job proximity. C) Concentration. D) Housing proximity. E) Continuity.

PAGE 86

86 Figure 4 1. Continued

PAGE 87

87 Figure 4 1. Continued

PAGE 88

88 CHAPTER 5 DISCUSSIONS Th is chapter will present discussions of how the research conducted in this study addresses the question of what is the relationship between urban form and traffic congestion in Florida. Recall that the objectives of this study were to: 1. Adapt the multi dimensional app roach to measure urban form at sub metropolitan level; 2. Make a valuable contribution towards the understanding of urban sprawl in the context of Florida; and 3. Provide further insight of the impacts the built environment poses on traffic congestion. Although this study focuses on the relationship between urban form and traffic congestion in Florida, the same methodology could be used in other areas at both sub metropolitan and metropolitan levels. In addition, the same methodology applies to addressing the rel ationship between urban form and its other consequences or causes, such as VMTs, greenhouse gas emission, and land values. Discussion of Findings and Results As described in the introduction of this paper, although an increasing number of studies have been focusing on investigating the relationship between urban form and traffic congestion for the past two decades, the relation is found to be inconclusive. The divergence of findings in existing literature results from a combination of several factors, inclu ding methodology, data, and study population differences. In this regard this study intends to incorporate several methodologies developed by a few recent studies to rethink this issue. Specifically, this study expands and adapts the method in measuring ur ban form developed by Cutsinger et al. (2005) to county level UAs in Florida. Also, the disaggregate method in measuring congestion newly developed by Blanco et al. (2010)

PAGE 89

89 is applied to this study. In terms of linking urban form and congestion, this study adapts the statistical method developed by Sarzynski et al. ( 2006). Some findings of this study are found to be consistent with those in previous studies, and can be summarized as follows: The mean values of housing indices of urban form dimensions are alw ays lower than job indices values, indicating jobs are located in a more compact pattern than housing units. Some of the conceptually distinct urban form dimensions are highly correlated with one another. Specifically, three dimensions of urban form dens ity, continuity, and mixed use, are found to be highly correlated in both studies. By using factor analysis, urban form dimensions can be group into smaller number of factors due to high inter correlations. Some UAs always rank high/low in congestion ranking across all three congestion measures. Other UAs, however, rank high in sprawl ranking in some urban form factors while low in others, indicating sprawl is a relative term. Divergent correlations exist between urban form factors and congestion measu res. Both studies find that density and continuity dimensions of urban form are significantly positively correlated to congestion measures. On the other hand, there are findings in this study that differ from their counterparts, which are summarized as fol lows: Mean values of three urban form dimensions density, concentration, and proximity, are lower than those in Cutsinger et al. Mean value of RCI is lower than that in Blanco et al. The concentration dimension of urban form that is not correlated to any of other indices in this study is correlated to most dimensions in Cutsinger et al. The grouping of urban form indices is different between this study and Cutsinger et Th is study yields stronger relationship between urban form and traffic congestion use and housing proximity dimensions of urban form are significantly correlated to congestion.

PAGE 90

90 As explained in the results chapter of this paper, reasons of these differences are diverse. Each difference can be generated by a single factor, or combined effect s of several factors. Those reasons are summarized below: This study uses local data to measure urban form whi ch are more accurate and up to date comparing to the national data used in previous studies. In addition, more complete data is used in calculating congestion measures. The distinctive demographic characteristics in Florida may contribute to the difference For example, higher portion of retired people living in the State results in a lower average density than the national level. In addition, the dominance of counties on coastal areas results in a higher average centrality than the national level. Previous studies selected study areas with relatively similar land size s whereas this study includes UAs of all sizes. This may cause a problem when a uniformly sized areal unit is applied to all UAs, especially to those with small area. Although scenario analysi s is used in this study to minimize the areal unit problem, it cannot be avoided when comparing the results with other studies. The fact that urban areas are sometimes separated in one county and are cut by the county boundary may affect correlation result s amongst urban form dimensions. Also, it may affect the correlation s between urban form factors and congestion measures if considerable number of residents living or working in those urban areas that are cut by county boundaries travel more to the adjacen t county than urban areas in the same county. However, this is not the case in other studies as their study areas are not affected by political boundaries. Different data sources are used in classifying land types. Although the data used in this study are developed land tends to be over estimated because of the way land uses are grouped. Cutsinger et al., on the other hand, over estimate area of developable land, resulting in continuity values in two studies incomparable. Limitations of this Study In this study, there are several limitations in measuring urban form, which, if were solved, would have yielded more accurate results in correlation analyses. Although Florida Parcel Data by County used in th is study is the most accurate and up to date data set available, it still tends to be not ideal. A test of Gainesville UA, which is familiar by the researcher, concludes that the number of housing units in some parcels is

PAGE 91

91 misreported. For example, for the parcel ID 07176014000, it is reported that there are in total 1068 housing units, which is considerably higher than the real number 360 seem not to update number of h ousing units in their data sets. For example, of more than 51,000 parcels designated as residential land use in Indian River County, over 40,000 are recorded to have zero number of residential units. Similar cases are found in Sumter County and Walton Coun ty. As described in the methodology part of this paper, this issue was adjusted by manually populating values into parcels with zero housing units. However, this process over estimated the real situation since non occupied residential units were included. In addition, one important fact is that mobile homes are not included because the number of mobile homes is not available in the parcel data set. Yet, not surprisingly, considerable number of people live s in mobile homes in Florida and they should contribu te a significant amount of daily travel. The degree to which this data set would affect the result is hard to assess. To clean this data manually is not practical. Given that this is the most accurate data set available, and given that most of the issues d escribed above are common across the State, it is believed that they are not detrimental in conducting the overall research analysis for this project. In addition, the 2.5*2.5 mile cell was selected to minimize the Modifiable Area Unit Problem. Although th is size is validated by Arafat et al. (2010), it might cause problem when calculating dimensions of urban form in small UAs. For example, the total area for the Desoto UA is about 13.5 square miles, which only allows two intact cells to be included within the UA boundary. This can obscure the accuracy of measuring urban form results. What is more, the fact that cells are cut by UA boundary can also generate

PAGE 92

92 inaccurate results. One good example is the Monroe UA which has a stripe shape. Indeed, for all 55 ce lls that are included in the Base Scenario, only one is intact. In this case, the variation of cell size and shape could considerably reduce the accuracy of urban form measures. One possible solution could be to use a smaller size cell. However, a careful analysis needs to be done in advance to balance counteracting effects. In this respect, this model generates more accurate results for larger county UAs than smaller county UAs. T he relationship between urban form and congestion might be under estimated du e to inconsistent measures of the variables. O n the congestion side a lthough i t is asserted that only peak period travels are considered, it is in essence strongly related to the average daily travel since by definition peak period travel is calculated th rough dividing daily travel in half. Therefore, non home work travels, which consist of a significant portion of daily travel in both peak and off peak periods are included while measuring congestion. On the urban form side, however, only residential and job related land uses are considered, meaning home work travel is the primary concern. In this regard measures of urban form do not account for the entire human activities and tra vel patterns that contribute to the congestion. To eliminate this problem, the possible solutions, in theory, could be either excluding non home work travel when measuring congestion or including other land uses when measuring urban form. However, both sol utions would require major modification in methodology and are not practical in this study due to time constraints. T he issue of inconsistency between urban form and traffic congestion is also reflected by the fact that congestion measures only include fre eways and major arterial

PAGE 93

93 streets. T he exclusion of minor streets in the calculation process may not fully capture the travel consequence that result s from urban form especially urban form characterized by compact development and a high level of clustering For example, during peak periods congestion is very likely to happen in minor streets within o r near working district. I f not incorporated into the calculation, excluding this factor would result in underestimation of the link between urban form and con gestion. Also it is likely to underestimate congestion and the value of a more connected network that allows vehicles to take alternatives to the arterials when there is congestion. Moreover, the interaction of housing and jobs is not well explored in this study. Of all seven urban form dimensions, only mixed use dimension considers both housing and jobs in one index. However, common knowledge tells us that many trips happen between different land uses. Sarzynski et al., for instance, measured the proximity of housing to jobs in their study and found this index is highly correlated with commute times. As explained in the methodology part of this paper, this index is excluded in this study because of time constraint. If the same index was employed in this stu dy, high correlation would have been expected between it and congestion measures. Last but not the least, as described in the introduction and literature review part of this paper, the relationship between urban form and traffic congestion can be intervened by many other factors, such as socio economic conditions, public interventions, and changes of these factors in time. These intervening variables can be controlled using regression analysis and multivariate time lag model (Sarzynski et al., 2006 ). T he causal relationship between urban form and congestion might have been more apparent if these factors were taken into account in the model design. T his study, however, is

PAGE 94

94 cross ltaneity 21 explained by Sarzynski et al. (p.605 606). Opportunities of Future Research The research on topics of understanding urban form, especially sprawl like urban development, is diverse. Current trend has been towards seeking the consequences of sprawl, among which traffic congestion draws increasing attentions. Using multi dimensional approach to quantify urban form, this study examines UAs of more than two third of the counties in Florida and provides a big picture to interpret the general char acteristics of built environment in Florida. Findings of this study on urban form suggest the necessity to conduct more detailed research on individual cases if clearer and more in depth understanding of sprawl is needed. Case study on one or several count ies would be a useful tool to examine characteristics of urban form that cannot be studied at the metropolitan level, such as street connectivity and design elements. Also, appropriate study area needs to be defined which includes sprawl like development b eyond the UA boundary and is suitable across areas of all sizes. This, if achieved, could yield a stronger relation between urban form and congestion. Another way to obtain a clearer relationship between urban form and congestion would be to modify the mod el on the congestion side. It would be desirable if congestion measures only included peak periods on weekdays, and trips were classified into different trip purposes, such as home work, home shopping, etc. The portion of each trip purpose could then be us ed as a weighting factor for different land uses when 21 Simultaneity bias refers to the different responses in time for the change of transport network and the change of travel demand in response to traffic congestion. Most transport projects take 10 15 years to complete; yet during this time, congestion can ch ange significantly due to change in travel demand (Sarzynski et al., 2006).

PAGE 95

95 exploring urban form and congestion correlations. In addition, congestion measures could be separated by travel mode, such as automobiles and public transport; each mode might respond differently to dim ensions of urban form. M oreover, since many studies examine the urban form travel behavior and travel behavior tr affic consequence relations, the study model can be refined by incorporating travel behavior as a bridging factor between urban form and traffic congestion. This factor, if considered, would not only better interpret how urban form and congestion are related, but also provide evidence about the causal relationship among these t hree variables. In addition, s ince some aspects of urban form reveal a strong statistical correlation with congestion, it would be meaningful to look at the relation under the spatial context. The fact that both sprawl and congestion are regional issues indicates that spatial analysis, if conducted, could reveal interesting patterns between adjacent UAs. In fact, because in many cases UAs are cut by county boundary, activities that hap pen in one UA might be more associated with adjacent UAs than those within the same county boundary, which also supports the idea of conducting spatial analysis. In this regard, GIS spatial analysis to measure urban form and congestion, and their relations is suggested for future studies. The ultimate purpose of studying the impacts of urban form on congestion, arguably, is to relate the findings and conclusions with policy. Many studies have addressed the impacts of lan d use policies on changes of urban form and the related transport consequences. For example, one recent study found that state growth management programs effectively reduced urban sprawl in terms of population density

PAGE 96

96 and land use mixture ( Sun & Yin, 2007 ) However, few studies ha s addressed the reversed relationship; that is, using the findings of the impacts of urban form on travel patterns to identify, evaluate, and eventually guide policies and practices that tend to induce changes on built environment and human activities. To achieve this requires researchers think out of the box, meaning other variables that influence the urban system need to be investigated and their relative power to explain urban conditions need to be weighted.

PAGE 97

97 CHAPTER 6 CONCLUSIONS This study has explored the relationship between urban form and traffic congestion by using data from 45 county level UAs as of 2007 in Florida, measuring 12 indices of urban form and 3 measures of traffic congestion, grouping urban form indices into 5 di stinct factors, and conducting bivariate analyses between variables. The correlation results reveal that divergent relationships exist between urban form factors and congestion measures. Size/density/mixed use factor is significantly positively correlated to TTI and RCI; continuity factor is significantly positively correlated to all three congestion measures; housing proximity factor is significantly negatively correlated to all three congestion measures; and neither centrality/job proximity factor nor con centration factor has statistically significant correlation with any of the congestion measures. Only the relationship between housing proximity factor and congestion supports the idea of Comparisons of the results t o national averages suggest that all urban form indices are meaningful and reasonable to describe each conceptually distinct urban condition. Further correlation and factor analyses prove some urban form indices are correlated and they can be grouped into smaller number of factors. The fact that the grouping of urban form factor is different from other studies suggest that this process should be done in every like study using multi dimensional approach to measure urban form due to the unique characteristics of study population. Ranking of urban form factors depicts general sprawl patterns for UAs in each county. It is found that UAs of all sizes in the State are experiencing urban sprawl in different aspects. Some UAs tend to be consistent in the ranking acr oss several factors.

PAGE 98

98 Other UAs, however, rank high in some factors but low in others, indicating that sprawl is a relative term that depends on which urban form dimension and which kind of land use is addressed. Further spatial analyses of urban form facto rs reveal that the size/density/mixed use factor is spatially related to county population. Also, counties with low values of centrality/job proximity and housing proximity factors concentrate in coastal areas and/or have geographically separated UAs, indi cating that the centrality and proximity dimensions of urban form in the model are sensitive to the number and spatial location of UAs in each county. Methodologically, contributions of this study to the field include: adapting the prevailing integrated tr ansportation land use model to sub metropolitan level; developing scenario analysis to calculate urban form in order to minimize the Modifiable Areal Unit Problem; calculating values for each urban form factor using standardized z score of related urban fo rm indices; and using more accurate and up to date data sources for both urban form and congestion calculations. In addition, this research adopted the newly developed disaggregate method to calculate congestion measures, which generate more accurate outco mes compare to the aggregate method developed by Texas Transportation Institute. As many practitioners and researchers perceive, traffic congestion has become increasingly severe and harder to mitigate. In many cases, conventional attempts about road construction and widening in order to catch the worsening of congestion prove to be expensive, unfeasible, and even shortsighted. While other solutions, such as economical approach and travel demand management, may offer some relief, the fact that travel is a derived demand should encourage planners and policy makers to

PAGE 99

99 consider directing urban form patterns as an alternative approach to handling traffic congestion. For example, this study implies that increasing the proximity of housing units may shorten the delay time for travelers and reduce congestion level. Such effort the transport network to deal with travel demand. In addition, this study indicates that developi ng compact urban form by increasing population density and mixed use of housing units and jobs may induce area attempt to mitigate congestion by controlling urban sprawl needs to be carefully planned and evaluat ed since different aspects of sprawl like development can have totally opposite effects on travel outcomes.

PAGE 100

100 APPENDIX A CLASSIFICATION OF LA ND TYPES Table A 1. Land use codes to category of land types lookup table Land use code Description Land type 1 Single Family Developed 2 Mobile Homes Developed 3 Multi family 10 units or more Developed 4 Condominia Developed 5 Cooveratives Developed 6 Retirement Homes Developed 7 Miscellaneous Residential (migrant camps, boarding homes, etc.) Developed 8 Multi family less than 10 units Developed 11 Stores, one story Developed 12 Mixed Use store and office or store and residential or residential combination Developed 13 Department Stores Developed 14 Supermarkets Developed 15 Regional Shopping Centers Developed 16 Community Shopping Centers Developed 17 Office buildings, non professional service buildings, one story Developed 18 Office buildings, non professional service buildings, multi story Developed 19 Professional service buildings Developed 20 Airports (private or commercial), bus terminals, marine terminals, piers, marinas. Developed 21 Restaurants, cafeterias Developed 22 Drive in Restaurants Developed 23 Financial institutions (banks, saving and loan companies, mortgage companies, credit services Developed 24 Insurance company offices Developed 25 Repair service shops (excluding automotive), radio and T.V.repair, refrigeration service, electric repair,laundries, laundromats Developed 26 Service stations Developed 27 Auto sales, auto repair and storage, auto service shops, body and fender shops, commercial garages, farm and machinery sales fender shops, commercial garages, farm and machinery sales and services, auto rental, marine equipment, trailers and related equipm ent, mobile home sales motorcycles, construction and vehicle sales. Developed

PAGE 101

101 Table A 1. Continued Land use code Description Land type 28 Parking lots (commercial or patron) mobile home parks Developed 29 Wholesale outlets, produce houses, manufacturing outlets Developed 30 Florist, greenhouses Developed 31 Drive in theaters, open stadiums Developed 32 Enclosed theaters, enclosed auditoriums Developed 33 Nightclubs, cocktail lounges, bars Developed 34 Bowling alleys, skating rinks, pool halls, enclosed arenas Developed 35 Tourist attractions, permanent exhibits, other entertainment facilities, fairgrounds (privately owned). Developed 36 Camps Developed 37 Race tracks; horse, auto or dog Developed 39 Motels, Hotels Developed 41 Light manufacturing, small equipment manufacturing plants, small machine shops, instrument manufacturing printing plants Developed 42 Heavy industrial, heavy equipment manufacturing, large machine shops, foundries, steel fabricating plants, auto or aircraft plants Developed 43 Lumber yards, sawmills, planing mills Developed 44 Packing plants, fruit and vegetable packing plants, meat packing plants Developed 46 Other food processing, candy factories, bakeries, potato chip factories Developed 47 Mineral processing, phosphate processing, cement plants, refineries, clay plants, rock and gravel plants. Developed 48 Warehousing, distribution terminals, trucking terminals, van and storage warehousing Developed 49 Open storage, new and used building supplies, junk yards, auto wrecking, fuel storage, equipment and material storage Developed 71 Churches Developed 72 Private schools and colleges Developed 73 Privately owned hospitals Developed 74 Homes for the aged Developed 75 Orphanages, other non profit or charitable services Developed 76 Mortuaries, cemeteries, crematoriums Developed 77 Clubs, lodges, union halls Developed 78 Sanitariums, convalescent and rest homes Developed 79 Cultural organizations, facilities Developed 81 Military Developed

PAGE 102

102 Table A 1. Continued Land use code Description Land type 83 Public county schools include all property of Board of Public Instruction Developed 84 Colleges Developed 85 Hospitals Developed 86 Counties (other than public schools, colleges, hospitals) including non municipal government. Developed 87 State, other than military, forests, parks, recreational areas, colleges, hospitals Developed 88 Federal, other than military, forests, parks, recreational areas, hospitals, colleges Developed 89 Municipal, other than parks, recreational areas, colleges, hospitals Developed 0 Vacant Residential Developable 9 Undefined Resewed for Use by Department of Revenue Developable 10 Vacant Commercial Developable 38 Golf courses, driving ranges Developable 40 Vacant Industrial Developable 45 Canneries, fruit and vegetable, bottlers and brewers distilleries, wineries Developable 50 Improved agricultural Developable 51 Cropland soil capability Class I Developable 52 Cropland soil capability Class I1 Developable 53 Cropland soil capability Class 111 Developable 54 Timberland site index 90 and above Developable 55 Timberland site index 80 to 89 Developable 56 Timberland site index 70 to 79 Developable 57 Timberland site index 60 to 69 Developable 58 Timberland site index 50 to 59 Developable 59 Timberland not classified by site index to Pines Developable 60 Grazing land soil capability Class I Developable 61 Grazing land soil capability Class I1 Developable 62 Grazing land soil capability Class I11 Developable 63 Grazing land soil capability Class IV Developable 64 Grazing land soil capability Class V Developable 65 Grazing land soil capability Class VI Developable 66 Orchard Groves, Citrus, etc. Developable 67 Poultry, bees, tropical fish, rabbits, etc. Developable 68 Dairies, feed lots Developable 69 Ornamentals, miscellaneous agricultural Developable 80 Undefmed Reserved for future use Developable

PAGE 103

103 Table A 1. Continued Land use code Description Land type 82 Forests, parks, recreational areas Developable 90 Leasehold interests (government owned property leased by a non governmental lessee) Developable 91 Utility, gas and electricity, telephone and telegraph, locally assessed railroads, water and sewer service, pipelines, canals, radioltelevision communication Developable 92 Mining lands, petroleum lands, or gas lands Developable 93 Subsurface rights Deve lopable 94 Right of way, streets, roads, irrigation channel, ditch, etc. Developable 97 Outdoor recreational or parkland, or high water recharge subject to classified use assessment Developable 98 Centrally assessed Developable 99 Acreage not zoned agricultural Developable 95 Rivers and lakes, submerged lands Undevelopable 96 Sewage disposal, solid waste, borrow pits, drainage reservoirs, waste land, marsh, sand dunes, swamps Undevelopable Note: Land use codes and description are available from FGDL site: http://www.fgdl.org/metadataexplorer/full_metadata.jsp ?docId=%7B6FFD8FE0 E960 4199 828F B5816841ACB0%7D&loggedIn=false

PAGE 104

104 APPENDIX B PROCESS OF QUANTIFYI NG URBAN FORM This section presents an example of step by step procedures on how to calculate each dimension of urban form for the Base Scenario of a chosen county using ArcGIS and Microsoft Excel Software. Urban form of all 45 counties that experienced traffic congestion on freeways and major arterials in year 2007 was measured following the same process. Also, other three scenarios were conduc t ed foll owing the same steps and using the same GIS layers as Base Scenario except for the unique locations of areal unit for each scenario. Base Scenario Steps 1 6 Data Management 1. Created a new Personal Geodatabase in ArcCatalog and imported datasets described below. STATE_CNTBND (polygon layer of county boundaries in Florida from the Census site: http://www2.census.gov/cgi bin/shapefiles2009/stat e files?state=12 ) STATE_ URBAN (polygon layer of urban boundaries in Florida in April 2007, from FGDL site: http://www.fgdl.org/metadataexplorer/explorer.jsp ) STATE_FISHNET_BASE (statewide p olygon layer of areal unit system for Base Scenario) STATE_JOB (statewide job information at census block level from the Census OnTheMap database: http://lehd.did.census.gov/led/onthemap/fl/wa c/ Filename: fl_wac_total_ja_2007_1.csv.gz) STATE_EMPRES (statewide employed resident information at census block level from the Census OnTheMap database: http://lehd.did.census.gov/led/onthe map/fl/rac/ Filename: fl_rac_total_ja_2007_1.csv.gz) [CNTNAME]_LANDTYPE (polygon layer of land type classification, where [CNTNAME] = The name of a chosen county.) [CNTNAME]_CENSUSBLK (polygon layer of census block for a chosen county from the Census site : http://www.census.gov/cgi bin/geo/shapefiles/state files?state=12 ) [CNTNAME]_PARCEL (parcel data of a chosen county from FGDL) 2. by intersecting STATE_CNTBND with [CNTNAME]_URBAN.

PAGE 105

105 3. Clipped STATE_FISHNET_BASE with [CNTNAME]_URBAN and named [CNTNAME]_CELL. 4. Created point layer of housing unit. 1) Selected residential land uses in [CNTNAME]_PARCEL by using the following query: "DORUC" = '001' OR "DORUC" = '002' OR "DORUC" = '003' OR "DORUC" = '004' OR "DORUC" = '006' OR "DORUC" = '008' Land use codes that belong to this category are: 001 Single Family 002 Mobile Home 003 Multi Family that is greater than or equal to 10 housing units 004 Condominium 006 Retirement Homes 008 Multi Family that is less than 10 housing units 2) Populated number of housing units for those parcels with zero number of housing units. For each land use type: 001, 002, 004, and 006 1 dwelling unit 003 10 dwelling units 008 5 dwelling units 3) Selected housing units built through 2007 by using the following query: "EFFYRBLT" <= 2007 OR "ACTYRBLT" <= 2007 4) Exported selected parcels and n amed [CNTNAME]_HSUNITS_PARCEL. 5) Created point layer for residential units by using Feature to Point tool and named [CNTNAME]_HSUNITS. 5. Created polygon layer of job location. 1) Joined STATE_JOB to [CNTNAME]_CENSUSBLK (kept only matching records) and exported to a new layer named [CNTNAME]_JOB. 2) Calculated job density by adding a field (set type as Double) and using following equation in Field Calculator: DENSITY = [total] / [Shape_Area] 6. Created polygon layer of employed resident location. 1) Joined STATE_EMPRES to [CNTNAME]_CENSUSBLK (kept only matching records) and exported to a new layer named [CNTNAME]_EMPRES.

PAGE 106

106 2) Calculated job density by adding a field (Double) and using following equation in Field Calculator: DENSITY = [total] / [Shape_Area] Step 7 Population size calculations 7. In ArcMap, calculated number of employed residents in the UA. 1) Intersected [CNTNAME]_CELL with [CNTNAME]_EMPRES and named the new polygon layer as CELL_EMPRES_INTERSECT. 2) Calculated number of employed residents in the new layer by adding a new field (Long Integer) and using following equation in Field Calculator: TOTAL_NEW = [DENSITY] [Shape_Area] 3) Output the sum of TOTAL_NEW in Excel spreadsheet. Step 8 11 Density calculations 8. In ArcMap, calculated number of housing units in each cell. 1) Interse cted [CNTNAME]_CELL with [CNTNAME]_HSUNITS and named the new point layer as CELL_HSUNITS_INTERSECT. 2) Used Summary Statistics tool to calculate number of housing units: Input CELL_HSUNITS_INTERSECT Statistics Field(s) NORESUNTS Statistic Type SUM Case Field FID_[CNTNAME]_CELL 9. Calculated number of jobs in each cell. 1) Intersected [CNTNAME]_CELL with [CNTNAME]_JOB and named the new polygon layer as CELL_JOB_INTERSECT. 2) Calculated number of employed residents in the new layer by adding a new field (Long Int eger) and using following equation in Field Calculator: TOTAL_NEW = [DENSITY] [Shape_Area] 3) Used Summary Statistics tool to calculate number of residential units: Input CELL_ JOB _INTERSECT Statistics Field(s) TOTAL_NEW Statistic Type SUM Case Field FID_[CNTNAME]_CELL

PAGE 107

107 10. In Excel spreadsheet, calculated total area of developed and developable land in UA using area of UA to subtract undevelopable land area 22 11. Housing Unit (Job) Density was calculated through dividing total number of housing units (jobs) [result from Step 8.2 (Step 9.3)] by total area of developed and developable land (result from Step 10). Step 12 Concentration calculations 12. Housing Unit Concentration and Job Concentration were calculated through: 1) In Excel spreadsheet, inputted UA area, a rea of each cell, number of housing units in each cell (results from Step 8.2) and number of jobs in each cell (results from Step 9.3). 2) Housing unit (job) density per cell was calculated through dividing number of housing units (jobs) [results from Step 8. 2 (Step 9.3)] by area of each cell. 3) For those cells that have housing unit (job) density per cell (results from Step 12.2) greater than Housing Unit (Job) Density (results from Step 11), calculated the excess number of housing units (jobs) in each cell using Eq. B 1 Excess number of Housing Units (Jobs) in Each Cell = Number of Housing Units (Jobs) in Each Cell [results from Step 8.2 (Step 9.3)] [Area of Each Cell Housing Unit (Job) Density (results from Step 11)] ( Eq. B 1 ) 4) Housing Unit (Job) Concentration was solved using Eq. B 2 Housing Unit (Job) Concentration = Sum of Excess number of Housing Units (Jobs) in Each Cell (results from Step 12.3) / Total Number of Housing Units (Jobs) [result from Step 8.2 (Step 9.3)] ( Eq. B 2 ) Step 13 Mixed use calculations 13. Mix of Housing to Job and Mix o f Job to Housing were calculated through: 1) In Excel spreadsheet, inputted number of housing units in each cell (results from Step 8.2) and number of jobs in each cell (results from Step 9.3). 2) For each cell, calculated Housing Unit (Job) Exposure Index using Eq. B 3 Housing Unit (Job) Exposure Index in Each Cell= {Number of Housing Units (Jobs) in Each Cell [results from Step 8.2 (Step 9.3)] / Total Number of Housing Units (Jobs) [result from St ep 8.2 (Step 9.3)]} {Number of Jobs (Housing Units) 22 Originally, the sum of developed and developable land area was used as dividend to calculate density. However, it was found that the parcel layer does not contain roads, which sh ould be included as part of developed land. Without considering area of road, total area of developed and developable land is under estimated, and therefore the density is overestimated. This problem was solved using area of UA to subtract undevelopable la nd area

PAGE 108

108 in Each Cell [results from Step 9.3 (Step 8.2)] / the sum of Total Number of Housing Units and Jobs (results from Step 8.2 and Step 9.3)} ( Eq. B 3 ) 3) Mix of Housing to Job (Job to Housing) was solved usin g Eq. B 4 Mix of Housing to Job (Job to in each cell (results from Step 13.2) Number of Housing Units (Jobs) in Each Cell [results from Step 8.2 (Step 9.3)]} ( Eq. B 4 ) Step 14 15 Continuity calculations 14. In ArcMap, calculated areas of three land types for each cell: 1) In ArcMap, Intersected [CNTNAME]_CELL with [CNTNAME]_LANDTYPE and named the new polygon layer as CELL_LANDTYPE_INTERSECT. 2) Used Summary Statistics tool to calculate areas of three land types for each cell: Input CELL_LANDTYPE_INTERSECT Statistics Field(s) shape_area Statistic Type SUM Case Field FID_[CNTNAME]_CELL and LANDTYPE_CODE 15. In Excel spreadsheet, Continuity was calculated through: 1) For each cell, calculated the ratio of developed lan d area 23 to developable land area. 2) Counted the number of ratios that are higher than 0.5, then divided by total number of cells. Step 16 21 Centrality calculations 16. In ArcMap, created a point layer for the location of county government building: 1) Selected the parcel in [CNTNAME]_PARCEL where the county government building is located, exported it to a new polygon layer. 2) Used Feature to Point tool to convert the polygon layer into a point layer named [CNTNAME]_GOV_POINT. 17. Calculated distance between centroid of e ach cell and the location of county government building: 1) Converted [CNTNAME]_CELL into a point layer named [CNTNAME]_CELL_POINT, using Feature to Point tool. 2) Tools, generated a distance table using Distances Between Points (Between Layers) tool in Analysis Tools: 23 Developed land area includes results from Step 14.2 and area of infrastructure that is not included in the parcel layer.

PAGE 109

109 Source Point Layer [CNTNAME]_GOV_POINT Unique Source ID Field objectedID Target Point Layer [CNTNAME]_CELL_POINT Unique Target ID Field objec ted 18. Aggregated housing unit at parcel level to census block level: 1) Intersected [CNTNAME]_CENSUSBLK with CELL_HSUNITS_INTERSECT and named the new point layer as HSUNITS_CENSUSBLK_INTERSECT. 2) Used Summary Statistics tool: Input HSUNITS_CENSUSBLK_INTERSECT S tatistics Field(s) NORESUNTS Statistic Type SUM Case Field FID_[CNTNAME]_CENSUSBLK 3) Joined newly created table with [CNTNAME]_CENSUSBLK, then exported to a new polygon layer named [CNTNAME]_HSUNITS_CENSUSBLK. 4) Converted [CNTNAME]_HSUNITS_CENSUSBLK into a point layer named [CNTNAME]_HSUNITS_CENSUSBLK _POINT, using Feature to Point tool. 19. Calculated distances between locations of housing units and the location of county government building: 1) en Points (Between Layers) tool in Analysis Tools: Source Point Layer [CNTNAME]_GOV_POINT Unique Source ID Field objectedID Target Point Layer [CNTNAME]_HSUNITS_CENSUSBLK _POINT Unique Target ID Field SUM_NORESUNITS 20. Calculated distance between locations of jobs and the location of county government building: 1) Converted CELL_JOB_INTERSECT into a point layer named [CNTNAME]_JOB_POINT, using Feature to Point tool. 2) (Between L ayers) tool in Analysis Tools: Source Point Layer [CNTNAME]_GOV_POINT Unique Source ID Field objectedID Target Point Layer [CNTNAME]_JOB_POINT Unique Target ID Field TOTAL_NEW 21. In Excel spreadsheet, Housing Unit Centrality and Job Centrality were ca lculated through: 1) Inputted results from Step 18, 19 and 20. 2) Calculated Average Distance between Cell and Government Building using Eq. B 5

PAGE 110

110 Average Distance between Cell and Government Building = sum of distances between cells and government building (resul ts from Step 18) / total number of cells ( Eq. B 5 ) 3) Calculated Weighted Average Distance between Housing Unit (Job) and Government Building using Eq. B 6 Weighted Average Distance between Housing Unit (Job) and Government ts (jobs) in each census block [results from Step 19 (Step 20)] distance between housing unit (job) and government each census block [results from Step 19 (Step 20)]} ( Eq. B 6 ) 4) Housing Unit (Job) Centrality was solved using Eq. B 7 Housing Unit (Job) Centrality = Average Distance between Cell and Government Building (result from Step 21.2) / Weighted Average Distance between Housing Unit (Job) and Government Building (result from Step 21.3) ( Eq. B 7 ) Step 22 24 Proximity calculations 22. In ArcMap, calculated distances between cells: 1) (Within Layer) tool in Analysis Tools: Source Point Layer [CNTNAME]_CE LL_POINT Unique ID Field objectedID Output 23. Calculated average distance between housing units (jobs) in one census block and those in the rest of census blocks: 1) In Ha (Within Layer) tool in Analysis Tools: Source Point Layer [CNTNAME]_HSUNITS_CENSUSBLK _POINT ([CNTNAME]_JOB_POINT) Unique ID Field TOTAL_NEW Output tics only (minimum, maximum, mean, 24. In Excel spreadsheet, Housing Unit Proximity and Job Proximity were calculated through: 1) Inputted results from Step 22 and 23. 2) Calculated Average Distance between Cells using Eq. B 8

PAGE 111

111 Average Distance between Cells = sum of distances between cells (results from Step 12) / total number of cells ( Eq. B 8 ) 3) Calculated Weighted Average Distance between Housing Units (Jobs) using Eq. B 9 ho using units (jobs) in each census block (results from Step 23) average distance between housing units (jobs) in one census block and those in the rest each census block (resu lts from Step 23)] ( Eq. B 9 ) 4) Housing Unit (Job) Proximity was solved using Eq. B 10 Housing Unit (Job) Proximity = Average Distance between Cells (result from Step 24.2) / Weighted Average Distance between Housing Units (Jobs) (result from Step 24.3) ( Eq. B 10 )

PAGE 112

112 APPENDIX C URBAN FORM OUTCOMES BY SCENARIO AND MEAN VALUES Table C 1 Urban form outcomes Base Scenario County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Alachua 79,918 948.83 1252.69 0.24 0.40 3743.10 5908.36 0.71 1.40 1.88 1.55 1.71 Bay 69,322 414.90 367.70 0.48 0.58 1575.13 2413.30 0.40 1.03 2.23 1.00 1.72 Bradford 3,335 244.79 364.09 0.16 0.49 353.60 522.11 0.40 1.64 6.73 1.32 2.43 Brevard 213,531 656.98 570.55 0.24 0.49 1304.94 1897.85 0.72 1.14 1.38 1.14 1.29 Broward 796,783 1388.31 1910.34 0.15 0.28 4804.45 5753.16 0.93 1.19 1.44 1.18 1.25 Charlotte 50,294 792.15 402.46 0.19 0.57 1217.33 2387.12 0.40 1.12 1.65 1.19 1.41 Citrus 21,608 403.83 176.87 0.16 0.39 410.74 723.88 0.60 1.11 1.22 1.10 1.12 Clay 54,396 409.46 390.22 0.23 0.51 988.12 1555.83 0.63 1.09 1.13 1.23 1.32 Collier 108,761 531.30 690.18 0.21 0.44 1584.29 2325.48 0.75 1.19 1.13 1.22 1.39 Desoto 5,536 479.11 596.29 0.25 0.44 1578.48 2223.37 0.43 1.82 6.34 2.08 3.30 Duval 428,926 627.52 880.45 0.37 0.49 2969.54 3896.22 0.76 1.38 1.64 1.47 1.56 Escambia 125,354 560.37 733.19 0.44 0.56 1190.10 1289.99 0.58 1.49 2.07 1.76 2.07 Flagler 19,783 499.68 181.65 0.46 0.37 519.69 696.29 0.54 1.27 1.37 1.82 1.77 Hendry 9,284 521.58 570.96 0.50 0.30 200.88 379.62 0.62 0.99 1.15 1.07 1.09 Hernando 42,234 555.05 286.90 0.24 0.32 921.00 1009.10 0.62 0.90 1.34 1.48 1.15 Highlands 24,612 482.09 352.80 0.10 0.38 774.15 1368.78 0.39 1.35 1.87 1.16 1.30 Hillsborough 531,022 829.14 1059.39 0.33 0.52 3344.25 4559.86 0.80 1.37 2.00 1.44 1.57 Indian R iver 47,658 478.24 416.54 0.19 0.43 906.97 1426.46 0.68 1.28 2.06 1.15 1.40

PAGE 113

113 Table C 1. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Lake 81,537 618.28 447.27 0.13 0.37 1055.46 1621.99 0.80 1.13 1.51 1.14 1.28 Lee 206,137 551.63 500.69 0.24 0.46 1517.84 2882.37 0.79 1.30 1.72 1.23 1.46 Leon 114,843 481.77 776.76 0.37 0.56 3442.12 4919.91 0.68 1.49 2.68 1.34 1.90 Manatee 124,812 653.61 757.46 0.25 0.50 2504.52 4187.99 0.76 1.43 2.17 1.47 1.72 Marion 79,836 360.99 339.55 0.22 0.56 711.57 1600.28 0.49 1.11 2.11 1.13 1.56 Martin 52,310 451.11 479.14 0.45 0.57 1232.66 1754.97 0.60 1.56 2.60 1.47 1.63 Miami Dade 956,389 1410.62 2268.62 0.29 0.40 5185.11 5414.92 0.82 1.29 1.71 1.29 1.48 Monroe 30,560 869.83 609.82 0.25 0.46 1162.29 1640.25 0.89 1.14 1.60 0.97 0.96 Nassau 14,449 367.40 368.91 0.30 0.54 1059.08 1497.76 0.65 1.32 1.34 2.06 2.42 Okaloosa 73,233 627.09 681.57 0.45 0.44 1309.24 1525.00 0.80 1.27 1.45 1.17 1.28 Okeechobee 8,650 338.75 343.84 0.14 0.47 638.70 1073.48 0.67 1.36 2.07 1.51 1.83 Orange 492,969 838.65 1217.15 0.33 0.51 3381.31 4278.25 0.70 1.52 1.68 1.51 1.56 Osceola 108,483 412.51 319.13 0.29 0.52 983.71 1606.66 0.40 1.20 1.71 1.25 1.43 Palm B each 525,997 818.77 934.95 0.28 0.46 2927.48 3795.05 0.75 1.30 1.38 1.28 1.28 Pasco 151,539 647.16 298.27 0.28 0.38 1558.48 1868.68 0.47 1.43 1.44 1.10 1.11 Pinellas 425,630 1604.80 1787.52 0.15 0.33 3690.73 4163.92 0.84 1.17 1.28 1.27 1.37 Polk 214,746 450.11 437.12 0.29 0.52 1366.86 2376.83 0.68 1.07 1.43 1.19 1.37 Putnam 14,148 376.88 439.90 0.23 0.57 776.40 1189.34 0.63 1.31 4.83 1.13 2.43 Santa R osa 52,202 299.93 187.84 0.29 0.46 636.08 916.05 0.52 0.97 1.24 1.09 1.10 Sarasota 134,311 706.09 675.64 0.19 0.43 1687.65 2610.73 0.77 1.23 2.30 1.18 1.43 Seminole 193,539 875.91 986.38 0.19 0.34 2802.03 3398.78 0.96 0.92 1.06 1.22 1.27

PAGE 114

114 Table C 1. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity St Johns 60,319 559.49 461.27 0.25 0.40 1310.62 1819.16 0.78 1.24 1.61 1.38 1.50 St Lucie 96,384 416.64 274.60 0.38 0.49 1207.07 2306.47 0.47 0.99 1.56 1.32 1.37 Sumter 5,950 605.70 180.88 0.33 0.55 668.06 1067.61 0.50 0.83 0.83 1.58 1.41 Taylor 3,348 287.73 374.15 0.14 0.23 627.49 598.38 0.83 1.47 1.94 1.66 1.79 Volusia 197,849 615.42 538.75 0.25 0.49 1464.09 2403.89 0.74 1.08 1.06 1.10 1.17 Walton 4,938 709.97 369.01 0.23 0.44 629.49 868.82 0.93 0.96 0.96 1.26 1.33 Identical among scenarios. Table C 2 Urban form outcomes West Scenario County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Alachua 79,918 948.83 1252.69 0.24 0.42 3552.15 4779.06 0.77 1.48 2.01 1.59 1.76 Bay 69,322 414.90 367.70 0.48 0.58 1430.78 2276.85 0.40 1.03 2.23 1.01 1.74 Bradford 3,335 244.79 364.09 0.20 0.57 534.75 948.54 0.54 1.72 7.06 1.33 2.45 Brevard 213,531 656.98 570.55 0.26 0.46 1324.39 1751.40 0.72 1.14 1.39 1.14 1.29 Broward 796,783 1388.31 1910.34 0.18 0.30 4738.77 5788.07 0.89 1.22 1.49 1.20 1.26 Charlotte 50,294 792.15 402.46 0.21 0.56 1018.00 2399.54 0.36 1.16 1.71 1.20 1.42 Citrus 21,608 403.83 176.87 0.16 0.43 472.59 822.89 0.58 1.06 1.16 1.12 1.13

PAGE 115

115 Table C 2. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Clay 54,396 409.46 390.22 0.25 0.52 1043.92 1876.09 0.69 1.15 1.36 1.23 1.31 Collier 108,761 531.30 690.18 0.17 0.47 1381.93 1968.87 0.77 1.19 1.13 1.20 1.36 Desoto 5,536 479.11 596.29 0.19 0.37 881.82 1118.83 0.42 1.81 6.29 1.93 3.06 Duval 428,926 627.52 880.45 0.37 0.49 2985.68 3946.89 0.75 1.38 1.65 1.48 1.57 Escambia 125,354 560.37 733.19 0.33 0.58 2087.34 3685.82 0.55 1.47 2.04 1.70 2.00 Flagler 19,783 499.68 181.65 0.28 0.37 748.20 950.96 0.52 1.41 1.52 1.93 1.88 Hendry 9,284 521.58 570.96 0.16 0.34 615.97 990.16 0.55 0.96 1.10 1.08 1.10 Hernando 42,234 555.05 286.90 0.26 0.30 924.81 1132.97 0.50 0.90 1.34 1.40 1.09 Highlands 24,612 482.09 352.80 0.13 0.35 755.67 1095.24 0.26 1.27 1.76 1.17 1.31 Hillsborough 531,022 829.14 1059.39 0.33 0.53 3580.68 5340.80 0.80 1.42 1.42 1.44 1.62 Indian R iver 47,658 478.24 416.54 0.20 0.43 933.42 1436.87 0.69 1.25 2.02 1.16 1.41 Lake 81,537 618.28 447.27 0.15 0.41 1039.42 1758.32 0.69 1.15 1.54 1.14 1.28 Lee 206,137 551.63 500.69 0.23 0.43 1506.59 2489.24 0.72 1.35 1.78 1.31 1.55 Leon 114,843 481.77 776.76 0.36 0.62 3357.10 5152.65 0.69 1.50 2.68 1.33 1.88 Manatee 124,812 653.61 757.46 0.23 0.49 2151.29 3603.03 0.78 1.44 2.19 1.45 1.70 Marion 79,836 360.99 339.55 0.24 0.54 745.56 1760.81 0.43 1.10 2.09 1.13 1.55 Martin 52,310 451.11 479.14 0.21 0.45 1156.92 1598.33 0.64 1.66 2.75 1.57 1.74 Miami Dade 956,389 1410.62 2268.62 0.18 0.35 5758.26 5585.77 0.71 1.26 1.67 1.24 1.42 Monroe 30,560 869.83 609.82 0.26 0.47 1085.14 1944.95 0.91 1.15 1.62 0.95 0.95 Nassau 14,449 367.40 368.91 0.30 0.53 598.18 1040.80 0.75 1.38 1.40 2.11 2.48 Okaloosa 73,233 627.09 681.57 0.34 0.45 2054.36 2252.09 0.81 1.28 1.46 1.15 1.27

PAGE 116

116 Table C 2. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Okeechobee 8,650 338.75 343.84 0.12 0.30 747.67 783.74 0.50 1.35 2.06 1.45 1.75 Orange 492,969 838.65 1217.15 0.31 0.47 3249.75 3784.08 0.71 1.48 1.63 1.47 1.52 Osceola 108,483 412.51 319.13 0.27 0.54 881.34 1501.45 0.40 1.16 1.66 1.21 1.38 Palm B each 525,997 818.77 934.95 0.28 0.47 2983.42 3760.38 0.73 1.30 1.37 1.28 1.28 Pasco 151,539 647.16 298.27 0.28 0.38 1445.34 1913.22 0.42 1.40 1.41 1.08 1.09 Pinellas 425,630 1604.80 1787.52 0.19 0.34 3689.88 4501.70 0.85 1.14 1.25 1.03 1.11 Polk 214,746 450.11 437.12 0.28 0.50 1244.93 2051.04 0.61 1.07 1.43 1.21 1.40 Putnam 14,148 376.88 439.90 0.24 0.59 933.35 1451.45 0.72 1.26 4.66 1.09 2.35 Santa R osa 52,202 299.93 187.84 0.30 0.51 733.67 1185.08 0.58 0.99 1.27 1.10 1.11 Sarasota 134,311 706.09 675.64 0.20 0.43 1838.73 3003.18 0.80 1.27 2.38 1.17 1.42 Seminole 193,539 875.91 986.38 0.20 0.37 2711.80 3454.30 0.92 0.93 1.07 1.22 1.27 St Johns 60,319 559.49 461.27 0.26 0.44 1447.44 2411.47 0.82 1.27 1.65 1.38 1.50 St Lucie 96,384 416.64 274.60 0.38 0.47 1145.52 1940.09 0.45 1.02 1.61 1.32 1.38 Sumter 5,950 605.70 180.88 0.31 0.58 622.15 1160.70 0.50 0.81 0.81 1.47 1.31 Taylor 3,348 287.73 374.15 0.13 0.26 541.28 654.27 0.83 1.60 2.11 1.72 1.84 Volusia 197,849 615.42 538.75 0.24 0.47 1486.17 2359.49 0.70 1.02 1.06 1.11 1.18 Walton 4,938 709.97 369.01 0.31 0.47 655.26 925.38 0.92 0.97 0.96 1.21 1.27 Identical among scenarios.

PAGE 117

117 Table C 3 Urban form outcomes South Scenario County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Alachua 79,918 948.83 1252.69 0.26 0.43 4074.06 5612.59 0.67 1.39 1.87 1.49 1.64 Bay 69,322 414.90 367.70 0.47 0.57 1478.44 2266.97 0.42 1.02 2.22 1.01 1.73 Bradford 3,335 244.79 364.09 0.19 0.48 518.34 701.21 0.60 1.45 5.96 1.36 2.50 Brevard 213,531 656.98 570.55 0.23 0.47 1406.25 2203.52 0.75 1.12 1.35 1.12 1.27 Broward 796,783 1388.31 1910.34 0.18 0.31 4866.89 5541.52 0.70 1.17 1.43 1.16 1.22 Charlotte 50,294 792.15 402.46 0.19 0.60 1203.73 3111.85 0.47 1.18 1.74 1.21 1.43 Citrus 21,608 403.83 176.87 0.15 0.40 461.99 811.61 0.66 1.09 1.19 1.11 1.12 Clay 54,396 409.46 390.22 0.18 0.51 1017.53 1296.77 0.70 1.08 1.29 1.19 1.28 Collier 108,761 531.30 690.18 0.20 0.45 1574.14 2417.51 0.69 1.22 1.16 1.26 1.43 Desoto 5,536 479.11 596.29 0.16 0.31 1037.81 1376.06 0.41 2.15 7.48 2.24 3.55 Duval 428,926 627.52 880.45 0.35 0.47 3116.57 3971.57 0.77 1.36 1.62 1.46 1.55 Escambia 125,354 560.37 733.19 0.32 0.55 2125.45 3025.43 0.57 1.47 2.05 1.74 2.04 Flagler 19,783 499.68 181.65 0.28 0.39 808.53 919.14 0.52 1.30 1.40 1.96 1.90 Hendry 9,284 521.58 570.96 0.15 0.36 1021.15 1333.50 0.58 0.90 1.03 1.06 1.09 Hernando 42,234 555.05 286.90 0.23 0.32 832.73 890.66 0.70 0.89 1.33 1.44 1.12 Highlands 24,612 482.09 352.80 0.09 0.37 675.03 1185.18 0.35 1.22 1.69 1.14 1.29 Hillsborough 531,022 829.14 1059.39 0.33 0.53 3507.53 4523.61 0.82 1.43 2.02 1.46 1.65 Indian R iver 47,658 478.24 416.54 0.27 0.47 1136.36 1909.64 0.66 1.23 1.99 1.13 1.37 Lake 81,537 618.28 447.27 0.11 0.32 962.39 1363.75 0.78 1.11 1.49 1.14 1.28 Lee 206,137 551.63 500.69 0.22 0.45 1506.97 2649.70 0.80 1.25 1.65 1.22 1.45 Leon 114,843 481.77 776.76 0.36 0.62 3284.50 4823.81 0.68 1.47 2.64 1.32 1.86

PAGE 118

118 Table C 3. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Manatee 124,812 653.61 757.46 0.24 0.50 2366.91 4237.35 0.77 1.42 2.16 1.47 1.73 Marion 79,836 360.99 339.55 0.25 0.54 758.05 1703.05 0.49 1.09 2.07 1.11 1.52 Martin 52,310 451.11 479.14 0.19 0.45 1100.94 1725.67 0.61 1.50 2.48 1.45 1.61 Miami Dade 956,389 1410.62 2268.62 0.19 0.35 5149.70 5601.20 0.75 1.26 1.68 1.26 1.44 Monroe 30,560 869.83 609.82 0.26 0.48 936.46 1127.01 0.89 1.07 1.50 0.93 0.92 Nassau 14,449 367.40 368.91 0.34 0.51 1215.43 1689.33 0.58 1.36 1.37 2.10 2.47 Okaloosa 73,233 627.09 681.57 0.35 0.45 2258.98 2867.17 0.87 1.30 1.49 1.17 1.29 Okeechobee 8,650 338.75 343.84 0.17 0.50 660.85 1231.04 0.57 1.26 1.92 1.50 1.82 Orange 492,969 838.65 1217.15 0.33 0.52 3457.24 4456.24 0.66 1.52 1.67 1.50 1.55 Osceola 108,483 412.51 319.13 0.26 0.51 1011.96 1852.36 0.39 1.25 1.78 1.28 1.46 Palm B each 525,997 818.77 934.95 0.28 0.47 2895.93 3863.35 0.76 1.31 1.38 1.28 1.28 Pasco 151,539 647.16 298.27 0.27 0.39 1576.18 1859.71 0.43 1.39 1.40 1.07 1.08 Pinellas 425,630 1604.80 1787.52 0.18 0.33 3792.47 4353.62 0.84 1.20 1.31 1.32 1.41 Polk 214,746 450.11 437.12 0.27 0.51 1375.53 2584.09 0.67 1.09 1.45 1.20 1.38 Putnam 14,148 376.88 439.90 0.19 0.69 571.02 1494.47 0.61 1.39 5.16 1.16 2.49 Santa R osa 52,202 299.93 187.84 0.26 0.44 555.51 855.32 0.54 0.99 1.26 1.09 1.10 Sarasota 134,311 706.09 675.64 0.18 0.45 1718.26 2706.12 0.73 1.31 2.45 1.18 1.43 Seminole 193,539 875.91 986.38 0.20 0.37 2631.20 3428.47 0.91 0.94 1.08 1.26 1.31 St Johns 60,319 559.49 461.27 0.27 0.44 1224.38 1678.28 0.81 1.26 1.64 1.37 1.48 St Lucie 96,384 416.64 274.60 0.38 0.50 1164.05 2116.11 0.40 1.01 1.60 1.35 1.40 Sumter 5,950 605.70 180.88 0.32 0.49 703.94 987.39 0.58 0.85 0.84 1.63 1.45

PAGE 119

119 Table C 3. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Taylor 3,348 287.73 374.15 0.18 0.14 810.98 743.62 0.71 1.45 1.92 1.63 1.75 Volusia 197,849 615.42 538.75 0.26 0.51 1459.27 2646.53 0.74 1.01 1.06 1.10 1.16 Walton 4,938 709.97 369.01 0.26 0.43 977.11 1447.16 0.88 0.97 0.97 1.21 1.27 Identical among scenarios. Table C 4 Urban form outcomes Southwest Scenario County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Alachua 79,918 948.83 1252.69 0.27 0.46 4244.83 6196.21 0.74 1.39 1.89 1.54 1.71 Bay 69,322 414.90 367.70 0.48 0.58 1423.18 2295.11 0.40 1.01 2.20 1.00 1.73 Bradford 3,335 244.79 364.09 0.24 0.57 800.89 1281.99 0.75 1.48 4.70 1.36 2.51 Brevard 213,531 656.98 570.55 0.24 0.47 1403.70 2104.59 0.75 1.14 1.39 1.15 1.31 Broward 796,783 1388.31 1910.34 0.18 0.30 4773.17 5429.28 0.92 1.20 1.45 1.17 1.23 Charlotte 50,294 792.15 402.46 0.23 0.55 1110.07 2902.19 0.36 1.19 1.75 1.22 1.44 Citrus 21,608 403.83 176.87 0.15 0.44 495.36 988.94 0.67 1.10 1.20 1.14 1.15 Clay 54,396 409.46 390.22 0.19 0.48 979.50 1348.27 0.72 1.09 1.30 1.19 1.27 Collier 108,761 531.30 690.18 0.15 0.48 1354.56 2117.67 0.70 1.17 1.11 1.19 1.35 Desoto 5,536 479.11 596.29 0.15 0.38 541.33 728.40 0.33 2.17 7.54 2.15 3.41

PAGE 120

120 Table C 4. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Duval 428,926 627.52 880.45 0.37 0.47 3104.36 3978.39 0.80 1.37 1.63 1.47 1.56 Escambia 125,354 560.37 733.19 0.33 0.57 2124.40 3221.96 0.57 1.44 2.01 1.70 1.99 Flagler 19,783 499.68 181.65 0.29 0.42 759.21 972.65 0.44 1.30 1.40 2.06 2.01 Hendry 9,284 521.58 570.96 0.12 0.37 770.93 1316.18 0.58 1.02 1.18 1.07 1.10 Hernando 42,234 555.05 286.90 0.25 0.33 817.98 911.47 0.58 0.86 1.28 1.40 1.09 Highlands 24,612 482.09 352.80 0.11 0.26 699.12 984.79 0.28 1.26 1.74 1.17 1.32 Hillsborough 531,022 829.14 1059.39 0.33 0.52 3733.16 4913.85 0.78 1.43 2.02 1.46 1.65 Indian R iver 47,658 478.24 416.54 0.23 0.44 1081.16 1898.92 0.68 1.24 2.01 1.14 1.38 Lake 81,537 618.28 447.27 0.12 0.39 968.09 1518.37 0.69 1.14 1.52 1.16 1.30 Lee 206,137 551.63 500.69 0.23 0.43 1485.69 2154.97 0.71 1.30 1.72 1.28 1.51 Leon 114,843 481.77 776.76 0.38 0.66 3058.57 4970.01 0.66 1.49 2.67 1.32 1.87 Manatee 124,812 653.61 757.46 0.23 0.50 2241.43 3907.45 0.77 1.40 2.12 1.43 1.68 Marion 79,836 360.99 339.55 0.24 0.54 750.22 1725.88 0.43 1.08 2.05 1.11 1.53 Martin 52,310 451.11 479.14 0.20 0.46 1073.45 1617.59 0.70 1.57 2.60 1.53 1.70 Miami Dade 956,389 1410.62 2268.62 0.18 0.36 5575.64 5585.77 0.73 1.24 1.64 1.23 1.40 Monroe 30,560 869.83 609.82 0.27 0.47 897.82 1598.06 0.97 1.26 1.59 0.94 0.94 Nassau 14,449 367.40 368.91 0.27 0.47 637.94 1149.76 0.70 1.36 1.38 2.07 2.43 Okaloosa 73,233 627.09 681.57 0.34 0.46 2172.17 2450.93 0.86 1.37 1.56 1.20 1.32 Okeechobee 8,650 338.75 343.84 0.09 0.30 663.31 844.93 0.67 1.26 1.92 1.49 1.81 Orange 492,969 838.65 1217.15 0.33 0.51 3296.12 4087.26 0.65 1.46 1.61 1.45 1.49 Osceola 108,483 412.51 319.13 0.25 0.49 1037.08 1840.09 0.38 1.22 1.74 1.24 1.42

PAGE 121

121 Table C 4. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Palm B each 525,997 818.77 934.95 0.28 0.48 3031.97 3981.76 0.77 1.27 1.34 1.25 1.26 Pasco 151,539 647.16 298.27 0.28 0.38 1510.87 1878.44 0.41 1.38 1.38 1.07 1.07 Pinellas 425,630 1604.80 1787.52 0.18 0.33 3840.58 4567.11 0.80 1.16 1.27 1.29 1.39 Polk 214,746 450.11 437.12 0.28 0.50 1261.32 2133.90 0.67 1.07 1.46 1.22 1.40 Putnam 14,148 376.88 439.90 0.21 0.70 662.60 1789.71 0.64 1.41 5.23 1.14 2.45 Santa R osa 52,202 299.93 187.84 0.30 0.48 643.52 1183.10 0.64 1.03 1.31 1.11 1.12 Sarasota 134,311 706.09 675.64 0.21 0.43 1874.88 2995.29 0.70 1.32 2.47 1.18 1.43 Seminole 193,539 875.91 986.38 0.19 0.38 2624.66 3631.29 0.91 0.95 1.09 1.26 1.32 St Johns 60,319 559.49 461.27 0.26 0.46 1338.07 2211.52 0.77 1.28 1.67 1.38 1.49 St Lucie 96,384 416.64 274.60 0.37 0.47 1179.56 1747.98 0.43 1.04 1.65 1.37 1.42 Sumter 5,950 605.70 180.88 0.34 0.54 829.09 1164.54 0.59 0.83 0.83 1.51 1.34 Taylor 3,348 287.73 374.15 0.10 0.24 679.69 800.05 0.75 1.55 2.05 1.68 1.80 Volusia 197,849 615.42 538.75 0.24 0.50 1501.80 2571.59 0.76 1.01 1.06 1.11 1.18 Walton 4,938 709.97 369.01 0.32 0.47 1097.67 1535.02 0.88 0.98 0.97 1.20 1.27 Identical among scenarios.

PAGE 122

122 Table C 5 Mean values for urban form outcomes County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Alachua 79,918 948.83 1252.69 0.25 0.43 3903.54 5624.06 0.72 1.41 1.91 1.55 1.70 Bay 69,322 414.90 367.70 0.48 0.58 1476.88 2313.06 0.40 1.02 2.22 1.01 1.73 Bradford 3,335 244.79 364.09 0.20 0.53 551.89 863.46 0.57 1.58 6.11 1.34 2.47 Brevard 213,531 656.98 570.55 0.24 0.47 1359.82 1989.34 0.73 1.13 1.38 1.14 1.29 Broward 796,783 1388.31 1910.34 0.17 0.30 4795.82 5628.01 0.86 1.20 1.45 1.18 1.24 Charlotte 50,294 792.15 402.46 0.21 0.57 1137.28 2700.17 0.40 1.17 1.71 1.21 1.42 Citrus 21,608 403.83 176.87 0.16 0.41 460.17 836.83 0.63 1.09 1.19 1.11 1.13 Clay 54,396 409.46 390.22 0.21 0.51 1007.27 1519.24 0.69 1.10 1.27 1.21 1.29 Collier 108,761 531.30 690.18 0.18 0.46 1473.73 2207.38 0.73 1.19 1.13 1.22 1.38 Desoto 5,536 479.11 596.29 0.19 0.38 1009.86 1361.66 0.40 1.99 6.91 2.10 3.33 Duval 428,926 627.52 880.45 0.37 0.48 3044.04 3948.27 0.77 1.37 1.63 1.47 1.56 Escambia 125,354 560.37 733.19 0.35 0.56 1881.82 2805.80 0.57 1.47 2.04 1.73 2.03 Flagler 19,783 499.68 181.65 0.33 0.39 708.91 884.76 0.50 1.32 1.42 1.94 1.89 Hendry 9,284 521.58 570.96 0.23 0.34 652.23 1004.87 0.58 0.97 1.12 1.07 1.10 Hernando 42,234 555.05 286.90 0.24 0.32 874.13 986.05 0.60 0.89 1.32 1.43 1.11 Highlands 24,612 482.09 352.80 0.11 0.34 725.99 1158.50 0.32 1.27 1.76 1.16 1.31 Hillsborough 531,022 829.14 1059.39 0.33 0.53 3541.40 4834.53 0.80 1.41 1.87 1.45 1.62 Indian R iver 47,658 478.24 416.54 0.22 0.44 1014.48 1667.97 0.68 1.25 2.02 1.14 1.39 Lake 81,537 618.28 447.27 0.13 0.37 1006.34 1565.61 0.74 1.13 1.52 1.15 1.29 Lee 206,137 551.63 500.69 0.23 0.44 1504.27 2544.07 0.75 1.30 1.72 1.26 1.49 Leon 114,843 481.77 776.76 0.37 0.61 3285.57 4966.59 0.68 1.49 2.67 1.33 1.88

PAGE 123

123 Table C 5. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Manatee 124,812 653.61 757.46 0.24 0.50 2316.04 3983.95 0.77 1.42 2.16 1.45 1.71 Marion 79,836 360.99 339.55 0.24 0.55 741.35 1697.50 0.46 1.10 2.08 1.12 1.54 Martin 52,310 451.11 479.14 0.26 0.48 1140.99 1674.14 0.64 1.57 2.61 1.51 1.67 Miami Dade 956,389 1410.62 2268.62 0.21 0.37 5417.18 5546.92 0.75 1.26 1.68 1.26 1.44 Monroe 30,560 869.83 609.82 0.26 0.47 1020.43 1577.57 0.92 1.16 1.58 0.95 0.94 Nassau 14,449 367.40 368.91 0.30 0.51 877.66 1344.41 0.67 1.35 1.37 2.09 2.45 Okaloosa 73,233 627.09 681.57 0.37 0.45 1948.69 2273.80 0.84 1.31 1.49 1.17 1.29 Okeechobee 8,650 338.75 343.84 0.13 0.39 677.63 983.30 0.60 1.30 1.99 1.49 1.80 Orange 492,969 838.65 1217.15 0.32 0.50 3346.10 4151.46 0.68 1.50 1.65 1.48 1.53 Osceola 108,483 412.51 319.13 0.27 0.52 978.52 1700.14 0.39 1.20 1.72 1.25 1.42 Palm B each 525,997 818.77 934.95 0.28 0.47 2959.70 3850.13 0.75 1.30 1.37 1.27 1.27 Pasco 151,539 647.16 298.27 0.28 0.38 1522.72 1880.01 0.43 1.40 1.41 1.08 1.08 Pinellas 425,630 1604.80 1787.52 0.18 0.33 3753.41 4396.59 0.83 1.17 1.28 1.23 1.32 Polk 214,746 450.11 437.12 0.28 0.51 1312.16 2286.46 0.66 1.08 1.44 1.20 1.39 Putnam 14,148 376.88 439.90 0.22 0.64 735.84 1481.24 0.65 1.34 4.97 1.13 2.43 Santa R osa 52,202 299.93 187.84 0.29 0.47 642.19 1034.89 0.57 1.00 1.27 1.10 1.11 Sarasota 134,311 706.09 675.64 0.19 0.44 1779.88 2828.83 0.75 1.28 2.40 1.18 1.43 Seminole 193,539 875.91 986.38 0.20 0.36 2692.42 3478.21 0.92 0.93 1.07 1.24 1.29 St Johns 60,319 559.49 461.27 0.26 0.43 1330.13 2030.10 0.79 1.26 1.64 1.38 1.49 St Lucie 96,384 416.64 274.60 0.38 0.48 1174.05 2027.66 0.44 1.01 1.60 1.34 1.39 Sumter 5,950 605.70 180.88 0.33 0.54 705.81 1095.06 0.54 0.83 0.82 1.55 1.38

PAGE 124

124 Table C 5. Continued County n ame Population size Housing u nit d ensity Job d e n sity Housing u nit c oncentration J ob c oncentration Mix of j ob/ h ousing Mix of h ousing/ j ob Continuity Housing u nit c entrality Job c entrality Housing u nit p rox i mity Job proxi mity Taylor 3,348 287.73 374.15 0.14 0.22 664.86 699.08 0.78 1.52 2.01 1.67 1.80 Volusia 197,849 615.42 538.75 0.25 0.49 1477.83 2495.38 0.73 1.03 1.06 1.11 1.17 Walton 4,938 709.97 369.01 0.28 0.45 839.88 1194.10 0.91 0.97 0.96 1.22 1.29 Identical among scenarios.

PAGE 125

125 Table C 6 Composite z scores of urban form factors County name Size/ density/ mixed use factor Centrality/job proximity factor Concen tration factor Housing proximity factor Continuity factor Miami Dade 14.336 0.326 1.351 0.235 0.561 Broward 12.384 1.143 2.499 0.502 1.217 Hillsborough 6.247 0.76 0 1.636 0.423 0.835 Palm Beach 4.825 0.762 0.454 0.179 0.569 Orange 5.842 0.732 1.322 0.53 0 0.134 Duval 3.849 0.31 0 1.59 0 0.484 0.685 Pinellas 9.543 1.239 2.082 0.329 1.038 Polk 1.01 0 1.312 0.862 0.41 0 0.003 Brevard 0.267 1.338 0.076 0.633 0.457 Lee 0.286 0.046 0.414 0.225 0.563 Volusia 0.103 2.22 0 0.358 0.741 0.45 0 Seminole 3.219 2.339 1.55 0 0.291 1.572 Pasco 1.073 0.7 00 0.402 0.838 1.346 Sarasota 0.644 0.299 0.849 0.505 0.548 Escambia 0.355 1.923 2.275 1.358 0.539 Manatee 1.8 00 1.218 0.295 0.436 0.68 0 Leon 2.699 2.193 2.913 0.006 0.111 Collier 0.612 1.125 0.763 0.372 0.416 Osceola 2.477 0.538 0.797 0.273 1.576 St. Lucie 2.24 0 1.422 1.736 0.044 1.312 Lake 1.77 0 1.257 2.228 0.605 0.471 Alachua 5.879 0.97 0 0.234 0.747 0.393 Marion 2.906 0.435 0.775 0.696 1.161 Okaloosa 0.068 0.588 1.343 0.525 1.053 Bay 1.737 0.235 3.84 0 1.082 1.514 St. Johns 1.446 0.242 0.128 0.169 0.813 Clay 2.664 1.532 0.069 0.392 0.161 Martin 2.153 2.05 0 0.363 0.613 0.116 Santa Rosa 4.037 2.311 0.625 0.77 0 0.524 Charlotte 0.632 0.697 0.603 0.406 1.55 0 Indian River 2.328 0.189 0.448 0.618 0.111 Hernando 2.948 2.67 0 1.445 0.353 0.354 Monroe 0.871 1.789 0.242 1.276 1.529 Highlands 3.113 0.465 2.803 0.561 2.024 Citrus 4.153 1.951 1.486 0.716 0.189 Flagler 3.631 0.598 0.224 2.091 0.919

PAGE 126

126 Table C 6 Continued County name Size/ density/ mixed use factor Centrality/job proximity factor Concen tration factor Housing proximity factor Continuity factor Nassau 3.225 1.787 1.171 2.571 0.071 Putnam 3.075 4.535 1.465 0.664 0.062 Hendry 2.762 2.565 1.338 0.865 0.453 Okeechobee 3.785 0.814 1.995 0.552 0.338 Sumter 3.238 2.753 1.705 0.75 0 0.681 Desoto 2.333 10.322 1.529 2.614 1.55 0 Walton 2.362 2.302 0.344 0.357 1.47 0 Taylor 4.093 1.624 3.683 1.172 0.726 Bradford 4.227 6.416 0.118 0.063 0.511 Note: This table is sorted by the number of employed residents in each county level UA from the highest to the lowest value.

PAGE 127

127 APPENDIX D TRAFFIC CONGESTION O UTCOMES Table D 1. Traffic congestion outcomes County n ame RCI Delay per Capita TTI Miami Dade 1.62 45.2 1.62 Broward 1.62 59.7 1.67 Hillsborough 1.57 46.7 1.48 Palm Beach 1.31 35.6 1.35 Orange 1.43 43 .0 1.45 Duval 1.24 29.9 1.3 0 Pinellas 1.63 32.4 1.44 Polk 1.17 21.3 1.22 Brevard 1.37 46.6 1.34 Lee 1.39 28.6 1.34 Volusia 1.09 24.5 1.2 0 Seminole 1.42 31.7 1.4 0 Pasco 1.65 45.1 1.46 Sarasota 1.49 30.7 1.38 Escambia 1.16 21.3 1.23 Manatee 1.38 33.2 1.32 Leon 1.23 24.2 1.27 Collier 1.3 0 19.8 1.32 Osceola 1.47 48.6 1.36 St. Lucie 1.09 36.1 1.23 Lake 1.4 0 33.2 1.36 Alachua 1.15 19.8 1.21 Marion 1.06 26.5 1.17 Okaloosa 1.77 57.9 1.51 Bay 1.48 36.1 1.39 St. Johns 1.41 13.1 1.37 Clay 1.83 50.7 1.52 Martin 1.2 0 41.6 1.24 Santa Rosa 1.49 38.8 1.41 Charlotte 1.15 22.2 1.24 Indian River 0.92 9.4 1.13 Hernando 1.02 14.2 1.2 0 Monroe 1.56 79.8 1.42 Highlands 1.2 0 31.8 1.31 Citrus 1.19 34.5 1.26 Flagler 1.12 20.2 1.09 Nassau 1.11 4.4 1.34 Putnam 1.15 25 .0 1.24

PAGE 128

128 Table D 1. Continued County n ame RCI Delay per Capita TTI Hendry 0.93 5.2 1.09 Okeechobee 1.08 27.6 1.21 Sumter 0.96 12.6 1.11 Desoto 0.9 0 6.9 1.07 Walton 1.65 131.7 1.46 Taylor 0.56 0.2 1 .00 Bradford 1.25 42.6 1.27 Note: This table is sorted by the number of employed residents in each county level UA from the highest to the lowest value.

PAGE 129

129 LIST OF REFERENCES Anderson, W. P., Kanaroglou, P. S., & Miller, E. J. (1996). Urban form, energy and the environment: A review of issues, evidence and policy. Urban Studies, 33 (1), 7. Arafat, A., Steiner, R. L., Zwick, P., & Srinivasan, S. (2010, August). Urban form measurements for land use transportation coordination: An optimal zonal size approach to reduce the effect of the modifiable areal unit problem. Pr esented at the Palestinian Faculty Development Program 2010 West Bank Academic Colloquium, Ramallah, Palestine. Banister, D. (1996). Energy, quality of life and the environment: The role of transport. Transport Reviews, 16 (1), 23 35. Bartuska, T. J., Bar tuska, T. J. B. e., & McClure, W. R. (2007). The built environment : A collaborative inquiry into design and planning (2nd ed.). Hoboken: John Wiley & Sons. Boarnet, M. G., & Sarmiento, S. (1998). Can land use policy really affect travel behaviour? A stud y of the link between non work travel and land use characteristics. Urban Studies, 35 (7), 1155 1169. Boarnet, M. G., & Crane, R. (2001). Travel by design : The influence of urban form on travel New York: Oxford University Press. Cervero, R., & Gorham, R (1995). Commuting in transit versus automobile neighborhoods. Journal of the American Planning Association, 61 (2), 210 225. Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity, and design. Transportation Research Part D: T ransport and Environment, 2 (3), 199 219. Crane, R. (1996). Cars and drivers in the new suburbs: Linking access to travel in neotraditional planning. Journal of the American Planning Association, 62 (1), 51 65. Cutsinger, J., Galster, G., Wolman, H., Hanso n, R., & Towns, D. (2005). Verifying the multi dimensional nature of metropolitan land use: Advancing the understanding and measurement of sprawl. Journal of Urban Affairs, 27 (3), 235. Downs, A. (200 4 ). Still stuck in traffic : Coping with peak hour traff ic congestion anthony downs. [electronic resource] : Anthony downs (rev ed.). Brookings: Institution. Downs, A., & Brookings Institution. (1992). Stuck in traffic : Coping with peak hour traffic congestion Washington, D.C.; Cambridge, Mass.: Brookings In stitution; Lincoln Institute of Land Policy.

PAGE 130

130 Ewing, R., DeAnna, M. B., & Li, S. C. (1996). Land use impacts on trip generation rates. Transportation Research Record: Journal of the Transportation Research Board, 1518 ( 1), 1 6. Ewing, R., Pendall, R., & C hen, D. (2002). Measuring sprawl and its impact. Washington DC: Smart Growth America, Friedman, B., Gordon, S., & Peers, J. (1994). Effect of neotraditional neighborhood design on travel characteristics. Transportation Research Board, 1466 63 70. Galste r, G., Hanson, R., Ratcliffe, M. R., Wolman, H., Coleman, S., & Freihage, J. (2001). Wrestling sprawl to the ground: Defining and measuring an elusive concept. HOUSING POLICY DEBATE WASHINGTON 12 (4), 681 718. Gillham, O., & MacLean, A. S. (2002). The li mitless city : A primer on the urban sprawl debate Washington, DC: Island Press. Gordon, P., Kumar, A., & Richardson, H. W. (1989). Congestion, changing metropolitan structure, and city size in the united states. International Regional Science Review, 12 (1), 45. Gordon, P., Kumar, A., & Richardson, H. W. (1989). The influence of metropolitan spatial structure on commuting time. Journal of Urban Economics, 26 (2), 138 151. Gordon, P., Lee, B., & Richardson, H. W. (2004). Travel trends in US cities: Explai ning the 2000 census commuting results. Lusk Center for Real Estate, University of Southern California, Working Paper, 1007 Gordon, P., & Richardson, H. W. (1994). Congestion trends in metropoliran areas. Curbing Gridlock: Peak Period Fees to Relieve Traf fic Congestion, 2 1 31. Gordon, P., & Richardson, H. W. (1991). The commuting paradox. Journal of the American Planning Association, 57 (4), 416. Guo, J. Y., & Bhat, C. R. (2004). Modifiable areal units: A problem or a matter of perception in the context of residential location choice modeling. Transportation Research Record, 1898 138 147. Hess, G., Daley, S. S., Dennison, B. K., Lubkin, S. R., McGuinn, R. P., Morin, V. Z., Potter, K. M., Savage, R. E., Shelton, W. G., & Snow, C. M. (2001). Just what is sprawl, anyway. Carolina Planning, 26 (2), 11 26. Hills, P. J. (1996). What is induced traffic? Transportation 23(1), 5 16 Izraeli, O., & McCarthy, T. R. (1985). Variations in travel distance, travel time and model choice among SMSAs. Journal of Transport Economics and Policy, 139 160.

PAGE 131

131 Jaret, C., Ghadge, R., Reid, L. W., & Adelman, R. M. (2009). The measurement of suburban sprawl: An evaluation. City & Community, 8 (1), 65 84. Knaap, G. J., Song, Y., Ewing, R., & Clifton, K. (2009). Seeing the elephant: Multi disciplinary measures of urban sprawl. College Park. Online: Http://www.Smartgrowth.Umd.edu/research/pdf/KnaapSongEwingEtAl_ Elephant_022305.Pdf, Verfgbar Am, 17 Kulash, W., Anglin, J., & Marks, D. (1990). Traditional neighborhood development: Will the traffic work. Development, 21 21 24. Kwan, M., & Weber, J. (2008). Scale and accessibility: Implications for the analysis o f land use travel interaction. Applied Geography, 28 (2), 110 123. doi:10.1016/j.apgeog.2007.07.002 Litman, T. (2004). Generated traffic and induced travel. Implications for Transport Planning.Victoria: Victoria Transport Policy Institute, Malpezzi, S. (1 999). Estimates of the measurement and determinants of urban sprawl in US metropolitan areas. Unpublished Paper.University of Wisconsin, Madison Center for Urban Land Economics Research, easures of urban form in US metropolitan areas. The Center for Urban Land Economics Research, University of Wisconsin, Madison, WI.( Http://www.Bus.Wisc.edu/realestate/d ocs/docs/Alternative% 20 Measures% 20of% 20Urban% 20Form.Doc), Massey, D. S., & Denton, N. A. (1988). The dimensions of residential segregation. Social Forces, 67 (2), 281 315. McNally, M. G., & Kulkarni, A. (1997). Assessment of influence of land use tra nsportation system on travel behavior. Transportation Research Record: Journal of the Transportation Research Board, 1607 ( 1), 105 115. Messenger, T., & Ewing, R. (1996). Transit oriented development in the sun belt. Transportation Research Record: Journa l of the Transportation Research Board, 1552 ( 1), 145 153. Openshaw, S. (1983). The modifiable areal unit problem Norwick Norfolk: Geo Books. Owens, S. E. (1986). Energy, planning, and urban form London: Pion; Distributed by Methuen USA. Sarzynski, A. Wolman, H. L., Galster, G., & Hanson, R. (2006). Testing the conventional wisdom about land use and traffic congestion: The more we sprawl, the less we move? Urban Studies, 43 (3), 601.

PAGE 132

132 Stead, D., & Marshall, S. (2001). The relationships between urban fo rm and travel patterns. an international review and evaluation. EJTIR, 1 (2), 113 141. Taylor, B. D. (1995). Public perceptions, fiscal realities, and freeway planning. Journal of the American Planning Association, 61 (1), 43. Tsai, Y. H. (2005). Quantifyi ng urban form: Compactness versus' sprawl'. Urban Studies, 42 (1), 141. Wheaton, W. C. (1998). Land use and density in cities with congestion* 1. Journal of Urban Economics, 43 (2), 258 272. YIN, M., & SUN, J. (2007). The impacts of state growth management programs on urban sprawl in the 1990s. Journal of Urban Affairs, 29 (2), 149 179. doi:10.1111/j.1467 9906.2007.00332.x

PAGE 133

133 BIOGRAPHICAL SKETCH Ruoniu Wang was born in Chengdu, China in 1985. He attended Sichuan University and graduated in July 2008 with university and departmental honors and a degree in urban planning and design. During the undergraduate studies he spent one year on exchange at University of Washington Seattle majoring in urban planning and design, from where he decided to p ursue the graduate study in the U.S. As a graduate student at the University of Florida in the Department of Urban and Regional Planning Ruoniu worked at assistantships with the Center for Building Better Communities (CBBC) and later with the Shimberg Cent er for Housing Studies (SCHS). At the CBBC Ruoniu served as the GIS Analyst for the development of School Concurrency Model. At the SCHS he served as the GIS Assistant for the management of Florida Assisted Housing Inventory. In addition, he worked as a Re search Assistant on the project T he Economic Cost of Traffic Congestion in Florida funded by Florida Department of Transportation (FDOT). He was a co writer of the paper on the same topic that was presented at the Association of Collegiate Schools of Planning (ACSP) 2010 Annual Conference During his time at the University of Florida Ruoniu was awarded the David K. Maltby International Travel Memorial Scholarship. In Summer 2010 semester, he attended the University Abroad Program in Curitiba, Brazil.