|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 TRENDS IN IRRIGATION USE OF POTABLE WATER BY THE SINGLE FAMILY RESIDENTIAL SECTOR IN ALACHUA COUNTY FLORIDA By JOHN EDWARD PALENCHAR A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILL MENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2009
2 2009 John Edward Palenchar
3 To my wife and children
4 ACKNOWLEDGMENTS My thanks go out to Dr. Jim Heaney, Mr Ken Freidman and the Conserve Florida Water Clearinghouse team for asking the right questions, editing, data support, and general help along the way. It would have been impos si ble to perform any of the case study analysis without the data provided by Ri ck Hutton and Melissa Stewart of Gainesville Regional Utilities. I would like to thank Dave Bracciano of Tampa Bay Water Inc. as well as Norm Davis and John McCary at Hillsborough County Water Resources Services for their support in preliminary studies. Last ly, I thank the financial contributors that support the Conserve Florida Water Clearinghouse. In this regard my gratitude goes out to the Florida Department of Environmental Protection, South anagement District, and the Southwest Florida Water Management District.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF FIGURES ................................ ................................ ................................ ......................... 9 ABSTRACT ................................ ................................ ................................ ................................ ... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 14 Problem ................................ ................................ ................................ ................................ ... 14 Proposed Approach ................................ ................................ ................................ ................. 15 2 FLORIDA DATA SOURCES FOR WATER USE ANALYSIS ................................ ........... 19 Introduction ................................ ................................ ................................ ............................. 19 FDEP water production and facility data ................................ ................................ ................ 19 FAWN and NOAA /NCDC Climatic Data ................................ ................................ ............. 20 FD OR Parcel Data ................................ ................................ ................................ .................. 21 U.S. Census Bureau Block Level Data ................................ ................................ ................... 22 ................................ ................................ ........... 23 Water Management District Service Utility Service Area Boundaries ................................ .. 24 Summary and Conclusions ................................ ................................ ................................ ..... 25 3 DETERMINATION OF IRRIGABLE AREA FOR SINGLE FAMILY RESIDENCES ..... 28 Introduction ................................ ................................ ................................ ............................. 28 Methods t o Estimate Irrigable Landscape A rea ................................ ................................ ...... 28 Terminology ................................ ................................ ................................ .................... 28 Measurement Methods ................................ ................................ ................................ .... 30 Previous Studies ................................ ................................ ................................ .............. 30 Irrigable Area Using a Parcel Area Budget ................................ ................................ ............ 33 Introduction ................................ ................................ ................................ ..................... 33 To tal Parcel Area ( TA ) ................................ ................................ ................................ ..... 35 Impervious Area Components ( IA ) ................................ ................................ ................. 38 Footprint o f s tructure ( FS ) ................................ ................................ ........................ 39 Associated i mpervious a rea ( AIA ) ................................ ................................ ............ 42 Non Applicable Area ( NA ) ................................ ................................ .............................. 44 Irrigable o r Pervious Area ( PA ) ................................ ................................ ....................... 45 Prevalence o f In Ground Irrigation Systems ................................ ................................ .......... 47 Summary a nd Conclusions ................................ ................................ ................................ ..... 49 4 DETERMINATION OF IRRIGATION WATER USE ................................ ......................... 65 Background ................................ ................................ ................................ ............................. 65
6 Results o f Previous Studies o f Outdoor Water Use ................................ ................................ 67 University o f Florida Studies ................................ ................................ ........................... 67 Studies Performed b y Aquacraft Inc ................................ ................................ .............. 69 Flori da Studies b y Whitcomb ................................ ................................ .......................... 69 Effect o f In Ground Irrigation Water Use i n t he Alachua County Study Area ...................... 71 Motivation ................................ ................................ ................................ ....................... 71 Analysis o f SFR Water Use ................................ ................................ ............................. 72 SFR Water Use Patterns ................................ ................................ ........................... 73 Water Use Trends ................................ ................................ ................................ ..... 77 Determining Reasonableness o f a n Irrigable Area Application Rate .............................. 78 Conclusions ................................ ................................ ................................ ............................. 79 5 CLUSTERING THE IRRIGATION USERS IN THE POTABLE WATER SYSTEM ........ 92 Background ................................ ................................ ................................ ............................. 92 Potable Water Offset ................................ ................................ ................................ ....... 93 K Means Clustering ................................ ................................ ................................ ......... 95 Prior Measures o f Indoor Use ................................ ................................ ................................ 96 Clustering o f To tal Water Use f or All Accounts ................................ ................................ .... 99 Clustering Results b y Homes Built Before 1985 Versus During Or After 1985 .......... 101 Extrapolation o f Gainesville Results t o Other Florida Utilities ................................ .... 102 Generalization o f Clusters Using Discriminant Analysis ................................ ..................... 104 Conclusions ................................ ................................ ................................ ........................... 105 6 SUMMARY AND CONCLUSIONS ................................ ................................ ................... 118 APPENDIX A ANNUAL TRENDS OF PARCEL SQUARE FOOTAGE IN THE ALACHUA COUNTY STUDY AREA ................................ ................................ ................................ ... 123 B ANNUAL TIME SERIES SHOWING BREAKDOWN OF 1, 1.5, 2, 2.5, AND 3 STORY HOMES IN THE ALACHUA COUNTY STUDY AREA ................................ .... 125 LIST OF REFERENCES ................................ ................................ ................................ ............. 127 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 132
7 LIST OF TABLES Table page 3 1 Comparison of parcel areas reported in Aquacraft Inc. reports (DeOreo et al 2008; Mayer and DeOreo 1999; Mayer et al 2009). ................................ ................................ .... 58 3 2 Summary of datasets used in the Alachua County study area analysis ............................. 58 3 3 Complete table of all GRU parcels that linked to DOR use code ................................ ...... 5 9 3 4 Summary statistics of parcel areas showing single, dual meter groups and sprinkler system groups in Alachua County ................................ ................................ ..................... 59 3 5 Median parcel total area for the three period groups used to describe the Alachua Counts Study area. ................................ ................................ ................................ ............. 60 3 6 Heated area, effective area, and gross area for three periods along with the ratios that relate these areas to each other ................................ ................................ ........................... 60 3 7 Summary statistics comparing various measur es of structure area ................................ ... 61 3 8 Comparison table showing the contribution of various associated impervious area reported by ACPA ................................ ................................ ................................ .............. 62 3 9 Average dimensions for a swimming pool based on the observed data, Adapted from (Mayer and DeOreo 1999) ................................ ................................ ................................ 63 3 10 Summary statistics comparing parcel area and two estimations of irrigable ar ea between dual meter, single meter, sprinkler, non sprinkler, and overall groupings ......... 63 3 11 Summary of attributes for single and dual metered customers for Alachua County area ................................ ................................ ................................ ................................ ..... 64 3 12 Summary of attributes for accounts with and without sprinkler systems tagged by the ACPA ................................ ................................ ................................ ................................ 64 4 1 Summary of irrigation application rates ......... 89 4 2 Count of accounts from the utilities in Whitcomb's database with parcel area and irrigation from the potable water system ................................ ................................ ........... 89 4 3 Summary of irrigation application rates from 409 Florida accounts within 4 profile groups, using data provided by Whitcomb, 2005 ................................ .............................. 89 4 4 In gro und versus surface irrigation for 88 homes in Denver, CO (adapted from Stadjuhar (1997)) ................................ ................................ ................................ ............... 90 4 5 Comparison of accounts with/without in ground irrigation and dual meters in the Alachua County s tudy area ................................ ................................ ................................ 90
8 4 6 Monthly rainfall and ET 0 for the Alachua County study area from October 2007 to September 2008 ................................ ................................ ................................ ................. 91 5 1 GRU rate structure for irrigation and regular metered accounts during period of observation ................................ ................................ ................................ ....................... 114 5 2 Results of current methods for calculating the indoor water use in the study area ......... 114 5 3 The k means cluster centroids and water use percentages ................................ ............... 114 5 4 Summary of selected cluster attributes form the Alachua County study area ................. 115 5 5 Comparison table of selected attributes for pre 1985 group and the 1985 to 2007group ................................ ................................ ................................ ........................ 116 5 6 Default savings estimate for BMP 1 (non potable irrigation source replacements) for single family residential users ................................ ................................ .......................... 116 5 7 Training dataset discriminant analysis function parameters ................................ ............ 117 5 8 Confusion matrix for the results of the 1.50 cut off discriminant analysis run on the training dataset ................................ ................................ ................................ ................. 117 5 9 Confusion matrix for the results of the 1.42 o ptimized cut off discriminant analysis run on the training dataset ................................ ................................ ................................ 117
9 LIST OF FIGURES Figure page 2 1 Florida's publicly supplied water production f rom January 1999 through December of 2008, as reported by Florida Department of Environmental Protection ........................ 26 2 2 Schematic Representation of table and GIS relationships for Florida Department o f Revenue provided data 2009 ................................ ................................ .............................. 27 3 1 Parcel area budget ................................ ................................ ................................ .............. 50 3 2 Overview of Alachua County study area showing spatial distribution of single and dual metered customers ................................ ................................ ................................ ..... 50 3 3 Time series showing fluctuation in median parcel area by year from the Whitcomb (2005) database ................................ ................................ ................................ .................. 51 3 4 Image of Alachua County, zero lot line, single family parcels ................................ ......... 52 3 5 Example of median dual metered customers ................................ ................................ ..... 52 3 6 Example of median single metered customers ................................ ................................ .. 53 3 7 Moving average trend line showing median parcel areas in the Alachua County study area from 1960 to 2007 ................................ ................................ ................................ ...... 53 3 8 Time Series Comparison showing annual average of heated area, effective area, and gross area of the structure in the Alachua County Study area for single family residences built in the indicated year ................................ ................................ ................. 54 3 9 Cumulative count of homes built showing dominance of single story homes with a growth in popularity of two story homes ................................ ................................ ........... 54 3 10 Time se ries showing: annual average footprint of structure estimated using property ACPA ................................ ................................ ................................ ................................ 55 3 11 Linear regression o f associated impervious area as a function of effective area for 30,910 parcels the line fit equation is (y=0.69x) with an R2 value of 0.60 ....................... 55 3 12 Annually time series of the cumulative compo nents of total parcel area (TA) which are: pervious area (PA), associated impervious area (AIA), and the footprint of the structure (FS) on the property ................................ ................................ ............................ 56 3 13 Percent of single family reside nces in Alachua County Florida with in ground sprinklers ................................ ................................ ................................ ............................ 56
10 3 14 Time series comparing irrigable area for home with and without in ground irrigation systems ................................ ................................ ................................ ............................... 57 3 15 Annual time series showing the change in percent imperviousness of median parcels for the Alachua County study group from 1960 to 2007 ................................ ................... 57 4 1 Overall seaso ridge collected over two years (addapted from Baum 2005) ................................ ............. 82 4 2 Average monthly application rate time series from 409 Flor ida accounts within 4 profile groups, using data provided by Whitcomb (2005) ................................ ................. 82 4 3 Average indoor and outdoor water use for 1,402 dual metered (left figure) and 29,504 single metered (right f igure) residential accounts in the Alachua County Study area ................................ ................................ ................................ ........................... 83 4 4 Comparison of Alachua County customer with and without sprinklers ............................ 84 4 5 Alachua County study area water use compared to occurrence of in ground irrigation systems from 1960 to 2007 ................................ ................................ ................................ 85 4 6 Alachua County study area average peak month and average annual water use compared to cumulative occurrence of in ground irrigation systems from 1960 to 2007 ................................ ................................ ................................ ................................ .... 86 4 7 Trend of percent homes with irrigation to annual average GPAD ................................ .... 87 4 8 CDF and PDF of annual rainfall data from the NOAA site at Gainesville Regional Airport ................................ ................................ ................................ ................................ 88 5 1 Cumulative frequency distributions of total water use f or 29,507 single and 1,403 dual metered SFR accounts in the GRU service area ................................ ...................... 107 5 2 Scatter plot of the regular metered flow from the dual metered accounts. An extruded view details the accou nts that are expected to represent the distribution of actual indoor use in the area ................................ ................................ ................................ ....... 108 5 3 Frequency histogram of the total annual averages (x1) and peak months flows (x2) of all accounts in the Alachua County study area. Red indicates frequencies between 0 and 49; orange indicates frequencies between 50 and 499; green indicates frequencies of at least 500 ................................ ................................ ................................ 109 5 4 K means cluster s for the overall dataset showing scatter and marginal distributions of x 1 and x 2 for each cluster: 1) Minimal/Offline, 2) Mid range, and 3) Upper .................. 110 5 5 Discriminant analysis utilizing dual metered accounts as training data .......................... 111
11 5 6 Discriminant analysis applied to the 30,910 clustered users in the Alachua County study area ................................ ................................ ................................ ......................... 112 5 7 Regression of estimated irrigation water use in gpad verses discriminant score ............. 113
12 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of th e Requirements for the Master of Engineering TRENDS IN IRRIGATION USE OF POTABLE WATER BY THE SINGLE FAMILY RESIDENTIAL SECTOR IN FLORIDA By John Edward Palenchar December 2009 Chair: James P. Heaney Major: Environmental Engineering Sciences In 2005 p ublic water supply in Florida accoun ted for an average 2.5 billion gallons per day of fresh water withdrawals. A large percentage of this water is used to irrigat e urban landscapes The purpose of this study is to develop methods for estimating average an d peak water use rates for outdoor water use in urban areas. These estimates are used to support the development of water conservation planning methodologies that are being developed by the Conserve Florida Water Clearinghouse. This study uses locally ava ilable and statewide available data to show how trends in home construction and in ground irrigation have led to higher water use in single family residences (SFR) built after the mid 1980 s. Irrigation water use coefficients are presented as inches of wate r per unit of irrigable area for three classes of users: accounts with in ground irrigation system s that are reported to the property appraiser, accounts without sprinklers, and accounts with separate irrigation meters. Additionally, SFR accounts that subs titute potable water with alternative sources was estimated for the study area. was used to extract detailed information on irrigable area, impervious area, pervious area, and the presence of in ground irrig ation in each SFR parcel. This information was used in conjunction with a novel k means
13 clustering approach for hydrograph separation of indoor base and seasonal irrigation water flow based on the peak and average of single metered monthly billing recor ds from October 2007 to September 2008 The disaggregated irrigation use was then compared using property appraiser groupings denoting in ground irrigation and year built. A k means clustering algorithm was used to determine whether irrigators use potable water or irrigate with a substitute water source. An application rate was calculated for the accounts that use the potable system to irrigate. Results for Alachua County confirm previous studies, which show that indoor water use can be estimated as a stead y state rate with an average daily flow of 65 gallons per person per day Additionally, i rrigation is a variable rate with an average peak month use of 684 gallons per day per account and an annual average use of 320 gallons per day per account from Octobe r 2007 to September 2008, for the subset of customers that irrigate The median irrigable area for homes using the potable system for irrigation was 10,383 square feet. Thus, the expected water use coefficients for the Gainesville observations, are: an an nual average of 1. 5 inches per month per square foot of irrigable area and a peak rate of 3. 1 inches per month per square foot of irrigable ar ea. The occurrence of in ground sprinkler systems in new homes has increase d from o 80% of home s built in 2007. In ground irrigation homes make up 27% of the cumulative housing stock as of 2007 This cumulative percentage is expected to increase if the popularity of in ground irrigation remains at current levels For the SFR accounts i it is estimated that at least 44% of accounts with in ground irrigation do not use water from the potable system for irrigation but have substitute sources such as: wastewater reuse, private wells, and/or surface water sources. It is also estimated that about 2 0 % of the accounts without in ground irrigation systems do irrigate. The resulting weighted percentage of accounts using potable water for irrigation is 3 0 %.
14 CHAPTER 1 INTRODUCTION Problem The traditional means of accommodatin g the water use demands in Florida by tapping low cost ground and surface water sources is less viable for meeting future demands. Alternative sources such as desalination, wastewater reuse, aquifer storage and recovery are needed to meet future demands. Water conservation is most critical in water use caution areas where available supplies are limited and come at high costs. Water conservation can be an effective way to reduce average and peak demands and lessen the need to expand the water supp ly systems (Dziegielewski et al. 1993). The standard supply projection assumes that all water needs of the area can be described in terms of the residential population. Future average water use is determined by multiplying population projections by a gall ons per capita per day coefficient (gpcd), derived from historical metered data (Dzurik 2003). Peak water use is estimated by multiplying average water use by a peak to average ratio or peaking factor. This approach assumes that the gpcd will remain consta nt over the planning horizon, i.e., the physical (irrigated area, fixture efficiency, etc.) and economic (cost of water, etc.) drivers of water use for that population remain constant (Dzurik 2003; Lewis et al. 1981; Mays 1996). Additionally, it assumes th at the water use for all sectors is proportional to the residential population (FDEP 2008b; Florida. Dept. of Community Affairs 1989; Gonzalez and Yingling 2008; Mays 2002; Northwest Florida Water Management District 2007) Capital expansion costs are a ma jor component of water supply plans. The need for capacity or storage expansions is dictated by peak usage (Mays 2002) Peak monthly demands in
15 Florida usually occur in April or May due to a combination of lower precipitation and higher evapotranspiration (ET) I r rigation accounts for nearly one third of all residential water use in the U.S. and this percentage increases in warmer climates (Mayer and DeOreo 1999; Vickers 2001). The single most significant driver of peak seasonal demand in public water supp ly is irrigation (Chesnutt et al. 2004; Dziegielewski et al. 1993; Marella 2004; Mayer and DeOreo 1999; Mays 2002; Vickers 2001; Whitcomb 2006) Thus, i n the creation of a demand management or conservation plan outdoor water users with higher average and peak water demands should be targeted(Beecher et al. 2001; Chesnutt et al. 2004) Outdoor water use is an important, and increasing, component of average and peak water use in the single family residential ( SFR ) sector. This outdoor water use is primaril y for landscape irrigation, the topic of this study The purpose of this study is to develop methods for estimating average and peak water use rates for outdoor water use in urban areas. These estimates are used to support the development of water conserv ation planning methodologies that are being developed by the Conserve Florida Water Clearinghouse. Proposed Approach A parcel level land use and water budget will be used to analyze irrigation uses and trends in the SFR sector. Landscape i rrigation water u se can be estimated as the product of the irrigated area ( typically reported as square feet) and application rate ( typically reported as inches per month) as shown in equation 1 1. (1 1) Where : q is the irrigation water use,
16 n is th e count of irrigators on the system, A i is the irrigated area of customer i, and AR i is the irrigation application rate of customer i. T he most important drivers of irrigation use are the area to be irrigated and the soil water moisture deficit. Given a soil moisture deficit, the irrigation rate is the amount of water that needs to be applied to reduce or eliminate this moisture deficit. This number can be converted to an application rate such as inches per year, as shown in equation 1 2 a standard metric in this type of analysis (Lee et al. 2005; Mayer et al. 2009; Mayer and DeOreo 1999; Sample and Heaney 2006) AR=V*1.604/ A i ( 1 2 ) Where A R =application rate, inches per month of water applied to the irrigated area, V = m onthly average o utdoor use in g allons 1.604 = constant conversion to give application rate in inches. A i = irrigated area in square feet (ft 2 ). This study will center on estimat ing irrigated area (A i ) as the irrigable area of the parcel and applying a reasonable irrigation application rate AR. Th is will be done with generally e detailed analysis using a case study to fill in the unknowns. Finally, a clustering algorithm will be used to separate users and determine the number of account s that use the potable water system for irrigation purposes The major thrust of this study is to develop irrigation coefficients and show use trends that support water conservation planning methodologies Models being developed by the Conserve Florida Wat er Clearinghouse ( http://ConserveFloridaWater.org a c c essed on December 2009 ) use standardized activity coefficients to estimate water use for each sector This estimated water use is then calibrated against some measure of known wate r use for a water supplier.
17 Historical data on water supply and water billing data by sector is a first and critical element in developing a reliable and comprehensive water conservation plan. The next step is to disaggregate the larger users of water int o sectors The most accurate methods to perform disaggregate analysis of water users is to link the billing data to the physical and socioeconomic attributes associated with each billed entity. Significant effort may be needed to make this connection if th e billing database was not created with the ability to link the physical location to the flow for that location. Without this link, it is necessary to rely on comparable data from other utilities. Location information on the physical and socioeconomic att ributes of a Florida user can be taken from the US Census Bureau, Florida Department of Revenue (FDOR) and county property appraisers. The census provides decennial housing, economic and population data at various levels of aggregation down to the census block. The FDOR supplies annual data on individual parcel and selected improvement characteristics. Similar data can be obtained from county property appraisers but this data differs from county to county. data can have si gnificantly more detail but does not have consistently defined fields. T he FDOR data contains consistent ly defined fields for the entire s tate of Florida. This FDOR dataset is assumed to be accurate since it serve s as the basis for property tax assessmen ts. Both the count y and the FDOR databases are updated annually and are publically available (FDOR 2009) The data available from FDOR is public and can be downloaded via the internet. The pro ot always downloadable and service charges often apply to obtain this data The FDOR offers the ability to group parcels based on land use type using the two digit FDOR use codes. For example the use code for SFR parcels is 01. By leveraging the FDOR data base statewide methodologies can be
18 developed for water conservation planning macro level data, such as monthly water production provided by Florida Department of Environmental Protection (FDEP), compr ehensive data driven conservation strategies can be developed Utilizing this micro level data t he remaining chapters will show : An estimation of a standardized parcel level irrigable area from statewide Florida data sources. As well as relationships, ti me series, and trends highlighting pervious and impervious area components. The c alc ulation of peak month and annual average water use along with discussion of the relevant trends and related customer attributes within the service area. A k means clusteri ng of peak month and annual average water use is used to d etermin e the accounts that utilize the potable system for irrigation purpo ses, their peak and average application rates, the savings potential, and a discriminant analysis rule to extrapolate these results to other utilities
19 CHAPTER 2 FLORIDA DATA SOURCES FOR WATER USE ANALYSIS Introduction The purpose of this chapter is to describe the following primary data sources that are used to analyze urban water use patterns Florida is unusual, if not uniq ue, in providing public access to important information for water supply assessments The databases used for this study are described in the next sections. FDEP water production and facility data All utilities in Florida are required to submit monthly oper ating reports (MORs) to the Florida Department of Environmental Protection (FDEP). Selected data from these MORs for each water treatment plant (WTP) have been published annually since 1999 on the FDEP web site ( http://www.dep.state.fl.us/water/drinkingwater/download.htm a c c essed on December 2009 supply identification number (PWS ID). Each water system in the database is identified by a unique seven digit number. For instance, the Murphree Water Treatment Plant in Gainesville is identified by the number 2010946. The first number of the identifier, "2", tells that the water plant lies in FDEP District 2, which is in Jac ksonville. The next two numbers, "01", identifies the county where the water plant is located. In this instance the county is Alachua. The last four numbers identify the specific water system within the county and DEP district. The Basic Facility Report ( BFR) contains the following relevant attributes: Population served Sells to population Design capacity, gal./day Service connections
20 The BFR report only contains the above information for the current year. It does not provide this information for previou s years. The MOR monthly flows are provided as yearly downloads onto MS Excel spreadsheets. The user must download the data one year at a time and then combine it into a single data set. For the purpose of this analysis the total available data was downl oaded and compiled into a single data set. This was performed in MS Access and is stored locally on a MS SQL Server database. Treated potable water flows for the S tate of Florida are shown in figure 2 1 On average Florida public water suppliers produce d 2,595 million gallons of water a day (MGD) in 2008. The maximum day flow is 3,902 MGD and the monthly peak of the average daily flows (ADF) was 3,022 MGD in 2008 Peak flow is reported as the peak days flow in gallons for the indicated month. The data ar e presented as spreadsheet files. Key attributes for our purposes are: Monthly average water delivered Peak monthly average water delivered Peak daily water delivered The rated capacity of the water treatment plant The FDEP monthly water use data need s to be checked for errors and some data are missing. Other errors observed include reporting conflicting units of gallons or million gallons per day. FAWN and NOAA /NCDC Climatic D ata Climatic data is needed to evaluate the relationship between climate and outdoor water use Fifteen minute p recipitation and evapotranspiration (ET) data in Florida can be acqui red from the ( http://fawn.ifas .ufl.edu/data/reports/ a c c essed on December 2009 ) FAWN reports an ET value that can be used as a measure of irrigation need in conjunction with a soil moisture water
21 budgeting model (Allen et al 2005; Jensen et al. 1990; Kirkham 2005; Smajstrla 1997) FAWN data are available at 35 stations in Florida. The National Oceanic and Atmospheric Administration (NOAA) produce climate reports through the National Climatic Data Center (NCDC). The NCDC precipitation data are available at many stations in Florida at 15 minute to one hour inte rvals and reporting accuracy of 0.01 or 0.1 inches. The duration of the FAWN and NCDC databases varies from station to station. FDOR Parcel Data Parcel level information can be used to characterize individual accounts. The FDOR database is a statewide and generally available data set used to do much of this work. This database is available at http://dor.myflorida.com/dor/property/ a c c essed on December 2009 The relevant data from FDOR comes in two forms ; a Name Address Legal (N AL) file which is a comma delimited text denoted by the *.csv extension, and a parcel file which is a spatial shapefile denoted by the *.shp extension. The NAL files are large and require either MS Excel 2007 or database software (SQL, MS Access, etc... ). The parcel files require Geographic Information System (GIS) software such as ESRI ArcGIS. The FDOR also provides the spatial geometry of the United States Census block groups with relational identifiers attached but no relevant data for analysis. A schematic representation of the database relationships for the FDOR database is shown in figure 2 2 The attributes used in this analysis of irrigation are: Parcel identification number labeled in the NAL file as (PARCELID) DOR use code labeled in the NAL file as (DORUC) E ffective year built labeled in the NAL file as (EFFYR) Effective area labeled in the NAL file as (TOTLAREA).
22 Other fields may be applied in different types of analyses such as: Just Value (JV), number of residential units (NORESUNTS), an d number of buildings (NOBULDNG). The spatial GIS data is used to calculate parcel area s The NAL table also contains a field that indicates the reported square footage of the parcel titled (SQFOOT). This field is sparsely populated within the database and so the GIS calculated area is preferred. GIS software must be used i n order to work with the spatial data. Commercial software is available from ESRI and other companies Free open source software such as MapWindow GIS downloadable from http://www.mapwindow.org/ a c c essed on December 2009 offers the ability to calculate the area of parcels quickly and easily. Once the area is calculated the data can then be linked back to the NAL attributes using the one to one relational identity PARCEL NO to PARCELID as shown in the FDOR data schematic ( figure 2 2 ) U S Census Bureau Block Level Data The final piece of statewide data needed for parcel level water use analysis is household size. Residential indoor water use can be estimated relatively ea sily for a utility service area if the average household size (average people per house) and the number of residential units served are known (Friedman 2009; Mayer and DeOreo 1999) Knowing the indoor water use the outdoor water use can be calculated as t otal water use minus indoor water uses. The U.S. Census Bureau provides data every decade for average household size. Census data is provided at many levels of aggregation. The smallest spatial element i s the Census block. Using relational identities provi ded with the data it is possible to aggregate it by b lock g roup, t ract, t raffic a nalysis z one (TAZ), z ip c ode t abulation a rea (ZCTA), or c ounty This data is available in GIS form so that it can be combined and manipulated in reference to other spatially aware datasets such as FDOR parcels (US Census Bureau 2000) Census blocks contain fields that indicate a variety of housing and demographic data. Census block data can be assigned to
23 each parcel to create a specialized data set for a utility. This is done using GIS techniques that assign the census data from each block to parcels located within that block. Census block data is less spatially accurate than parcels. To minimize error in misplacement of parcels the data is assigned based on the location of t he geometric centroid of the parcels. Thus, the centroid of the parcel is assigned the attributes of the census block in which it resides Data Parcel level data from county property appraisers can add significantly to t he characteristics available to describ e the indoor and outdoor water use of customer accounts. A Though all county property appraisers are required by the State of Florid a to maintain certain attributes the actual data maintained and the database structure varies widely f ro m county to county. The ACPA data is available to download in shapefile and tabular form http://arrow.acpafl.org/ a c c essed on December 2009 This is common practice for appraisal data. The shapefile contains selected joins of the attribute data pre The parcel area was calculated in the shapefile as square meters which can be easily converted to square feet. In addition to the shapefile data, various other tables are available online as MS Access extracts from the appraisers Oracle server. The tables of interest for land and water use analysis are building area (BLDGAREA), t he public building attributes (BLDGINGPUB), and the public miscellaneous features of the property (MISCPUB). The BLDGAREA table contains a breakdown of various sub areas of the building improvements on the property. For example BAS is listed as the base a rea of the structure this is typically equivalent to the heated or living area of a SFR structure. The BLDGINGPUB table is a summary of building attributes and contains the majority of data needed for analysis. Key fields
24 in this table include: the effecti ve year built (EFFYR), the number of bedrooms (BEDS), the number of bathrooms (BATHS), the number of stories (STORIES), the total or gross area (TOTAREA), and the heated or living area (HTDAREA). The MISCPUB table is the most specific table in the database The content of a miscellaneous features table has been observed to vary greatly from county to county. This table is quite extensive in the ACPA database. The degree to which fields can be assumed accurate varies and the ACPA makes no claim as to the val idity of the database. It was assumed that fields typically used in property valuation would be valid for this study. Each miscellaneous use for a property is associated with a value (UNITS). For uses that have an area such as drive and walkways (DRIVE/WAL K) the UNITS field is populated as square feet. For uses that have no area such as sprinkler systems (SPR SYSTEM) the UNITS field should shows a 1 if the property has the feature and a blank if it does not. This is not always true. In some cases the valu es are other than 1 or blank. These may be ignored as errors likely in entry. The key attribute for this study is SPR SYSTEM that tells us which of the SFRs have in ground sprinklers. Water Management District Service Utility Service Area Boundaries The GIS shapefiles of utility service area boundaries are available from three Florida water management districts. These shapefiles for the Southwest Florida Water Management District ( SWFWMD ) the St. Johns River Water Management District ( SJRWMD ) and the South Florida Water Management District ( SFWMD ) can be downloaded at the web links : http://www.swfwmd.state.fl.us/data/gis/ a c c essed on December 2009 for SWFWMD, http://www.sjrwmd.com/gisdevelopment/docs/themes.html a c c essed on December 2009 for SJRWMD, and http://my.sfwmd.gov/gisapps/sfwmdxwebdc/dataview.asp a c c essed on December 2009 for SFWMD. These shapefiles provide the basis for assigni ng a parcel to a utility To
25 perform this query a GIS operation known as a spatial join can be utilized based on the location of the parcel. Summary and Conclusions This chapter described the key publically available databases that are used to support urba n water supply evaluations. The next chapter describes how irr igable area of these parcels is determined.
26 Figure 2 1 Florida's publicly supplied water production fr o m January 1999 through Dec ember of 2008, as reported by Florida Department of Environme ntal Protection
27 Figure 2 2. Schematic Representation of t able and GIS relationships for Florida Department of Revenue provided data 2009
28 CHAPTER 3 DETERMINATION OF IRRIGA BLE AREA FOR SINGLE FAMILY RESIDENCES Introduction The purpose of this chapter is t o describe procedures used to estimate the irrigable area for single family residences (SFRs) using the databases that were presented in chapter 2. Irrigable area of the parcel is defined as the portion of the privately owned parcel which has the potentia l to support vegetative landscape and thus be irrigated. It was assumed that irrigable area is equivalent to the pervious area of the private lot. Irrigated area is the actual area irrigated by the utility customer. Irrigated area is difficult and expens ive to determine for every customer in a utility. Thus, irrigable are a will be used in this study. Methods to Estimate Irriga ble Landscape Area Terminology The total area of a SFR parcel is the sum of the impervious area, pervious area, and non applicable area, or TA = IA+PA + N A ( 3 1 ) Where T A = total parcel area (ft 2 ) I A = parcel impervious area (ft 2 ) P A = parcel pervious area (ft 2 ) and N A = non applicable or other area (ft 2 ) ( lakes, wetlands, etc .. .) TA is calculated using GIS parcel geometry provided by FDOR N A is estimated from landuse maps and may not be an important component in the land budget depending on how parcels are delineated. PA is the calculated residual pervious area and is equivalent to the irrigable area of a parcel. Impervio us area (IA) is calculated as follows: IA = FS+AIA (3 2 ) Where
29 FS = footprint of the heated and unheated portions of the structure on the parcel in ft 2 and AIA= the associated impervious areas on the parcel in ft 2 ( FS is calculate d based on reported values for the gross area (GA) of the structure and the heated area (HA) of the structure or using derived ratio to the effective area (EA) The GA and HA values are reported by the county property appraiser in many cases and are physic al measures of area (ft 2 ). EA is a n onphysical measure of area reported by the Florida Department of Revenue (FDOR) and can be used to estimate FS as will be shown GA is the gross area defined by Alachua County Property Appraiser (ACPA) as the sum of all improvement areas contiguous with the heated or living area (HA) on a parcel. For mobile homes GA is equivalent to HA because by definition a mobile home is mobile and therefore cannot be contiguous with any other areas on the parcel. In this case, for exa mple, a porch must be accounted for as an AIA. AIA is the sum of all impervious areas not included in the GA by ACPA. There are about 35 areas that are reported by ACPA that are included in the AIA calculation as will be shown Irrigable area can be assum ed to be any applicable pervious area in the parcel boundary. Applicable areas include all areas of the parcel not submerged or specifically set aside as undevelopable easements or setback. Actual irrigated area can be measured on site by running the irrig ation system and measuring the application area. The actual irrigated area may exceed the irrigable area if sprinklers are set to apply water to adjacent impervious areas such as driveways, sidewalks, etc. Actual irrigated area can also include adjacent r ight of way area. On site direct measurement of irrigated area is quite expensive and is impractical for widespread application. Thus, irrigable area, as defined by the pervious area in equation 3 1 will be used in this thesis.
30 Measurement Methods Irrigab le area is often more accessible than measuring actual irrigated area either through GIS techniques, through estimatio ns based on tax assessors data or a combination of the t wo With the recent popularity of web base d geographic information system (GIS) a pplications such as Google Earth high resolution imagery is available for nearly all of the populated areas in the United States (Peng and Tsou 2003) The irrigable areas can be digitized (drawn) over aerial imagery T hus all the impervious area can be a ccounted for if it is not hidden by tree cover (Milesi et al 2005) Of course this method requires a high level of effort for large data sets but it is currently the most accessible means of estimating detailed irrigable areas. From the macro level ligh t intensity has been used to estimate urban imperviousness B y utilizing this and a selected sample of aerial photos a predictive ratio between impervious surface area and lawn area has been developed (Milesi et al 2005) though this is not applicable at t he SFR parcel level. Upcoming methods exist that utilize a combination of parcel geometry, multi spectrum aerial imagery and high frequency elevation data from li ght d etection a nd r anging (LIDAR) techniques ( Zhou and Troy 2008) These methods require highl y detailed imagery and LIDAR in order to classify parcel sub areas accurately. As the national LIDAR database is developed this may prove to be an efficient and even more accurate method to determine actual irrigated area and differentiate between turf, sh rub, and tree cover Previous Studies Based on a national study of water use in 1,200 SFRs across the United States, Mayer and DeOreo (1999) recommend that irrigable area should be used instead of irrigated area. Irrigable area wa s defined by Mayer and De Oreo (1999) garage, driveway, sidewalk, or other impermeable material; it is the area which could support
31 o f th is definition, Mayer and DeOreo (1999) point out that irrigable area is a function of measurable imperviousness while irriga ted area i ncludes operational and irrigation design decisions that may apply water to impervious and off site areas Mayer and DeOreo (1999) calculated irrigab1e area as the lot size minus the building footprint and associated impervious area They estimated that non irrigable areas such as driveways and sidewalks to be 7.5 % of the total lot size. According to the 1999 REUWS datab ase provided by Aquacraft Inc. the average irrigable area of the 1,130 REUWS study homes was 8,266 square feet (sq ft ) and the median was 5,576 sq ft. Statistics from the 1999 study are shown in table 3 1. D ifferent methods of calculation were used in these studies depending on data availability and scope of the study. T he California Single Family Water Use Efficiency Study (CSFWUES) ( DeOreo et al 2008 ) provide s the best detail showing irrigated area and lot size. This study also has the smallest sam ple size. This sample shows that rough estimate of the irrigated area is about 50% of the parcel area for parcels smaller than a quarter acre. The other studies discuss areas in terms of more accessible measures of area such as landscape and irrigable area s. In DeOreo et al. (2008), eight detailed parcel sub areas were calculated using GIS digitization techniques over aerial imagery. The eight sub areas are : turf, non turf plants, xeriscape (irrigated), xeriscape (non irrigated), hardscape, pool, vegetable garden, and house footprint Th is study analyzed outdoor water use in 243 out of 343 homes in the northern California study group. The average parcel in the group was 8,060 s q ft. The average irrigable area was reported as 4,165 sq ft. The median values w ere reported below the mean which indicates that large areas were skewing the mean. Statistics are shown in table 3 1.
32 Mayer et al. (2009) performed on the landscape area of 1,987 residential smart c ontroller study sites. Statistics are shown in table 3 1. Some of the sites were measured physically or using GIS and digital imagery, but many were obtained from tax assessor records. In this study the researche r s opted for the term landscape area and off include all parts of a property that are landscaped or landscapable and are or could be irrigated. This measurement is alternatively known as the irrigated area or the irrigable area. definition seems somewhat ambiguous and as a result irrigable area will be applied in this study. Much work has been done to estimate impervious and pervious urban land areas to evaluate stormwater runoff. ( Heaney et al ., 1999 ; Lee and Heaney 2003 ) For stormwater analysis, the SFR parcel level analysis includes the private parcel lot and the associated right of way (ROW) The lot portion of the area is divided into the following components: House Garage Part of driveway Yard Walkway to dwelling unit P ool Deck/shed The ROW portion of the area is divided into the following components: One half of street consisting of driving and parking lanes Curb and gutter, part of which is used as part of the parking lane Pervious area between curb and sidewalk Sidew alk Pervious area between sidewalk and property line. One half of an alley in some neighborhoods Part of driveway Literature estimates of impervious and pervious areas are available from the stormwater literature. However, these areas are for the sum of the lot and the ROW. For water use analysis,
33 the primary interest is in the lot area only plus that portion of the ROW area that is irrigated by the parcel occupant. Numerous efforts have been made to estimate irrigable and irrigated area using GIS. Howev er, these estimates are of limited accuracy at the parcel level of aggregation. Thus, a different, data driven, approach is evaluated as part of this study. The shift is to use the county and state property appraisal databases to estimate irrigable area. The two primary data (CPA) databases to make these estimates. The FDOR database is available for every parcel in the state of Florida whereas the CPA database varies in content and availability from county to county. The basic method is to estimate irrigable area as the difference between total and This direct measurement of impervious areas and l ot area provided by the CPA enables consistent relationships to be developed for the statewide FDOR database. This is a breakthrough giving the ability to efficiently analyze the impervious, pervious, and irrigable areas of millions of SFR parcels in the s tate of Florida. Irrigable Area Using a Parcel Area Budget Introduction This analysis of urban land use, as it relates to irrigation water use, has been done in cooperation with Gainesville Regional Utilities (GRU). A complete set of SFR billing data tagg ed with the county property appraisers ID was acquired for the analysis. Because of error in the database only a limited period was able to be used from October 2007 to September 2008. Within each lot a general SFR land use budget was assumed, as shown in figure 3 1 The irrigable area of a lot and the land use budget that are defined in literature (Field et al. 2000; Heaney et al. 1999; Lee and Heaney 2003; Mayer and DeOreo 1999) form the basis for developing this budget. A nuance addressed by Lee and Hean ey ( 2003 ) (in the context of storm
34 water) is the occurrence of areas within the parcel boundary that are not truly part of the private lot e.g. (conservation easement and waterway setbacks). These areas are essentially non applicable areas. Pool area is ad dressed separately in the budget because it is driven similar to irrigation but is more closely tied to indoor use. The land use budget can be simplified and described using mass balance equations as shown in equations 3 1 and 3 2 The applicable portion o f the lot or total area is the total parcel area (TA) minus non applicable areas ( NA ). In this chapter the analysis will illustrate how to estimate irrigable area by first determining the lot area from the various measures and then subtracting the impervio us components. The original dataset provided by GRU was composed of 35,768 residential metered customers. Of these customers 1,495 were dual metered residential units which have both an irrigation billing meter and a regular billing meter, as shown in figu re 3 2 Multiple data sets, shown in table 3 2, were acquired to perform analyses on the study area. In addition to the residential billing data, 2008 parcel level data (ACPA 2009) was obtained from the Alachua County Property Appraiser ( ACP A ) and similar data was obtained from the Florida Department of Revenue (FDOR). Of the 35,768 residential metered customers, occurrences of many to one and one to many relationships was a problem when the billing data tables were joined to both the FDOR (77,560 records containing improvements) and the ACPA data tables (86,992 card re cords containing improvements). Many of the billing records queried returned multiple sequence numbers or cards associated with the parcels. Using the FDOR database multi card entries were e liminated as all cards associated with a parcel must be summed by the property appraisers before they are given to FDOR. This is evident by the fact that FDOR reported 9,432 less records containing improvements. For the purpose of this analysis only singl e card parcels were used as it was
35 un certain if water use was associated with one or multiple cards. When all filter queries were completed on the data sets a total of 30,910 customer billing record s were kept. A summary of how the different FDOR classif ications compared against the utility classifications is shown in table 3 3. This indicates that of the 30,910 billing records carried through the analysis 30,866 were classified as single family residential (SFR). Verification by aerial imagery indicated that the mis classified homes appeared to be SFR so the ACPA classification was considered accurate and all 30,910 accounts were used. After initial queries to clean up the data set 29,507 single meter and 1,403 dual meter customers were carried thru the a nalysis. Of the dual metered customers the majority of them matching FDOR code 001. Of the dual meter customers only two are classified differently by the utility and FDOR O ne was listed as 077 ( Clubs lodges, union halls ) by FDOR and the other was classified as 008 ( Multi family less than 10 units ). Additionally, four of the dual meter customers were not found in the FDOR database. This is probably because the homes acquired the ir certificate of occ upancy too late in 2008 to make it onto the tax role. Overall, this samp le of 30,910 SFR homes with direct measurements of total lot area and total impervious area provides a unique database to evaluate irrigable area in a consistent manner. Total Parcel A rea ( TA ) Total parcel area from county property appraisers has been reported for Florida in water use studies performed by Whitcomb (2005). Whitcomb (2005) has compiled a database of 3,538 representative homes in 16 cities throughout the state as part of a residential water use study. The utilities included in the database are: Escambia, Hillsborough, Indian River, Lakeland, Melbourne, Miami Dade, Ocoee, Palm Beach, Palm Coast, Sarasota, Seminole, Spring Hill, St Petersburg, Tallahassee, Tampa, and Toho. Ho using characteristics were collected from county
36 property appraisers. A major limitation of thi s data is that parcel square footage is a sparsely populated field, and no measure of associated impervious area or footprint of structure is provided. From the Whitcomb database 1,150 or 32.5% of the records contained parcel areas. The median parcel area was 9,910 ft 2 just under acre. The time series of the median parcel area is shown in figure 3 3 The data shows no significant trend (R 2 = 0.0529) for the 1 ,150 parcel areas extracted from the Whitcomb (2005) database. The parcel area for the Alachua County s tudy area was reported by the FDOR. Unfortunately, only 31% of the parcels in all of Alachua County report this value. Fortunately FDOR provides 2008 GI S parcel geometries electronically as shapefiles (.shp) for nearly every parcel in the state of Florida and complete GIS coverage is expected by 2010. Total parcel area for every parcel in this study was calculated from the FDOR parcel geometry using ESRI ArcGIS software The billing data from GRU identified dual and single metered accounts. The ACPA lists the occurrence of sprinkler systems in the MISCPUB table. This is a major value added by utilizing the ACPA along with the component break down of imper viousness. Table 3 4 shows the results of the median parcel areas for the single and dual meter accounts as well as those accounts with and without sprinkler systems listed by the ACPA. Th e dual meter customers have a median parcel area of 19,166 ft 2 or 0.44 acres. The median parcel area for the single meter customers is 10,890 ft 2 or about 0.25 acres. The median parcel area for accounts with sprinkler systems is 14,538 ft 2 This is larger than those without sprinklers but smaller than dual meter accounts Median is the preferred statistic for the group because of the presence of a small number of extreme outliers which skew the arithmetic mean. Based on these
37 statistics dual metered accounts cannot be us ed to represent the general SFR population for this data set The result, shown in table 3 4 clearly confirms the intuitive assumption that customers with irrigation meters and/or irrigation systems typically have larger parcels. Much work on quantifying irrigation use in the past has been performed on du al metered accounts. These typical conservation savings rates are often reported as gallons per account per day ( gpad ) (Andrade and Scott 2002; Burton & Associates 2008; Hazen, and Sawyer 2003) Based simply on the fact that irrigation of larger lots requi res more water than smaller lots, it can be assumed that water use per account is greater for dual meter customers than single metered customers. One minor source of error noted with using parcel area is that the parcel geometry is not always equivalent to the potentially irrigable area This is of particular concern for condomini ums or multi family parcels but it does happen occasionally with single family parcels, as shown in figure 3 4 Only one instance of zero lot line development was contained in thi s data set. The parcels in this sub division were the individual homes, while the landscape was held as common area by the homeowner association (HOA), much like a condo situation. The customers do irrigate their own landscape as evident by the fact that two irrigation meters are present in the sub division. This error affected 232 or 0.79% (232/29,504 *100) of the single meter customers and 2 or 0.14% (2/1,402 *100) of the dual meter customers. The parcels were all about 2,800 ft 2 This is a small error i n the case of the study area but may be more of an issue in other utilities. By l imiting the minimum parcel area to the fifth percentile of parcels negative or zero areas that occur when the impervious area is subtracted from these zero lot line parcels, can be reduced or eliminated. This error is ignored for the study area due to its insignificance.
38 The vast majority of parcels in the study area are like the lots pictured in figures 3 5 and 3 6 Since the general trends in this analysis focus on the med ian as the representative statistic the lower and upper percentile accounts will not affect the result. Both the FDOR and t he ACPA databases report the effective year built for a structure. This date is the actual year the home was built or the year of ma jor renovation. The annual median parcel size varied by effective year built with a slight downward trend in parcel areas since the figure 3 7 and Appendix A The overall median parcel area is about 0.26 acre with 2007 having a medi an area of about 0.20 acres. There is still a significant amount of undeveloped land in the service area boundary and parcels can be assumed to be developed at a rate of between 400 and 600 units per year based on velopment can be expected in the long term to grow at a rate of 80 to 120 acres per year for the SFR sector. The median parcel area can be determined for three period groups: pre 1980, 1980 to 1994 and post 1994 as shown in table 3 5. Impervious Area Compo nents ( IA ) L ot area must be separated from the impervious area to calculate irrigable area The impervious components consist of : 1. Footprint of the residential structure (FS) which includes the, o Heated (climate controlled) or living area footprint (HA), and o The unheated (non climate controlled) footprint. 2. Other associated impervious areas (AIA) such as driveways and walkways The term footprint is used to distinguish the area of the structure that covers the ground from the area available for use. In genera l the footprint of a structure is equivalent to the first floor. In this balance the gross or total area ( G A) of the residential structure is the sum of the heated area (HA) and the unheated area. Unheated area is not typically reported by county
39 propert y appraisers but can be calculated as the simple difference of the gross area and the heated area (Unheated portion of GA = G A HA). Thus the impervious area components can be simplified and calculated following equation 3 2 were FS and AIA can be broken d own further into constituent sub area s Footprint of s tructure ( FS ) In the statewide FDOR database (FDOR 2009) only one measure of the footprint of the structure (FS) associated with a parcel is provided. FDOR reports EA ) that is based on the economic value of the component The field listed in the NAL that applies to structure area is called Total Effective Area (Total Adjusted Area) The FDOR defines this fi el n the property which would be the total of all floors on any multi story buildings and the total of all cards on multi card parcels. This entry should be in square feet (FDOR 2008) Executive Director of Valuation at the ACPA was co ntacted. After this phone meeting the following general definition was developed. Effective a rea is an area of the structure, weight ed by an effective area factor, which takes into account all subareas of the structure. Subareas can be classified as prima ry or secondary. Each subarea is adjusted at the same percentage that the unit cost is adjusted. Primary areas are the interior finished living areas, as determined by a reference standard cost per square foot of construction, for that style and use catego ry. All primary areas by definition have an Effective Area Factor of 1. Secondary areas are defined as an area that has a cost per square foot of construction different from the living area reference standard. All secondary areas have an Effective Area F actor greater or less than 1 (Oliver 2009). By this definition equation 3 3 can be defined that describes how effective area is typically derived.
40 (3 3) Were; Pa i = Primary Areas, Sa i = Secondary Areas, and Eaf i = Effective Area Fac T rue measures of area are usually reported in Florida by the county property appraisers as the sum of sub areas which are the same as the un weighted sub areas used to determine effective area. It is lo gical to assume a strong correlation between these three measures of area. Thus, effective area can be used confidently as a surrogate in calculation of the main structure s footprint. By looking at the times series shown in figure 3 8 it can be seen that all three measures of structure area (heated, effective, and gross area) track each other from year to year. Moreover, each time series shows a correlation from one year to the next with an observably increasing trend. For a single story residence GA is t he sum of the heated and unheated footprint. It is not necessary to include the roof overhang as part of the footprint of the base structure, typical in storm water impervious calculations, as area under the overhang can be and often is landscaped or irrig able. Thus, for a single story structure FS is equivalent to G A; either directly from a EA/HA and G A/HA ratios shown in table 3 6 and the FDOR database These ratios were calculated by dividing the sums of the respective areas within each time period. This analysis needs to be refined to correct for the number of floors in the SFR The FDOR database does not list the number of stories (N) for the structure but the CPA data may have this
41 information This information can be used to relate FS and effective area The Alachua County study area shows some variation in the distribution of number of stories from year to year as shown in figure 3 9 and Appendix B. The year 1984 was the least popular year for one s tory single family homes. Thirty one percent of the homes built in 1984 ha d more than one story. Overall multi story residences are less popular making up only about 14% of the housing stock. Error associated with ignoring the impervious area added by mu lti story residences is minimal but should be taken into consideration if data on number of stories per residence is available. To account for the prevalence of multi story homes footprints of the structures in the study area were estimated using the algo rithm shown in equation 3 4 : FS = TA (HA/N) *(N 1) ; N>1 FS = TA ; N = 1 ( 3 4 ) This algorithm, shown in equation 3 4 takes advantage of heated area, gross area, and number of stories reported by ACPA to su btract all but the estimated first floor of the heated area from the gross area. Some error is introduced in that all floors are assumed the same size for integer number of stories. No accommodation is made to account of two story homes with living area ab ove the garage or unheated area above the first floor, a source of error in this method. For decimal number of stories like 1.5 the algorithm removes only one third of the heated area. Because 1.5 and 2.5 story structures comprise only 2% of the SFRs thi s approximation should be acceptable T he times series of footprint, shown in figure 3 10, follows the same trend and pattern as the other reported measures of structure area. Differences in the time series indicate periods when multi story homes were popu lar. Based on this observation and the compar ative statistics shown in table 3 7 it is appropriate to calculate a FDOR footprint ratio FS/EA The footprint was calculated using the algorithm described earlier. The fifth percentile and ninety fifth percenti le
42 bounds of the calculated footprint shows that homes with sprinklers typically have a greater impervious footprint. Homes with dual meters, a subset of the sprinkler homes, are even larger, making up the majority of the large homes within the study are a. The results of the ratio of effective area to footprint of structure yields equation 3 5 FS = FS/EA EA ; w h ere the ratio FS/EA =1.06 for the overall dataset ( 3 5 ) This varied if accounts had more or less stories as follows: For 1 story FS/EA = 1.15 F or 1.5 story FS/EA = 0.87 For 2 story FS/EA = 0.70, and For 3 story FS/EA = 0.55 Associated i mpervious a rea ( AIA ) To complete the budget associated impervious area (AIA) must next be taken into consideration as indicated by equation 3 2 This value has b een difficult to estimate without labor intensive methods (Lee and Heaney 2003 ) Mayer et al. (1999) estimated that non irrigable areas such as driveways and sidewalks were 7.5% of the total lot size. Various other methods of estimation have been used in t he storm water literature (Field et al. 2000; Heaney et al. 1999; Lee and Heaney 2003; Mayer and DeOreo 1999) as needed in runoff calculations. The ACPA database offers the unique opportunity to use parcel level measurements for AIA. The ACPA database cont ains miscellaneous features associated with the property. Miscellaneous features are defined by ACPA as features not contiguous with the main structure. By this definition a porch or carport, for example, may be reported as part of the total area or as a miscellaneous feature depending on whether the roof of the feature is tied in with the roof of the main house ( i.e. contiguous). The list of impervious miscellaneous features and overall contribution to impervious area is shown in table 3 8
43 It is assume d that the footprint of pool is included in the area reported by screen enclosure for those homes with pools. In order to avoid double counting only the screened enclosure was counted. Pool deck area may be calculated as this area minus the reported pool area. According to the ACPA records 58% of the homes with pools have screened enclosures. About 16% of all homes have pools and the average pool area was 497 sq. ft., as shown in table 3 8 This percentage is decreasing as SFR pools seem to be losing popul arity (Friedman 2009) The average screen enclosure was 2,036 sq. ft. with pool surface accounting for about 25% of that area. Impervious area for pool accounts that did not have screen enclosures is calculated as the total of the reported patio area and t he reported pool area, without subtraction. The average pool area is comparable to national data on pool dimensions, shown in table 3 9 collected from the Residential End Uses of Water Study (Mayer and DeOreo 1999) AIA from drive/walkways, screen enclos ures (including pool area), and patios make up 1,278 ft2 but this value varies widely from one house to the next, as shown by the summary statistics in table 3 8 A ny of the three measures of structure (heated, total, and effective area) are more correlated to AIA than the parcel area which was used in the 1999 REUWS (Mayer et al. 1999) As such the effective area can be used to estimate this value by means of the l inear regression with a zero intercept which is depicted in figure 3 11. Thus, AIA can be described solely in terms of the FDOR reported effective area, as shown in equation 3 12. AIA= EA ; w h ere AIA/EA ) = 0.6 9 for the overall dataset ( 3 6 ) This e quation is valid over the range of data but is most applicable within the region bound ed by the 5 th and 95 th percentile of the data as shown in figure 3 11.
44 Non Applicable Area ( NA ) In order to eliminate parcel sub areas that are not applicable to analysis for irrigation, a distinction between the applicable and non applicable parcel area s must be made following equation 3 1 The issue arises for larger parcels, typically over 1 acre, where a portion of the parcel is a lake, wetland, or forest easement. Thi s error can be reduced or eliminated by utilizing the median for analysis or bounding the lot area to the 95 th percentile of parcels. Without current land use maps there is no way to determine if a parcel contains submerged or some other type of un submer ged conservation area. The only means of clarifying this error is to overlay the parcel geometry on top of current aerial imagery. This can be a time consuming task for 30,9 10 parcels. The method taken in this project was to use existing 2004 land use/cov er GIS shapefiles 2004) These shapefiles are publicly available from SJRWMD or from the Florida Geographic Data Library (FGDL) at the University of Florida. The occurrences of non applicable a rea were checked by spatially joining the parcel to the landuse/cover data. A buffer area was applied to account for inaccuracy in the shapefiles. The results showed parcels that had non applicable areas within the parcels boundary. Non applicable areas we re defined as any area that was not listed in its level1 designation as out of 30, 910 parcels that contained non applicable areas. All but 17 of the parcels containing non irrigable areas within their parcel bou ndary had a ear B uilt prior to 2004. Thus, the majority of parcels showing non applicable areas had been developed after 2004 when the landuse/cover was created. As a result the non applicable area of these parcels only indicates the pre dev elopment classification ( e.g. Agriculture, Wetlands, etc ) Parcels that contained non applicable areas were checked for accuracy using 2007 oblique imagery provided by the GIS Division of the Alachua County Department of Growth
45 Management, under the Al achua County's Board of Commissioners (BoCC) This data can be viewed on line at the following web address: http://maps.alachuacounty.us/geogm/viewer.htm accessed July 2009. Based on this ana lysis, non applicable area was considered insignificant for the service area and is not carried through the analysis. Even with this small sample size, it can be assumed that non applicable areas only need to be considered with lots larger than 1 acre and then only on a county by county basis, depending on how easements are apportioned. This error can be minimized by using the median statistic for groups being examined. Irrigable or Pervious Area ( PA ) C ombining equations 3 1 and 3 2 gives : TA = PA ( FS + AIA ) ( 3 7 ) An annually calculated time series from Alachua County is shown in figure 3 12 The median lot area was relatively constant from 1960 to 1980 but has been decreasing from a 1980 level of about 14,000 ft 2 to 8,300 ft 2 in 2007 a 40% decrease. Pervi ous area is decreasing as a function of decreasing lot sizes and increasing impervious area. Irrigable area is equivalent to pervious area. Thus equation 3 7 may be rewritten as: A irrigable = PA = TA ( FS + AIA ) ( 3 8 ) Total parcel area can be calculated by the FDOR parcel area GIS with the most precision occurring when non applicable area is shown not to be a significant consideration for the service area. Footprint of structure can be estimated from FDOR effective area using equation 3 5 with the most pr ecision occurring when the service area has between 80% and 90% single story homes. Resolution may be added b y incorporating story specific ratios shown following equation 3 5 Associated impervious area can be estimated using equation 3 6 with the most p recision occurring for effective areas between 1,049 ft 2 and 3,717 ft 2 A simplified equation can
46 be made from combining equations 3 8, 3 5, and 3 6 as shown in equation 3 9 because FS and AIA are both estimated as functions of effective area. A Irrigabl e = TA 1.06 EA 0.6 9 EA o r A Irrigable = TA EA (1. 75 ) ( 3 9 ) This equation may be modified to accommodate other SFR areas. This may be done by using the varied FS/EA ratios for number of stories. However this equation should be fairly accurate beca use of the detailed analysis of a large sample of nearly 31,000 SFR parcels. This equation could be modified based on different building trends and the amount of non applicable area included in the parcels. Future work should include a ground truth of the impervious area to provide a measure of error in the estimation. Additionally, work should be performed on more counties in Florida. T he irrigable area may be calculated for each parcel or group of parcels using equation 3 9 and only the reported or measur ed areas or by adapting equation 3 15 which takes advantage of modeled impervious areas and the state FDOR database. The calculation following equation 3 9 has no lower or upper bound and so negative as well as multi acre irrigable areas occur. This is a r esult of zero lot line and non applicable areas, respectively. By assuming positive irrigable area with a maximum value of the 95 th percentile of TA a bound ed version of equation 3 9 can be made: Irrigable area = 0 ; if TA E A (1. 75 ) < 0 Irrigable area = TA EA (1. 75 ) ; if 95 th percentile TA < TA EA (1. 75 ) < 0 (3 10) Irrigable area = 95 th percentile TA ; if TA EA (1. 75 ) > 95 th percentile TA The summary statistics for the groups are shown in table 3 1 0 The mea ns are all larger than the median s indicating a skewed distribution. T he irrigable areas calculated using equation
47 3 10 have the 5 th and 95 th percentile of 986 ft 2 and 38,947 ft 2 respectively. Equation 3 9 is also used to calculate an irrigated area as sh own in table 3 1 0 This is done for each individual parcel by subtracting the calculated FS and the reported AIA from the FDOR calculated lot area. One result of this method is that 1,517 accounts are given an irrigable area less than 986 ft2 and 1,662 acc ounts have irrigable areas greater than 38,947 ft2. From this is can be inferred that using the difference of lot area and impervious area, without bounding upper and low limits of the potentially irrigated area, is likely to misrepresent or miscalculate a reas for many accounts. Overall the result is only a slight shift of the mean to the right but at the individual parcel level it can result in very erroneous estimates of irrigable area. Future work could include a statistically significant sample of this group measured by on site and through GIS techniques. It is recommend ed that irrigable area is bound either as shown in equation 3 1 0 or by using a utility specific upper and lower bound on the individually calculated irrigable areas. This bound can be us ed for analysis of individual parcels. For aggregate analysis the median irrigable area should be used. The median value is stable regardless of whether the data is bound ed or unbound ed dividually calculated irrigable areas. One drawback of using the median statistic is that the median does not satisfy continuity i.e., median*count is not equal to total area of this category. Prevalence of In ground Irrigation Systems Information on the prevalence of in ground sprinkler systems is generally unavailable. Fortunately, t he ACPA lists occurrence of in ground irrigation in the mi scellaneous features of properties table The number and percent of single family residences with in ground sprinkle rs for each year since 1960 in Alachua County Florida is shown in figure 3 1 3 A dramatic increase in the occurrence of in ground sprinklers has occurred during each of the past 30 years from less than 10% in 1968 to 85% in 2007. These results show an ove rall average of 22% of homes with
48 in ground sprinklers as of 2007 but virtual saturation of new homes deploying in ground sprinklers. I n ground irrigation systems began to be installed in more than 10% of new homes in Gainesville beginning in 1980 and this rapid increase has continued unabated to its present level of 85% in 2007. Without in ground systems, customers still used manual irrigation. Several other attributes for single and dual metered customers were determined to test whether dual metered cus tomers are a representative sample of a typical home in the Alachua County s tudy area Table 3 1 1 shows the average household year built and average house value (just value), which were directly based off FDOR data and aggregated for each customer group. The percent in ground irrigation in the table was taken from the ACPA data. Based on the analysis of FDOR and ACPA attributes for the two customer groups, dual metered customers were not representative of the study area. Dual metered customers had newer h omes which were larger in size and worth more than the single metered customers, which are representative of the study area. O nly 80% of dual metered homes were reported to have in ground irrigation when close to 100% of dual metered homes should actually have in ground irrigation market in ground irrigation systems which were not reported to the ACPA. Future work should be performed to evaluate the accuracy of the in ground irrigation tag provided by the ACPA. Based on this analysis, current application rates obtained from the dual meter accounts represent a small upper percentile of the housing stock Figure 3 14 and table 3 1 2 show that the median irrigable area for homes w ith in ground irrigation is larger than the irrigable area for home s without in ground irrigation. T he trend shown in figure 3 16 indicates a decline in irrigable area for both groups since 19 8 0
49 Even though homes with in ground systems have slightly larg er irrigable areas, the area normalized application rate can be used to judge if automatic sprinklers affect outdoor water use. Summary and Conclusions Irrigable area can be estimated from statewide FDOR data. The Alachua County Property Appraisers databa se has been used to add significantly to this data. Leveraging the ACPA and FDOR data for the 30,910 parcels t he irrigable area may be calculated for multiple group s of parcels using equation 3 10 Future work could include a comparison of the standardize d irrigable area to actual irrigated and on site measured irrigable (pervious) areas. Irrigable area = 0 ; if TA EA (1. 75 ) < 0 Irrigable area = TA EA (1. 75 ) ; if 95 th percentile TA < TA EA (1. 75 ) < 0 (3 10) Irriga ble area = 95 th percentile TA ; if TA EA (1. 75 ) > 95 th percentile TA If the number of stories is available the impervious footprint of the structure on the parcel can be calculated using the following ratios: For 1 story FS/EA = 1.15 For 1.5 story FS /EA = 0.87 For 2 story FS/EA = 0.70, and For 3 story FS/EA = 0.55 As the service area has developed the trend o f decreasing parcel area and increasing impervious area can be observed. Since 1960 lots have nearly doubled the percentage that is impervious (figure 3 1 5 ) and have halved the pervious or irrigable area. Equation 1 1 defines irrigation water use as a function of irrigable area and application rate. The irrigable area has been addressed in this chapter. The analysis of water use and irrigation ap plication rates is the focus of the next chapters.
50 Figure 3 1. Parcel area budget Figure 3 2. Overview of Alachua County study area showing spatial distribution of single and dual metered customers
51 Figure 3 3. Time series showing fluctuation in me dian parcel area by year from the Whitcomb (2005) database.
5 2 Figure 3 4. Image of Alachua County, zero lot line, single family parcels. Figure 3 5. Example of median dual metered customer s
53 Figure 3 6. Example of median single metered customer s Figu re 3 7. M oving average trend line showing median parcel areas in the Alachua Coun ty study area from 1960 to 2007
54 Figure 3 8. Time s eries c omparison showing annual average of heated area, effective area, and gross area of the structure in the Alachua Cou nty s tudy area for single family residences built in the indicated year Figure 3 9. Cumulative count of homes built showing dominance of single story homes with a growth in popularity of two story homes.
55 Figure 3 10. Time series showing: annual averag e footprint of structure estimated using property Florida Department of Revenue and annual average stories from Alachua County Property Appraiser Figure 3 11. Linear regression of associated impervious area as a function of effective area for 30,910 parcels the line fit equation is (y = 0.69x) with an R 2 value of 0.60.
56 Figure 3 12. Annually time series of the cumulative components of total parcel area (TA) which are: pervious area (PA), associated impervio us area (AIA), and the footprint of the structure (FS) on the property. Figure 3 13. Percent of single family residences in Alachua County Florida with in ground sprinklers.
57 Figure 3 14. Time series comparing irrigable area for home with and without in ground irrigation systems Figure 3 15. Annual time series showing the change in percent imperviousness of median parcels for the Alachua County study group from 1960 to 2007
58 Table 3 1. Comparison of parcel areas reported in Aqu a craft Inc. reports ( DeOreo et al 2008; Mayer and DeOreo 1999; Mayer et al 2009) Study(area) Mean Median Standard Deviation Min. Max. Count Mayer and DeOreo 1999 (Irrigable sq. ft.) 8,266 5,576 11,308 289 156,345 1,130 DeOreo et al 2008 (Irrigated sq. ft.) 4,165 2,872 4 ,565 327 41,357 243 DeOreo et al 2008 (Lot sq. ft.) 8,060 6,579 6,076 1,263 48,596 243 Mayer et al 2009 (Landscape sq. ft.) 14,706 4,890 41,094 240 472,185 1,987 Table 3 2. Summary of datasets used in the Alachua County study area analysis Data Se t Source Application Year of Data Monthly billing data GRU Metered customer level analysis 2007 2008 Parcel geometry and Name Address Legal (NAL) F DOR Customer level analysis 2008 Parcel subareas and extra features ACPA Customer level analysis: Determin ation of impervious area, In ground irrigation p ool, etc... 2008 Census Block US Census Average household size 2000
59 Table 3 3. Complete table of all GRU parcels that linked to DOR use code DOR code Description Count 00 0 Vacant Residential 1 00 1 Singl e Family 30,866 00 2 Mobile Homes 2 00 7 Miscellaneous Residential 1 00 8 Multi family less than 10 units 22 0 12 Mixed use store and office or store and residential or residential combination 3 0 52 Cropland soil capability Class I1 1 0 54 Timberland site index 90 and above 1 0 61 Grazing land soil capability Class I 2 0 68 Dairies, feed lots 5 0 69 Ornamentals, miscellaneous agricultural 1 0 71 Churches 4 0 77 Clubs, lodges, union halls 1 TOTAL 30,910 Table 3 4. Summary statistics of parcel area s showing single, dual meter groups and sprinkler system groups in Alachua County Group Mean (ft 2 ) Median (ft 2 ) 5 th Percentile(ft 2 ) 95 th Percentile(ft 2 ) Count Dual meter 24,034 19,166 7,950 51,624 1,40 3 Single meter 16,660 10,890 2,800 45,049 29,50 7 Overa ll 16,997 11,214 2,800 45,374 30,910 With sprinkler 22,012 14,538 5,046 56,179 8,304 Without sprinkler 15,158 10,782 2,405 43,395 22,606 Overall 16,997 11,214 2,800 45,374 30,910
60 Table 3 5. Median parcel total area for the three period groups u sed to describe the Alachua Counts Study area. Period Median Parcel ft 2 Count of Parcels Pre 1980 12,300 12,852 1980 94 11,375 10,201 1995 2007 9,438 7,857 Overall 11,214 30,910 Table 3 6. Heated area, effective area, and gross area for three periods along with the ratios that relate these areas to each other Term Pre 1980 1980 to 1994 1995 to 2007 Overall HA 1,655 1,778 2,171 1,827 EA 1,862 2,059 2,542 2,100 GA 2,133 2,350 2,944 2,411 EA/HA 1.12 1.16 1.17 1.15 GA/HA 1.29 1.32 1.36 1.32
61 Table 3 7. Summary statistics comparing various measures of structure area. Area Source Group Mean (ft 2 ) Median (ft 2 ) 5 th Percentile (ft 2 ) 95 th Percentile (ft 2 ) Count FS, Calculated Dual Meter 3,2 57 3,165 1,898 5,070 1,40 3 Single Meter 2,186 2,078 940 3,692 2 9,507 With sprinkler 2,984 2,840 1,818 4,653 8,304 Without sprinkler 1,960 1,898 858 3,205 22,606 Overall 2,235 2,118 956 3,838 30,9 10 HA, ACPA Dual Meter 2, 698 2,563 1,518 4,303 1,40 3 Single Meter 1,786 1,630 953 3,117 29 ,507 With sprinkler 2,444 2,302 1,412 3,999 8,304 Without sprinkler 1,600 1,491 912 2,642 22,606 Overall 1,827 1,660 960 3,241 30,9 10 EA, FDOR Dual Meter 3,14 2 2,976 1,832 5,044 1,40 3 Single Meter 2,050 1,893 1,042 3,579 29 ,507 With sprinkler 2,854 2,686 1,694 4,629 8,304 Without sprinkler 1,823 1,717 980 2,988 22,606 Overall 2,100 1,931 1,049 3,717 30,9 10 GA, ACPA Dual Meter 3,64 3 3,452 2,114 5,840 1,40 3 Single Meter 2,352 2,184 1,138 4,128 29,507 With sprinkler 3,304 3,092 1,966 5,401 8,304 Without sprinkler 2,083 1,974 1,059 3,439 22,606 Overall 2,411 2,228 1,152 4,315 30,9 10
62 Table 3 8. Comparison table showing the contribution of various associated impervious area reported by ACPA Associated Impervious Area (AIA) Overall Area (ft 2 ) Mean (ft 2 ) % Overall Area Count % Occurrence based on 30,910 accounts Barns and Pole Barns 52,752 805 0.1% 66 0.2% Basketball Courts 1,600 1,600 0.0% 1 0.0% Bath Houses 1,036 104 0.0% 10 0.0% Brick Drive/Walkways 96,417 451 0.2% 214 0.7% Bridges 448 448 0.0% 1 0.0% Cabanas 161 81 0.0% 2 0.0% Canopies 556 278 0.0% 2 0.0% Carports 244,189 299 0.6% 816 2.6% Decks 1,325,081 266 3.1% 4,987 16.1% Drive/Walkways General 22,781,132 800 54.0% 28,465 92.1% Enclosed Porches 62,950 245 0.1% 257 0.8% Garages 93,737 499 0.2% 188 0.6% Gazebos 13,575 109 0.0% 125 0.4% Green Houses 39,651 173 0.1% 229 0.7% Kennels 1,833 141 0.0% 13 0.0% Open Porches 278,216 188 0.7% 1,477 4.8% Pati os 5,560,366 392 13.2% 14,171 45.8% Pavilions 2,057 229 0.0% 9 0.0% Paving 290,147 1,649 0.7% 176 0.6% Pools 2,385,177 497 5.7% 4,795 15.5% Pump Houses 1,524 109 0.0% 14 0.0% Ramps 154 154 0.0% 1 0.0% Racquetball Courts 3,120 780 0.0% 4 0.0% Screened Enclosures 6,929,526 2,036 16.4% 3,404 11.0% Screened Porches 907,640 230 2.2% 3,942 12.8% Screened Rooms 5,370 413 0.0% 13 0.0% Sheds 141,799 203 0.3% 700 2.3% Shops 15,248 508 0.0% 30 0.1% Slabs 124,872 187 0.3% 666 2.2% Stables 10,694 1,528 0.0% 7 0.0% Storages 538,515 14 3 1.3% 3,769 12.2% Tennis Courts 269,096 7,081 0.6% 38 0.1%
63 Table 3 9. Average dimensions for a swimming pool based on the observed data Adapted from (Mayer and DeOreo 1999) Study site Count Length (ft) Width (ft) Depth ( ft) Area (ft2) Tampa, FL 59 27.2 15.3 5.2 418 All Swimming Pools 943 30.1 15.7 5.6 489 Table 3 10. Summary statistics comparing parcel area and two estimations of irrigable area between dual meter, single meter, sprinkler, non sprinkler, and overall groupings Area Source Group Mean Median 5 th Percentile 95 th Percentile Count (ft 2 ) (ft 2 ) (ft 2 ) (ft 2 ) Lot FDOR Calculated Dual meter 24,034 19,166 7,950 51,624 1,403 Single meter 16,660 10,890 2,800 45,049 29,507 With sprinkler 22, 012 14,538 5,046 56,179 8,304 Without sprinkler 15,158 10,782 2,405 43,395 22,606 Overall 17,000 11,214 2,800 45,374 30,910 Irrigable, equation 3 10 Dual meter 15,850 12,941 4,642 38,947 1,403 Single meter 11,183 8,338 986 38,94 7 29,507 With sprinkler 13,562 9,590 2,285 38,947 8,304 Without sprinkler 10,599 8,282 986 37,274 22,606 Overall 11,395 8,447 1,047 38,947 30,910 Irrigable, equation 3 9 Dual meter 18,038 12,941 4,642 43,446 1,403 Single meter 13,270 8,338 931 39,539 29,507 With sprinkler 16,876 9,590 2,285 48,183 8,304 Without sprinkler 12,241 8,282 642 37,274 22,606 Overall 13,486 8,447 1,047 39,725 30,910
64 Table 3 11. Summary of attributes for si ngle and dual metered customers for Alachua County area Group Sample Size Avg. house value Avg. Year Built Mean Irrigable ft 2 Median Irrigable ft 2 % In ground Irrigation Single m etered 29,507 $ 190 ,000 1983 11,183 8,338 24% Dual Metered 1,403 $ 346 ,000 19 94 15,850 12,941 80% Table 3 12. Summary of attributes for accounts with and without sprinkler systems tagged by the ACPA Group Sample Size Avg. House Value Avg. Year Built Median Irrigable Area (ft2) Sprinklers 8,304 $302,652 1994 9,590 No Sprinklers 22,606 $158,327 1 980 8,282
65 CHAPTER 4 DETERMINATION OF IRRIGATION WATER USE Background Water demand in the single family residential (SFR) sector is usually the most important component of urban water use comprising 60 to 90 percent of total urban water use (Mayer and DeOreo 1999). A calibrated end use water budget is an essential component of water conservation analysis. End uses can be defined at various levels of disaggregation ranging from single family residential seasonal and non seasonal use, to m ore detailed end users such as seasonal irrigation use by customers with in ground sprinkling systems. The more macro level of disaggregation for the study of irrigation use is the evaluation of irrigation use by SFR customer categories. This study utilize s state and county wide database s to capture not just a sample but the entire population of individual water users wherever possible. This data driven approach enables the analysis to target accounts that share common traits making them the most likely to benefit from a conservation program. This data also supplies a baseline of water use that is vital in assessing the potential and realized saving of any conservation program. T hree SFR customer categories are identified in this chapter : 1) with in ground irrigation systems, 2) without in ground irrigation systems, and 3 ) accounts with separately metered irrigation usage (a sub set of category 1) Ideally, water utility metered billing records provide this information but this is seldom the case. In actua lity, the total water use by a customer is measured periodically, typically monthly. If one is interested in irrigation water use only, then it is necessary to partition total water use into its indoor and outdoor components. M ethods for doing this hydrog raph separation are presented in this chapter A general water budget for a residential account takes the form of equation 4 1 where each end use can be described as a rate
66 having specific coefficients similar to the irrigation rate coefficient AR present ed in equation 4 1. ( 4 1) The focus o f this study is on the irrigation component of outdoor water use. Pool water usually occurs outdoor. The major difference between pool and irrigation use is that the customer needs potable water f or pools (Friedman 2009). In contrast, the customer may irrigate with non potable substitutes from wells, and/or wastewater and stormwater, the focus of the next chapter. This section will focus on robust methods that utilize data driven approaches in dete rmining the effects of in ground irrigation on average and peak seasonal demand within the SFR sector. The selected study area is Gainesville, Florida where water supply is provided by a municipal utility called Gainesville Regional Utility (GRU). The SFR sector consists of 29,504 accounts whose monthly usage is measured with a single meter and 1,402 customers with separate indoor and irrigation meters. One year of monthly water use data was provided by GRU for the period from October, 2007 to September, 20 08. Climate data for this time period will be presented in this chapter. Information regarding all SFR parcels in Alachua County that includes all of GRU is available from the Florida Department of Revenue (FDOR) and the Alachua County Property (ACPA) databases. These property databases were linked to the customer billing data for this study. Previous studies have relied on more expensive survey data or physical measurement of a selected number of customers to obtain the following relevant infor mation: Persons per account Lot area Impervious area Pervious area
67 Irrigated area Presence of an in ground irrigation system Year occupied for account Current assessed value Presence of a swimming pool Number of bathrooms In order to frame th is work, it is necessary to briefly examine some of the body of work on the subject of irrigation application rates in Florida. As billing data is most often collected monthly this time step will be used in this study. Hourly and daily water delivery data are available at the source end of the water system. Results of Previous Studies of Outdoor Water Use University of Florida Studies (Baum 2005; Dukes et al. 2006; Haley et al. 2007) has shown that a variety of application rates are used to irrigate SFR landscapes O nsite measurements were made o f irrigated area and metered irrigation using an unobstructed flow meter after the irrigation line diverged from the house main line. This is an accurate but costly method of determining irri gation application rates. C ustomer s were grouped into three treatments: 1. Treatment one (T1) consisted of existing irrigation systems and typical landscape plantings, where the homeowner controlled the irrigation scheduling. Existing irrigation consiste d of rotary sprinklers and spray heads installed to irrigate both landscape and turfgrass during the same irrigation cycle.
68 2. Treatment two (T2) also consisted of existing irrigation systems and typical landscape plantings, but the irrigation scheduling was based on 60% of net irrigation requirements 3. Treatment three (T3) consisted of an irrigation system designed according to specifications for optimal efficiency including a landscape design that minimized turfgrass and maximized the use of native, d rought tolerant plants (Baum 2005) The sample sizes were large enough to be significant when using two year s of data aggregated on a monthly basis. The samples sizes in each group were small though with: T1 having between 5 and 8 participants depending on the year, T2 having between 6 and 9 participants depending on the year, and T3 having between 4 and 20 participants depending on the year. The summary of the data in table 4 1 shows a range of application rates from 40.7 in/yr for T3 to 69.0 in/yr for T 1. The study collected climate data from weather stations near the sites and calculated ET 0 using the FAO 56 Penman Monteith Equation. The EPA WaterSense program suggests application rates of 7 0% or less compared to ET 0 (US EPA WaterSense, 2008) Thus, th e application rates shown in table 4 1 are relatively high. Peak to average applications ranged from 1.19 for T1, to 1.74 for T3. As can be seen in figure 4 1 seasonal peaks occur twice during the year once in spring and once in fall Haley et al. (200 7) modeled daily theoretical application requ ir ements for the three treatment groups, taking into consideration effective precipitaion and crop coefficents. The results for the study group s indicate that both T1 and T2 irrigated with T1 appling 2.4 times t heir theoretical irrigation requirement and T2 appl y ing 1.7 times their theoretical irrigtion requirement. Thus, the overall pattern may not be representative of irrigators on a potable supply system unless the majority of users are overirrigating.
69 Studies Performed by Aquacraft Inc. According to a recent California study (Mayer et al. 2009) some customers over irrigate but many under irrigate. Mayer et al. (2009) evaluated the effect of weather based smart irrigation controllers in California. Their resul ts indicated that 41.8 % of customer s increased water use after installation of smart controllers designed to meet the landscapes water needs, 56.7% decreased water use, and 1.5% showed no change. Of the sites that decreased water use, only about 50% were irrigating in excess of theoretical requirements before smart controllers were installed. Outdoor u se equals t otal d emand minus i ndoor u se minus l eaks. This simple equation forms the basis of outdoor water use estimates. When some irrigation is expected year round outdoor use can be estimated as total billed minus 70 gpcd multiplied by the number of residents in a household (Friedman 2009 ). Florida Studies by Whitcomb Whitcomb (2005) analyzed SFR water use in 16 Florida cities The database provided by W hitcomb (2005) was examined for irrigation application rates. O nly accounts that have parcel area as a populated field and that the survey indicated irrigation from the potable water system are considered in the analysis. This first query resulted in 522 o f the original 3,538 accounts. Not a ll of the 522 accounts were used because Miami Dade and Tampa billed bi monthly during the study, which would significantly dampen the peak. Several other accounts were eliminated because the billing data did not contain a complete year. The resulting final monthly database consist ed of 409 accounts. The sample sizes by utility are shown in table 4 2. Outdoor use was assumed to be total billed use minus indoor use. No attempt in the database was made to account for establ ishment of new landscape or pool filling from new construction. The Whitcomb (2005) database includes survey results on the number of people in the home. Indoor use was estimated a s the maximum of 70 gpcd times the survey household size
70 (HHS) (Friedman 200 9), or zero. Billing data was taken from 1999 to 2004 in most cases. The monthly data was read date rectified so that the usage was assigned to the calendar month in which it occurred (Dziegielewki and Opitz 2002; Dziegielewski et al. 1993) Rainfall and E T ranged during the period of record f ro m severe drought in 2000 to moderately moist years in 2003 and 2004 ( Whitcomb 2005) As a result the irrigation use should represent an average mix of these climatic conditions. The ET ( N ET) was calculated by Whitcom b using Hargreaves (Hargreaves and Allen 2003) equation for limited data. This equation may introduce error when comparing with standard equations such as FAO 56 Penman Monteith Equation (Allen et al. 2005; Dukes 2009; Gaviln et al. 2006; Triebel 2005) The equation for calculating monthly application rates is shown in equation 4 2 ( 4 2 ) Only the building area was reported by Whitcomb. As a result, the standardized irrigable area presented earlier could not be used directly. In th is case a proxy irrigable area can be calculated as total parcel area minus reported building area. Using equation 4 2 with the proxy irrigable areas calculated as the parcel area minus the building area, monthly average application rates can be calculate d Within the queried data set of 409 po table irrigators the application varied among profiles as shown in table 4 3. cannot be broadly applied in Florida because the proxy irrigable area value used did not represent a consistently defined irrigable area. More importantly, the study showed relatively small samples of complete data from various util ities throughout Florida. Many areas in south Florida were under represented. The monthly
71 peak to average ra tio was stable between profiles, as shown in figure 4 3 It ranged from 1.52 for profile 1 to 1.43 for profile 4. The overall monthly calculated irrigation peak to average ratio is 1.47. Based on the datasets examined it is clear that irrigation use at t he account level varies from no use by some in the winter to over 6 inches per month for others during peak season. The numbers can vary widely on an account by account basis, often with little correlation to drivers such as rainfall or ET (Mayer and DeOr eo 1999; Stadjuhar 1997) I rrigation is a large user of water even if running efficiently. What are the trends in irrigation use and how can these trends be evaluated ? Effect of In Ground Irrigation Water Use in the Alachua County Study Area Motivation Acc ording to the 1999 REUWS home s with in ground irrigation systems used 35% more water than houses without and homes which only used hand (hose) water use 33% less water than those with in ground systems (Mayer and DeOreo 1999). This finding is suspect becau se of the small sample sizes and limited logging periods. Work performed by Stadjuhar ( 1997) in conjunction with the 1999 REUWS extensively examined 88 residential accounts in Denver, Colorado. Stadjuhar (1997) compared in ground irrigators with surface ir rigators. Surface irrigators use hoses and portable sprinklers. Over the four week logging period during June 1996 accounts with in ground irrigation showed higher water use. The study also estimated irrigable area for these accounts. The findings show t hat an increase in the irrigable area of in ground accounts corresponded to about a 30% increase in overall water use. Turf greenness and shaded area were also added as explanatory variable s but were shown to be insignificant in explaining irrigation water use. When an application rate was calculated hose irrigators actually used more water than in ground, as shown in table 4 4.
72 The work performed by Stadjuhar (1997) may not be applicable to Florida conditions because Colorado has distinct growing seasons Critically the logging period was short and only captured one rainfall event The conclusions of these studies show general trends in irrigation water use but do not give the clarity that utilities and planners need to analyze disaggregate uses and tren ds. Common features of these studies are the ir small sample sizes and or short data collection periods T he largest dataset was Whitcomb (2005) sample of 3,538 accounts out of an estimated 6 million SFR accounts in Florida (FDEP 2008a) Surveys associat ed with these studies require significant resources. In order to analyze nearly every residential account in a potable water system, customer billing data need s to be combined with parcel attribute data to do large scale evaluations. Florid a is fortunate in having such information available. These: state, county, and utility databases are the cornerstone of data driven approaches in water use and conservation analysis as will be described below Analysis of SFR Water Use With the cooperation of Gainesvil le Regional Utilities (GRU) one year of monthly water use data (October 2007 to September 2008) was collected for 1,402 dual metered customers and 29,505 single metered customers. These customer level billing data were combined with parcel level customer attribute databases from the Florida Department of Revenue (FDOR) and the Alachua County Property Appraiser (ACPA) Much of the information acquired from the ACPA such as occurrence of in ground irrigation is not realistic to gather for each account The r esult is that other researchers have had to rely on survey data and representative samples of the population. Information on the attributes of these SFR accounts was presented in Chapter 3. This chapter evaluates the associated water use patterns for thes e users based on one year of monthly billing data.
73 SFR w ater u se p atterns A total of 1,402 GRU SFR customers have two meters that record indoor and outdoor use separately Dual metered billing data is important because it isolates indoor and irrigation u sage. The single and dual monthly billing data are reported as integer values in thousands of gallons ( k g als /mo nth ). The major drawback of using dual meter customers to characterize the accounts in the serv ice area is that these accounts are highly unrep resentative of the majority of the SFR accounts Dual meter accounts represent large users with newer expensive home and larger lots. This was shown clearly in tables 3 10 and 3 12. As a result methods must be used that can take advantage of single metere d data. This involves performing hydrograph separation to disaggregate the base indoor flow and the seasonal irrigation flows. W astewater charges are calculated based on billed water use f or all residential regular billed customers within the Alachua Count highest monthly average water usage for the January or February billing periods. The smal k g al or 33 gallons per account per day ( gpad ). Wastewater charges are calculated based on the smaller of either the actual As of October 2008 the char ge for individual accounts wastewater was $ 4.9 0/kgal. Because of the strong financial incentive to minimize water use in January and February in conjunction with the natural minimal irrigation need i n those months the minimum of the months of Jan uary or F eb ruary may closely approximate indoor only use for the single meter customers. Based on work by Friedman (2009) minimum indoor use in the study area is close to 70 gpcd. The minimum month during the period of record was January with a total average use o f 181 gpad for single metered accounts as shown on the right side of figure 4 3 The average
74 household size or people per home (HHS avg ) for these accounts was 2.53. According to Friedman (2009) indoor use should be 70* HHS avg or about 177 gpad A differenc e of only 4 gpad is calculated between this and the minimum month for single metered accounts. By assuming this to represent non seasonal use a monthly patte rn of irrigation water use can be observed. The minimum month and the metered indoor use from dua l meter accounts all tend toward a constant indoor use reported by Friedman (2009), and shown in figure 4 3 N ot all customers without sprinkler systems irrigate by hose or otherwise. Additionally, not all accounts with sprinkler systems necessarily use t hem. What exists in a real system is some mix, the central tendency of which can be observed whe n a large sample of the p opulation is analy zed T hree groups are used as basis of comparison for this study: the in ground irrigation (sprinkler) group tagged by ACPA, those not tagged by ACPA as having in ground irrigation, and accounts from GRU with separate irrigation meters. R egular and irrigation metere d data was summed over the 30,910 accounts. Various attributes associated with the tagged account group s are summarized in table 4 5 Dual meter accounts are a sub set of the in ground sprinkler accounts. Water use for the in ground sprinkler group included dual meter accounts and water use for dual meter accounts is the sum of the regular and irrigation me tered usage. Summarizing table 4 5, the percent total accounts gives the percent of accounts that fall within the respective group. The average gpad is the annual average water use expressed as gallons per account per day. The percentage of each c ontribution to the total water use for the period of record from October, 2007 to September, 2008 is shown by % total annual average use Usage in peak month is the total use for the system peak production month of May divided by the number of day s in May and the number of accounts in the respective group. The
75 % peak month (May) use is similar to % total annual average use except that it is only based on the water use during the peak water production month of May. Persons per house are calculated using the 2000 census block average household size (HHS avg ) for each parcel in the respective groups. Average indoor gpcd is indoor water use normalized by household size and is calculated as the dual metered indoor use divided by the HHS avg This was held constant for all groups following Friedman (2009). Average indoor gpad is the product of gpcd and HHS avg except for the dual meter accounts where it is directly metered. Average irrigation gpad is the average annual irrigation water use expressed as gallons per acc ount per day. This value is calculated as Average gpad minus Average indoor gpad Peak month (May) irrigation gpad is similar to Average irrigation gpad except that it is calculated solely on the peak month of May. Irrigable Area Median ft 2 is the media n irrigable area for the respective group calculated using the technique discussed in detail in this chapter The field AR in./month is the annual average irrigation application rate calculated with Average gpad and Irrigable Area Median ft 2 for each g roup. The field peak AR in./month is similar to the annual average except that it is based solely on the peak month of May. Both Peak AR and the annual average AR are calculated using equation 4 3 : (4 3) Total peaking factor is th e u sage in peak month dived by the annual average gpad. Irrigation peaking factor is calculated similarly except that only the peak month is considered.
76 The fields effective area, effective yr. built, and just value are calculated as the average of each wi th in its respective group. These three fields can be found in the FDOR database. Some key points from table 4 5 are: Only 27% of customers have in ground irrigation systems Single meter accounts with in ground sprinklers use nearly twice as much water as customers without in ground sprinklers Dual meter accounts use nearly th ree times as much water as non sprinkler customers Household size is nearly constant for the three groups. Outdoor water use for homes without sprinklers is only 39 gpad as compared t o 214 and 435 gpad for groups sprinkler and dual meter accounts, respectively. Outdoor water use constitutes the majority of annual water use for the in ground sprinkler and the dual metered group s The in ground sprinkler and the dual metered group s have much newer homes with an average age of 14 years as compared to 28 years the no in ground sprinkler group. The median irrigable areas are 16 and 56% larger for in ground sprinkler and the dual metered group, respectively, as compared to the no in ground sp rinkler group. Application rates calculated based on the billed water use and irrigable area are much lower than expected, based on the studies examined. This is likely a result of many homes in each group which: o d o not utilize the potable water system for irrigation o i rrigate at low er rates, o only irrigate a portion of the irrigable area, or o some combination of the above three factors Dual meter accounts comprise about 80% in ground sprinkler accounts. In other words 13.5% of the in ground sprinkler accoun t s are also dual meter accounts. As can be seen in table 4 5 and figure 4 4 customers with in ground irrigation tend to use significantly more total water on average yet have similar household sizes to the rest of the population. T he total peak demands for the in ground sprinkler accounts are double that of non in ground sprinkler accounts. The total monthly peak to average ratio for sprinkler accounts is about 1.6 and the ratio for the non sprinkler accounts is about 1.4. This is consistent with the mon thly peak to average ratios calculated using the Whitcomb (2005) dataset. The outdoor peak to average ratios tell a different story. When indoor use is removed accounts without in ground systems are more variable than accounts with these systems. This may be due to any number of
77 causes including manual irrigation by the group. The peak monthly and average annual uses for the sprinkler accounts were double that of the non sprinkler accounts. The result is that in ground sprinkler accounts had a monthly sea sonal peak 300 gpad higher than the non sprinkler accounts and that the annual average use was 200 gpad higher t han the non sprinkler accounts. Water use trends To better understand the relationship of water use to the presence of in ground irrigation th e average annual water use was graphed as a time series using the effective year built as the time axis. The percent of home s with in ground irrigation for that year built was graphed on the same time axis. Microsoft Excel was used to fit trend lines to t he data. A clear break point for water use exists in 1985 which corresponds to the same break point in the occurrence of in ground irrigation systems. The results, shown in figure 4 5 and 4 6 indicate an obvious relationship between percent of homes built with sprinkler systems and average annual water use. The total gpad is plotted in figure 4 7 as a function of the % of homes in each year that have in ground irrigation systems using data for effective year built in each year from 1960 to 2007. The result ing regression equation is shown in equation 4 8. Total gpad = 191.2 + 252.88*% in ground irrigation systems ( 4 8) Absent in ground irrigation systems, the total gpad would be 191, close to the previously calculated indoor use of about 180 gpad The per cent in ground systems has grown dramatically from less than 10% in 1985 to about 80% of the new houses in 2007 having in ground systems. Using the regression equation, the gpad for these 2007 houses would be about 393 gpad about twice as much as it was for new homes in 1985. The intercept 191.2 gpad indicates the average annual gpad of the Alachua county system without any in ground irrigation. The indoor gpad for the system can be estimated as 177 gpad
78 (70*2.53). Thus, only about 14 gpad (5.53 gpcd) cannot be explained by the occu rrence of in ground irrigation. Determining Reasonableness of an Irrigable Area Application Rate The irrigation shown for the service area by dual metered and in ground sprinkler accounts is less than that of the studies exa mined. Thus, an analysis of the reasonableness of the calculated application rates is in order. The July 2009 draft version of the U.S. EPA Watersense water budget tool (US EPA Water Sense 2008) will be used to investigate this factor It is available onl ine at: http://www.epa.gov/watersense/specs/homes.htm a c c essed on December 2009 Analysis of irrigation rate for a system begins with placing the period of record on a rainfall distribution. This is needed to determine if the water use calculated is representative of typical conditions. From October, 2007 to September, 2008 total rainfall in the study area was about 43.5 inches or 3.6 in./month. Figure 4 8 shows annual rainfall for the period being examined is just abov e the lower twenty fifth percentile value of 42.5 inches. The result of somewhat low annual rainfall is that the water use during this period may be elevated. This relationship has been observed statewide during the 2000 drought year (Marella 2004) One c aution that must be added to such an assumption is the implementation of drought management strategies often done during low rainfall periods. A primary tool used in drought wn to be successful in reducing water use (Shaw and Maidment 1987; Whitcomb 2006) Thus, application rates calculated during irrigation restriction may result in a false low application rate. The logic being that once the drought is over and the restrictio n are removed most of the accounts will return to previous irrigation practices, which are of a higher use than the restricted practices.
79 For the period of observation, the study area was not under enforced irrigation restrictions. Next, ET 0 was acquired through the FAWN network. The FAWN (FAWN 2009) station closest to the study area is the Alachua station. The ET 0 over the year was 37 inches or 3.1 in./month. Dual meter accounts will be used to analy ze application rate s as it is unknown how many accounts with sprinklers use potable water for irrigation. To use the Water Budget tool the landscape area and average monthly ET 0 is input. The median irrigable area of the dual meter accounts from table 4 5 and the average ET 0 from table 4 6 were used. The peak landscape water allowance (LWA) is 3.78 inches per month. This is simply 70% of the peak use month (May) ET 0 requirements, shown in table 4 6 the default cut off for EPA W ater S ense new home specifications. This gives the upper limit on reasonable peak m onth irrigation. The dual meter peak irrigable application rate, shown in table 4 5 is 2.83 in./month, or about 750 gpad. Finally, using the EPA WaterSense calculator for landscape water requirements (LWR) yields an even higher water use. Knowing that th e May rainfall was 0.6 inches and assuming irrigation of the entire median irrigable area, a landscape coefficient (K L ) of 0.6, and a 70% lower quartile distribution uniformity (DU LQ ); the LWR yielded 1,171 gpad or 4.4 inches during May 2007. Thus, these d ual meter accounts either irrigate much less than the i r irrigable area, water much less during the peak month than the WaterSense LWR or a combination of the two. It can be generally accepted that an irrigable area application rate is less than and canno t be compared against the actual irrigated area application rate. This known the irrigable area application rate is a powerful tool to compare the irrigation use among group s Conclusions A pplication rates can be calculated for various account groups usin g monthly customer billing data. However, except for the dual metered accounts it is impos si ble to determine how
80 many of these accounts irrigate f ro m the potable water system. Accounts that use substitute sources for irrigation drive the billing data calc ulated application rate down and the result is a false low Because of the low cost of self supplied irrigation water, these users of alternative sources may actually use more water than accounts that irrigate with water fr o m the potable distribution syste m (Andrade and Scott 2002). Even if irrigation from the potable systems is known application rates cannot necessarily be compared directly to irrigated area application rates unless lots in a service area are small i.e. (less than a quarter acre). Some c omments about the three groups analyzed are as follows: The no in ground sprinkler group o These accounts do not appear to irrigate extensively from the potable water system. This result is a combination of no use, low use, and a small number of high users. o Low use accounts may irrigate only a very small portion of their irrigable area but this cannot be determined fr o m billing records. I rrigable area application rates can be used confidently as a basis of comparison between groups because overall reduction i n water use is the goal of conservation activities. This goal is independent of the method used to reduce water. Water use may be reduced by replacing high water demanding landscape with less water demanding types increasing the efficiency of the irrigati on system, or reducing the irrigated portion of the irrigable area. The in ground sprinkler group o This group has a r elatively low overall application rate for the calculated irrigable area. This result is a combination of high and low rate accounts. o The h igh rate accounts are typically newer more expensive homes, many of which have separate irrigation meters. o If trends continue as has been observed t he cumulative effect of newer homes with in ground sprinkler systems will increase the water use in the st udy area and likely in Florida unless changes are made to how much area these systems are irrigating, the water demand of that landscape, and/or the efficiency of the irrigation system. The dual meter group
81 o This group primarily uses public supply for irri gation. o Of this group 56 homes were built in 2007 and had an average application rate of 3.6 in/month T hese higher rates may indicate landscape establishment, as some o High use from thes e 2007 homes did not skew the mean because the mean of the group remained relatively unchanged by removing these 56 accounts. o This group is not representative of the population nor of the in ground irrigator group within that population and should not be us ed as a benchmark for average use. It has been shown that outdoor water use can be ascertained from known irrigation accounts on the potable water system. However, the percentage of accounts that irrigate from the potable water system varies widely in Flor ida (Whitcomb et al. 2005) Unfortunately, little information is known about which accounts irrigate. In order to consider how many accounts in a utility use the potable system for irrigation a new approach must be taken. The ability to determine irrigati on use from the potable water system and the savings that a utility may seek requires more detailed analysis of billing data as will be presented in the following chapter.
82 Figure 4 1. Overall seasonal average application rates (T1, T2, and T3) from Flor ridge collected over two years (addapted from Baum 2005) Figure 4 2 Average monthly application rate time series from 409 Florida accounts within 4 profile groups (Profiles are based on percentiles of: value, year built lot area, and buil ding area. The profile percentiles are: 1)25 th ,2)50 th 3)75 th and 4) 90 th ); using data provided by Whitcomb (2005)
83 Figure 4 3 Average indoor and outdoor water use for 1,402 dual metered (left figure) and 29,504 single metered (right figure) residenti al accounts in the Alachua County Study area
84 Figure 4 4 Comparison of Alachua County customer with and without sprinklers.
85 Figure 4 5 Alachua County study area water use compared to occurrence of in ground irrigation systems from 1960 to 2007
86 F igure 4 6. Alachua County study area average peak month and average annual water use compared to cumulative occurrence of in ground irrigation systems from 1960 to 2007
87 Figure 4 7. Trend of percent homes with irrigation to annual average GPAD
88 Figure 4 8. Cumulative probability distribution and frequency histogram of annual rainfall data (1961 to 1969 and 1984 to 2009) from the NOAA site at Gainesville Regional Airport
89 Use ft 2 throughout to be consistent. Table 4 1. Summary of irrigation application rates from a study in during 2002 to 2003 (Baum 2005) Study Site Mean Irrigated Area ( ft 2 ) Application (in. / month ) Average ET 0 (in. / month ) Peak Application Application a s Percent of ET 0 T1 14,498 5.75 4.36 6.8 5, May 132% T2 10,393 4.57 4.36 6.40, Sep 105% T3 9,149 3.39 4.36 5.89, Apr 78% Overall 11,149 4.67 4.36 5.97, Oct 107% Weighted average based on entire group. Table 4 2. Count of accounts from the utilities in Whitcomb's database with parcel area and irrigation from the potable water system. Utility Count City of Lakeland 119 City of Melbourne 32 Escambia County Utilities 22 Hillsborough County Utilities 31 Indian River County Utilities 1 Palm Beach County 1 Sarasota County Utilities 74 Sp ring Hill 124 Toho Water Authority 5 Total 409 Table 4 3. Summary of irrigation application rates from 409 Florida accounts within 4 profile groups, using data provided by Whitcomb, 2005. Profile ( n ) Proxy Irriga ble ( ft 2 ) Application (in. /month) Averag e ET (NET) (in. /month) Peak Application $ Application As Percent of ET (NET) $ 1 (103) 8,110 1.65 4.12 2.51, May 40% 2 (134) 10,437 1.88 4.13 2.84, May 46% 3 (94) 9,312 3.02 4.16 4.35, May 73% 4 (78) 12,165 3.14 4.12 4.48, May 76% Ove rall* 9,922 2.33 4.14 3.42, May 56%
90 Table 4 4. In ground versus surface irrigation for 88 homes in Denver, CO ( adapted from Stadjuhar ( 1997)) Group Mean ( gpad ) Std Dev. ( gpad ) Count Irrigable Area (ft 2 ) Application (in/month) In ground 917 1,439 38 12 ,329 2.7 Surface 335 651 50 4,401 3.0 Table 4 5. Comparison of accounts with/without in ground irrigation and dual meters in the Alachua County study area. Item No i n ground Sprinkler In ground Sprinkler Dual meter accounts Total or Overall Count 22,60 6 8,304 1,403 30,910 % total accounts 73% 27% 5% 100% Average gpad 209 # 393 # 612 # 259 # % total annual average use 59% 41% 11% 100% Usage in peak month, gpad 385 # 625 # 970 # 384 # % peak month (May) use 56% 44% 11% 100% Persons per house 2.50 2.62 2.59 2.53 Average indoor, gpcd 68.3* 68.3* 68.3 # 68.3* Average indoor gpad 170* 179* 177 # 173 Average irrigation gpad 39* 214* 435 # 86 Peak month (May) irrigation gpad 214* 445* 749 # 212 Irrigable Area Median ft 2 8,282 9,590 12,941 8,447 AR in./month 0.23 1.09 1.64 0.50 Peak AR in./month 1.26 2.27 2.83 1.22 Total peaking factor 1.84 1.59 1.59 1.49 Irrigation peaking factor 5.51 2.08 1.72 2.46 Effective area, average ft 2 1,823 2,854 3,144 2,100 Effective yr. b uilt 1980 1994 1994 1984 Just value $158,327 $302,652 $691,396 $197,106 *Flow estimated based on metered, # Flow directly metered
91 Table 4 6. Monthly rainfall and ET 0 for the Alachua County study area from October 2007 to September 2008. M on th Yr ET 0 (In/Month) P (In/Month) Oct 07 2.8 7.0 Nov 07 1.2 0.8 Dec 07 0.5 2.0 Jan 08 1.6 3.8 Feb 08 1.9 3.2 Mar 08 1.2 4.4 Apr 08 3.6 1.1 May 08 5.4 0.6 Jun 08 5.3 5.3 Jul 08 5.1 5.6 Aug 08 4.4 9.2 Sep 08 4.1 0.5 Total ( in./yr. ) 3 7 4 5 Average ( in./Month ) 3. 1 3.7
92 CHAPTER 5 CLUSTERING THE IRRIGATION USERS IN THE POTABLE WATER SYSTEM Background Use of potable water for irrigation can be reduced through conservation or may be substituted for a lower quality source such as reclaimed wastewater, surface water, or private well water. Ideally, t he utility has information on customers who are using reuse water or have private wells. Often reuse customers are not metered and c ustomers who use wells or surface sources for their irrigation water are very difficult to track. Private residential irriga tion wells, even if a permit is on record with a county health department, have no central data source to identify their prevalence or location. Whitcomb (2005) surveyed a cross section of 3,521 homes in 16 cities in Florida regarding their water use patt erns including their irrigation source. The percentage of irrigation that is provided by the utility ranges from as low as 21% for Melbourne and St. Petersburg to a high of 100% in Tallahassee with an average of 64%. The potential for source substitution in a given area is proportional to the accounts that draw or will draw irrigation water from the potable water system. Thus, a critical value which must be determined is the proportion of accounts in a proposed reclaimed area that use the potable supply fo r irrigation and the quantity of that use. The vast majority of single family residential ( SFR ) water customers are served by a single meter that records total use, typically on a monthly basis. However, some, typically, larger SFR customers, have two mete rs so that the regular metered indoor and irrigation metered outdoor uses are recorded separately. If dual metering is not used t he minimum month method is a popular way to perform single metered hydrograph separation. The minimum month method assumes that outdoor water use ceases in the winter because irrigation water is not needed (Dziegielewski et al. 1993; Vickers 2001) Unfortunately, this approach is less valid for warmer
93 climates like Florida where year round irrigation is practiced (DeOreo et al. 2008; Mayer et al. 2009) Aside from dual meter accounts it is difficult to find direct data in Florida on the proportion of total water use that is associated with indoor purposes or outdoor purposes. This information has been collected historical ly by: surveys, special metering of some or all of the customers, or by using dual metered billing records when available (Mayer and DeOreo 1999; Whitcomb et al. 2005) A direct way to estimate the offline customers has been developed using one or two year s of billing data to infer from the time series signatures of the water use patterns the proportion of accounts that only use potable water for indoor purposes and the volume of that use. Assuming a volumetric balance and ignoring leaks, outdoor use is equ al to total use minus indoor use or Q irrigation = Q total Q indoor ( 5 1) Potable w ater o ffset From the perspective of the utility, each account that deploys an alternative supply reduces irrigation water use by 100%. However, these users are drawing w ater from local supplies (typically wastewater reuse and/or shallow wells) at relatively low commodity charges As a result, these irrigators may use more water than if they were being served by the utility. p o effective potable water savings achieved by using alternative irrigation sources. 610(21), F.A.C.) is (Burton & Associates 2008) : amount of potable quality water (Class F I, G I, or G II groundwater or water meeting drinking water standards) saved through the use of reclaimed water expressed as a percentage of the total reclaimed water used. The potable quality water offset is calcu lated by
94 dividing the amount of potable water saved by the amount of reclaimed water used and In order to calculate the offset ( O ), it is necessary to know the usage for the offset purpose before ( Q b ) and after ( Q a ) the ch ange was made as shown in equation 5 2. ( 5 2) The Florida DEP Reuse Inventory (FDEP 2009) uses an offset of 40% for public water supply systems. The offset percentages are used in calculating a savings rate ( Q s ) as shown in equation 5 3. ( 5 3) The value of O is assumed to be between 0 and 100% under the assumption that the alternative water use will be higher than potable water use because it is available at a relat ively low or even zero commodity charge. Reclaimed water use could be less than the prior potable water use especially if the unit cost of reclaimed water is greater than it is for potable water. Burton and Associates (2008) studied 60 homes in Ocoee, Flo rida with separate irrigation meters. The average water use for these homes before switching to reuse was 346 gallons per account per day ( gpad ). The irrigation water use after switching to reuse decreased to 298 gpad This unexpected result may be due to the similarity in the charges for potable and reclaimed water and recent efforts encouraging water conservation. In the Ocoee case, all of the saved water should be credited since the irrigation use actually decreased. Based on review of numerous reuse st udies in SWFWMD, potable water offsets for residential use range from 25 35% for unmetered reuse to 45 55% for metered reuse with an average of 40% (Andrade and Scott 2002)
95 BMP 1 in the original Guide for water conservation planning in Florida estimates a savings rate ( Q s ) of 300 gpad for SFR accounts (Hazen and Sawyer 2003) This estimate was derived from SWFWMD reuse studies using accounts with separate irrigation meters. As a result, this savings rate may not be representative of typical accounts. The value of Q b in equation 5 3 can be assumed equivalent to Q irrigation in equation 5 1 if considering irrigation use before reclaimed water is supplied. As a result the focus will be on characterizing irrigation usage from the potable water system ( Q b irri gation ) for any grouping of users. By this source substitution market potential can be evaluated for various discrete strata of potential customers. K means Clustering A popular method to group a data set into categories that are of interest is using the k means clustering algorithm. XLSTAT Version 2009.4.03 (Addinsoft 2009) was used for this analysis. The algorithm optimizes clustering by minimizing the objective function shown in equation 5 4. ( 5 4 ) Where k = number of subsets in the global set = multidimensional centroid or mean point of cluster set C i = multidimensional data point in question, and E = sum of squares of errors in the Euclidean distances to th e iteration subset means The results are k clusters that are as c ompact and separate as possible (Everitt and Dunn 2001; Han and Kamber 2006; Xu et al. 2009) The k means cluster algorithm uses the following basic steps to determine appropriate assignment to clusters: Input the number of clusters. Assign each data point randomly to a cluster. The centroid of each cluster is determined, and then each point is assigned to the cluster whose centroid is closest to it by Euclidian distance.
96 This process is repe ated until no points switch clusters (convergence) indicating that the minimum value of E in equation 5 4 has been found (Han and Kamber 2006). Each data point is the mean (x1) and monthly peak (x2) of the billed water use for the customer. The mean give s a direct measure of the average water use whereas the peak measures the seasonality and peak seasonal water use for each customer. Thus, offline irrigation users would be expected to have a monthly water use pattern with relatively low mean and peak mont h usage For these indoor users, the mean water use is directly proportional to the persons per house. The users were divided into the following three clusters: 1. s maller users with low peak month use ; 2. m edium users with medium peak month use ; and 3. l arger u sers with high peak month use. The hypothesis is that cluster 1 members correspond to customers who are using other sources of irrigation water or have no or minimal demand for irrigation water. Prior Measures of Indoor Use The monthly use of water, for single and dual meter accounts, over the period of observation has been shown in figure 4 5. From this it can be observed that the indoor water use is relatively constant and averages less than 200 gpad This value of indoor use appears to be relatively c onstant independent of single or dual metering. D uring the period of observation GRU imposed a residential wastewater charge which was calculated as $4.10/k g al times the irrigation) usage. This offe red a strong incentive to minimize water use during those months. Because of this financial reinforcement the minimum month method may be a fair approximation of indoor use during that period. The minimum calculated average flow for all the regular meters in the month of January, 2008 is 184 gpad The 200 0 U.S. Census blocks yielded
97 an average household size ( HHS a vg ) of 2.53 people per house. At an HHS a vg of 2.53 this gives the service area a minimum month gpcd of 72.7. W ater use of dual metered accounts i s much larger than that of single metered accounts. From figure 5 1 the median use for dual metered accounts is 532 gpad with irrigation accounting for 355 gpad (532 2.53*70) or 67%. In contrast the single meter group has a median use of 192 gpad with i rrigation accounting for 15 gpad (192 2.53*70) or 8%. By comparing the median annual average usage for the 1,403 dual meter accounts to the 29,507 single meter accounts the dual meter accounts can be shown to be unrepresentative of the majority of users. Single meter accounts in general are minimal users of potable irrigation water. The cost of water for the GRU can be high as shown in table 5 1 especially if wastewater charges area included. W astewater charges are calculated based on billed water use f or all residential regular highest monthly avera ge water usage for the January or February g al or 33 gallons per account per day ( gpad ). Wastewater charges are calculated based on the smaller of either the actual monthly water usage or the es GRU 2008). As of October 2008, the charge for individual accounts wastewater was $4.90/kgal. Regular and irrigation meters are billed at different inverted block rate structures. As a result a customer would need to us e well o ver nine thousand gallons (k g al) per month or 300 gpad in order for a separate line to be cost effective. As a result only about the upper 25th percentile of single meter accounts use over 300 gpad M uch benefit in having an irrigation
98 meter may come fro m the reduction in the wastewater charge as many homes have some irrigation year round. Having a separate irrigation meter allow s the customer to irrigate year round without penalty. The regular metered usage for the indoor can be assumed to represent ind oor usage as irrigation is most likely accounted for with the irrigation meter. The regular metered flow for dual meter accounts shows an arithmetic mean of 178 gpad and standard deviation of 68 gpad Using Census block level data it was determined that th is group had an average household size (HHS a vg ) of 2.59 people per house. Thus, the average indoor per capita use for the area by this method was 68.7 gallons per person per day ( gpcd ). However by looking at the data in figure 5 2, it can be observed that 19 of the assumed indoor values are above 560 gpad Using 70 gpcd as an expected value this results in an average household size of over eight people, an unlikely event. The values above 560 gpad for annual average or peak month can be considered erroneous in determining expected indoor use. By looking at a scatter plot of the annual average and the peak billed usage month of May for each account the outlier data becomes quite evident. Thus a bounded dataset is used as shown by the exploded portion of figure 5 2. and a mean peak month usage of 179 gpad or 69.25 gpcd. These values are lower than the 72.7 gpcd calculated using the minimum month method. National studies performed by Aquacraft Inc. (Aquacraft 2005; DeOreo et al. 2008; Mayer and DeOreo 1999) have shown that indoor water use is relatively constant and averages around 70 gpcd. Thus, several calculations of indoor use ca n be produced for the study area, as shown in table 5 2. These results give a consistent value for indoor use in the study area of about 70 gpcd.
99 This value will be used as a check of the cluster assumed to have no or minimal demand for irrigation water. I n absence of direct data, 70 gpcd should be used to estimate indoor water use for SFR customer groups in Florida (Friedman 2009) Clustering Of Total Water Use for All Accounts To apply a k means clustering to indoor and outdoor use, an overall dataset of Q total was created fr o m the single and dual meter accounts. The regular and irrigation metered flow s for each of the dual total for each customer in the group. The single meter accou nts already report Q total The annual average and the monthly peak flows were calculated for the overall dataset. The data points for 1 ) and the peak month of May (x 2 ) were binned using widths calculated using the (Sco tt 1979) method. The binned data was then put into the three dimensional histogram shown in Figure 5 3. The k means clustering algorithm was applied to form three clusters based on x 1 and x 2 The numerical results are summarized in table 5 3. The percent of customers in each group illustrates the densities of x 1 and x 2 shown in figure 5 3. The centroids can be calculated as the mean of x 1 and the mean of x 2 to describe the central tendency of each cluster. The results in table 5 3 indicate that over 70% of the use in the service area is characteristic of indoor use with metered indoor usage. Thus, only 29% of accounts appear to be significant uses of potable water fo r irrigation. Strikingly, these accounts use 56% of the water on average and contribute to 66% of the usage during the peak month of May. The marginal probability density histograms are shown in figure 5 4. The histogram of the cluster marginal density fun ctions were binned using widths calculated by the Scott (1979) method in Matlab. Overlaps in the clusters can be observed in figure 5 4. By this visualization,
100 the annual average upper bound on minimal/offline use is about 275 gpad (109 gpcd) and the uppe r bound on monthly peak use for the minimal/offline group is about 500 gpad ( 198 gpcd). These observations do not say that all water use above these bounds is for irrigation. These marginal bounds simply state that as accounts go further past them the use is more likely to be for irrigation (Singh et al. 2007) The overall (all clusters) annual avera ge was 259 gpad and the peak month use was 384 gpad. This indicates that the majority of customers on the system do not use potable water for irrigation as evid enced by the frequency histogram in figure 5 3. Combining the ACPA with FDOR databases gives extensive data on property and structural attributes for each customer. The ACPA database has several important fields for analysis including: Presence of a sprin kler system, effective year built, stories, baths, gross area of structure, area of drive/walkways, patios, screened enclosures, etc... No differentiation is made between automatic or manual systems in this database. The effective year built is the actual year of construction or the year of major renovation. The area data in the ACPA database can be used in conjunction with FDOR data to determine the irrigable area for each parcel. Adding the census data allows the estimation of HHS av g for the clus ters. The median of the irrigable area (A irrigable ) was used to calculate the application rate (AR) in in./ month because the mean was skewed by outliers, e.g., the maximum value being 2,015,766 square feet, returning unreasonable irrigable area for SFR par cels. The peak and average AR were calculated following equation 4 3. Indoor use was assumed using Cluster 1 gpcd as this
101 value was close to the a priori assumed value. It is recommended that a minimum of three clusters be used but more should be added to individual analyses such that the annual average of the minimal cluster is as close to 70 gpcd as possible. Cluster 3, comprises newer, larger, and more expensive homes on larger lots. Cluster 3 homes contain the majority of dual meter accounts and sprink ler systems and have the highest water use and application rate. At least 3,700 accounts do not irrigate f ro m the potable system as indicated by the percent of accounts with sprinkler system in cluster 1. Overall only about one quarter of the accounts app ear to use the potable water system for irrigation. Gainesville Regional Utilities supplies over 1,100 reclaimed water accounts in the study area with over 2 million gallons per day (MGD) treated wastewater. This averages out to about 1,800 GPAD of irriga tion water Q a in equation 5 2. Interestingly there are about 3,700 customers in Cluster 1 with in ground sprinklers ; thus around one half would be expected to use reclaimed water and the other half would use private wells or surface water as a substitute to potable water for irrigation. Clustering results by homes built before 1985 versus during or after 1985 It can be observed from table 5 4 that newer homes tend to be in the upper use group (Cluster 3) as eviden ced by the average year built of 1992. T hese homes use much more water than the older homes. This is called out by the fact that the minimal/offline group (Cluster 1) has an average year built of 1982. This is counter intuitive as water saving indoor fixtures have been installed by code in newer homes (Friedman 2009) Thus, all other factors being equal newer homes should use less water. The first consideration is the household size. Indeed the newer homes do tend to have more people, 2.63 for Cluster 3 as opposed to 2.50 for Cluster 1. This doe s not explain the difference in water use I f the population normalized gpcd value is use d, the result is 360 gpcd for Cluster 3 as opposed to 62.2 for Cluster 1. Indeed the difference between
102 the groups is called out by the percent of accounts with sprink ler systems and the irrigable area for the respective groups. From table 5 4, it can be seen that Cluster 3 had 74% of its customer s with in ground irrigation while Cluster 1 had 17%. It has been shown in figures 3 15, 4 7, and 3 16 that in ground irrigat ion and water use are increasing and that irrigable area is decreasing. By looking at the summary attributes in table 5 5 it can be seen that homes built before 1985 closely resemble Cluster 1. This is logical as 80% of the accounts built before 1985 are i n Cluster 1. By comparison the homes built during or after 1985 tend to rese mble a mix of Cluster 1 and 2. Extrapolation of Gainesville r esults to o ther Florida u tilities Use of billing data linked to customer attributes is the most rigorous method to det ermine the water use and to target groups of customer s for conservation programs Thus the recommended approach for estimating the savings rates or irrigation offsets for SFR customers in Florida is to divide the customers into clusters as described in th is chapter using available software. The results of the clustering will tell the total average and peak month irrigation use at present. Total potable water use for the single family residential sector can be separated into its respective indoor and outdoo r components using a simple point estimation of the indoor portion. Additionally, a more disaggregated k means clustering algorithm may be used to differentiate between levels of outdoor use. A point estimate of indoor use of 70 gpcd and average household size (HHS avg ) from the U.S. Census can be use d to estimate indoor water use. Replacing this value for Q indoor in equation 5 1, the equation for irrigation use becomes: Q irrigation (gpad) = Q total (gpad) 70 gpcd*HHS av g ( 5 6 ) Outdoor use consists primari ly of irrigation and can be broken down into three groups using the k means clustering algorithm, as described in this chapter These groups are: 1. Minimal/Offline,
103 2. Mid range, and 3. Upper. It is recommended that a minimum of three clusters be used. More clu sters may be added to individual analyses such that the annual average of the minimal cluster is as close to the a priori value of 70 gpcd as possible. The first cluster represents those accounts that use little or no irrigation water because they do not irrigate or they irrigate using alternative sources, e.g. ., reuse water and private wells. These are essentially indoor only customers. The second cluster indicates the average irrigation users. The irrigation use may be calculated using equation 5 6 or by replacing the indoor 70 gpcd in equation 5 6 with the gpcd of the first cluster or: Q irrigation (gpad) = Q total (gpad) (Cluster 1 gpcd)*HHS av g (5 7) This result would be the annual average irrigation use in gpad. The third cluster represents the above average users of irrigation water. Their irrigation use can be calculated using equation 5 6 or equation 5 7 This group represents the largest potential for savings in irrigation use. The difference between Cluster 3 and Cluster 2 in table 5 6 is the sav ing s that may be expected by improving the system efficiency of this group (landscape or irrigation) to the average efficiencies of the area. For the annual average this would be a savings of 430 gpad. A peak month savings of 1,036 gpad per upper use accou nt during the month of May can be achieved. Potable water savings can also be realized by removal of Cluster 2 and Cluster 3 accounts from the potable water system. This is accomplished by providing an alternative supply. The se saving s may be less than the saving s from efficiency improvements if the offset percentage is taken into consideration. For reuse offsets simply multiply the present average annual water use
104 by 30 or 50% depending upon whether the option is unmetered or metered, respectively. Finall y, calculate the expected offset savings as shown in table 5 6 Both the efficiency improvement and the potable offset approaches suggest a clear strategy of targeting the higher end irrigation users with the highest variability in use. In the case of GRU this target population would be about the top 10% of the users. Generalization of Clusters Using Discriminant Analysis A discriminant analysis was performed in order to simplify classification of customers as indoor and outdoor users. The dual meter acc ounts were used as a training dataset. Indoor use was assumed to be regular metered usage and indoor and outdoor use was assumed to be the sum of the regular and irrigation metered usage. Indoor only accounts were given a discriminant value of 1 and indo or and outdoor accounts were given a value of 2. A least squares regression was performed which predicts the discriminant score (DS) based on annual average (x 1 ) and peak month (x 2 ) water use. This regression equation is the discriminant function. DS = 0 + 1 x 1 + 2 x 2 (5 8) A cut off is assigned to the DS based on the mid point between the average DS of each group. An un weighted discriminant function was used due to equal priors (Ragsdale, 2004). The parameters for the training set are shown in table 5 7. The error was calculated and only 14.7% of the accounts were misclassified. This is a good result given the equal priors. A confusion matrix shown in table 5 8 indicates that more indoor and outdoor users were misclassified as indoor only than indoor on ly are misclassified as indoor and outdoor. This is likely an error in the training data as many indoor and outdoor accounts had unreasonably low water use, as shown in figure 5 5. An optimization was performed using MS Excel nonlinear Solver to find the discriminant cut o f f value that minimizes the error. The optimized cut off was 1.42 as opposed to
105 1.50. By looking at the confusion matrix presented in table 5 9 it can be seen that the error is more evenly distributed. The optimized cut off returned an er ror of only 11.1%. The discriminant function with the 1.42 cut off matches the clustered 30,910 accounts almost exactly if the mid range and upper clusters are combined to form an indoor and outdoor group, as shown in figure 5 6 Conclusions The tendency of a customer to use the potable water for irrigation can be calculated based on the discriminant score simplifying the analysis process. Using the most recent years billing data, the peak month and the annual average gpad for each customer can be used to calculate their discriminant score (DS) as follows: DS = 1.16077 + 3.10 228 E 04 *Annual average, gpad + 4.03 355 E 04 Peak month, gpad ( 5 9 ) Decisions can be made knowing that if the DS is greater than 1.42 that account is more likely to use water for irriga tion purposes. Knowing this peak month and annual average savings, functions can be created. These are shown by the linear regression equation in figure 5 7 The comparative results of this analysis are dependent on the average people per house of the uti lity, the presence of irrigation restrictions, and the relative abundance of rainfall. The result of these relationships gives the ability to calculate peak month and annual average savings because of lowering the discriminant score. Additionally, the offs et can be calculated if the irrigation use is multiplied by its respective offset percentage, 30% for unmetered and 50% for metered alternative source. Thus, two simple equations and the DS calculated from billing data with equation 5 10 may be used: Peak month irrigation, gpad =0 ;if DS<1.42 Peak month irrigation, gpad =1,691.7*DS 2,130.4 ;if DS>1.42 a nd, ( 5 10 )
106 Annual avg. irrigation, gpad=0 ;if DS<1.42 Annual avg. irrigation, gpad= 781.12*DS 969.74 ;if DS>1.42 This method can be utilized at any lev el of aggregation from micro to macro. A single parcel may be classified as indoor only or indoor and outdoor. A utility wide discriminant score and respective peak and average irrigation use may be calculated and the irrigable area application rate may ev en be calculated if the parcels within the utility service area boundary are known from the FDOR database. The applications and conservation planning potential for utilizing the FDOR database and billing data characterization and clustering is extensive. T his predictor should be reliable since it is based on the large GRU database.
107 Figure 5 1. C umulative frequency distributions of total water use for 29,507 single and 1,403 dual metered SFR accounts in the GRU service area
108 Figure 5 2. Scatter plot of t he regular metered flow from the dual metered accounts. An extruded view details the accounts that are expected to represent the distribution of actual indoor use in the area
109 Figure 5 3. Frequency histogram of the total annual averages (x1) and peak mo nths flows (x2) of all accounts in the Alachua County study area. Red indicates frequencies between 0 and 49; orange indicates frequencies between 50 and 499; green indicates frequencies of at least 500.
110 Figure 5 4 K means clusters for the overall datas et showing scatter and marginal distributions of x 1 and x 2 for each cluster: 1) Minimal/Offline, 2) Mid range, and 3) Upper.
111 Figure 5 5. Discriminant analysis utilizing dual metered accounts as training data
112 Figure 5 6. Discriminant analysis applied to the 30,910 clustered users in the Alachua County study area.
113 Figure 5 7. Regression of estimated irrigation water use in gpad verses discriminant score.
114 Table 5 1. GRU rate structure for irrigation and regular metered accounts during period of observat ion. Metered Usage (k g al) Charge ($/k g al) Irrigation < 15 3.11 Irrigation > 15 5.50 Regular < 9 1.59 3.11 5.50 Winter Max or 1 4.90 Table 5 2. Results of current methods for calculating the indoor water use in the study area. Method GPCD Minimum month 72.7 Regular metered from d ual meter accounts 68.7 Referenced approx imation (Friedman 2009) 70.0 r egular metered from d ual meter accounts 63.7 Table 5 3. The k means cluster centroids and water use percentages. Irrigation Group Mean, gpad Peak month (May), gpad % of SFR Customers % of SFR annual average water use % of SFR peak water use 1.) Minimal/Offline 162 188 70% 44% 34% 2.) Mid range 429 708 25% 42% 47% 3.) Upper 861 1,746 4% 14% 19% Overall/Total 259 384 100% 100% 100%
115 Table 5 4. Summary of selected cluster attributes for the Alachua County study area I tem Cluster 1 Cluster 2 Cluster 3 Overall Count 21,763 7,878 1,269 30,910 Average gpad 162 429 861 259 Usage in peak month, gpad 188 708 1,746 384 Persons per house 2.50 2.60 2.63 2.53 Average indoor, gpcd 64.8 64.8 64.8 64.8 Average indoor gpad 162 168 170 164 Average irrigation gpad 0 260 690 95 Peak month (May) irrigation gpad 26 540 1,576 220 % Dual meter accounts 1% 10% 30% 5% % w ith sprinklers 17% 47% 74% 27% Irrigable Area Median ft 2 8,073 9,385 16,581 8,447 AR in./month 0.00 1.35 2. 03 0.55 Peak AR in./month 0.16 2.81 4.64 1.27 Total peaking factor 1.16 1.65 2.03 1.49 Irrigation peaking factor Undefined 2.07 2.28 2.33 Effective area, average ft 2 1,894 2,475 3,314 2,100 Effective yr. b uilt 1982 1989 1992 1984 Just value $171,467 $241,374 $361,954 $197,106
116 Table 5 5. Comparison table of selected attributes for pre 1985 group and the 1985 to 2007group. Item Pre 1985 1985 to 2007 Overall Count 17,295 13,615 30,910 Average gpad 212 317 259 Usage in peak month, gpad 293 501 384 Persons per house 2.49 2.59 2.53 Average indoor, gpcd 62.2 62.2 62.2 Average indoor gpad 155 161 157 Average irrigation gpad 58 157 101 Peak month (May) irrigation gpad 138 340 227 % Cluster 1 81% 57% 70% % Cluster 2 17% 36% 25% % Cluster 3 2% 7% 4% % Accounts with sprinklers 9% 49% 27% Irrigable Area Median ft 2 9,322 6,728 8,447 AR in./month 0.30 1.14 0.58 Peak AR in./month 0.72 2.47 1.31 Total peaking factor 1.38 1.58 1.48 Irrigat ion peaking factor 2.38 2.16 2.24 Effective area, average ft 2 1,851 2,416 2,100 Just value $154,274 $251,443 $197,074 Count 17,295 13,615 30,910 Table 5 6 Default savings estimate for BMP 1 (non potable irrigation source replacements) for single f amily residential users. Irrigation Group % of Total Irrigation average gpad Irrigation peak gpad Offset % Offset s avings gpad Minimal 70.6% 0 26 50% 0 Mid range 25.3% 260 540 50% 130 Upper 4.1% 690 1,576 50% 387 Wgt. Avg. irrigators 29.4% 320 684 50% 160 Wgt. Avg. total 100% 101 220 50% 51 *Offset = 30% for unmetered and 50% for metered alternative source.
117 Table 5 7. Training dataset discriminant analysis function parameters Group DS Function variable Training set result 1 1.30 0 1.16077 2 1 .71 1 3.10 228 E 04 cut off 1.50 2 4.03 355 E 04 Table 5 8. Confusion matrix for the results of the 1.50 cut off discriminant analysis run on the training dataset Predicted 1 2 Total Actual 1 1,355 36 1,391 2 374 1,018 1,392 total 1,729 1,054 2 ,783 Table 5 9. Confusion matrix for the results of the 1.42 optimized cut off discriminant analysis run on the training dataset Predicted 1 2 Total Actual 1 1,3 02 89 1,391 2 219 1, 173 1,392 total 1, 521 1, 262 2,783
118 CHAPTER 6 SUMMARY AND CON CLUSIONS Irrigation water use for a utility service area is equal to the sum of the product s of irrigated area and application rate for each SFR parcel in the service area following equation 1 1. (1 1) Where : q is the irrigation wat er use, n is the count of irrigators on the system, A i is the irrigated area of irrigator i, and A R i is the application rate of irrigator i. The irriga ble area is equivalent to pervious area and can be calculated by rearranging equation 3 1 as: PA = TA IA NA (6 1) Where PA = parcel pervious area, and TA = total parcel area, IA = parcel impervious area, NA = non applicable or other area (easements, etc...) Irriga ble area can be estimated from the FDOR database for a utility service area if the se rvice area boundary is known or is completely contained within municipal boundaries. W ith this list of parcel areas for the SFR sector simple GIS techniques can be used to calculate parcel area from the FDOR parcel GIS files. Aerial imagery or landuse/lan dcover maps should be used to estimate the percent of these parcels that falls over non applicable areas such as lakes or wetland easements and setbacks. If the parcels are drawn to ex clude non applicable areas as part of the SFR parcels th e n this term in equation 1 1 can be ignored. In the case of the Alachua County study area Non ap plicable area was assumed negligible
119 If the number of stories is available the impervious footprint of the structure on the parcel can be calculated using the following rati os: For 1 story FS/EA = 1.15 For 1.5 story FS/EA = 0.87 For 2 story FS/EA = 0.70, and For 3 story FS/EA = 0.55 The refined coefficients should be used if the housing stock distribution of stories varies significantly from the following: For 1 story 85.9%, For 1.5 story 1.7%, For 2 story 9.0%, For 2 .5 story 0.0%, and For 3 story 3.3% The associated impervious areas can be estimated as 0.6 9 the FDOR effective area. AIA= EA; w h ere AIA/EA ) = 0.6 9 for the overall dataset (3 6 ) Utilizing total parcel a rea (TA) and effective area (EA) from the FDOR database irrigable area is calculated as the difference of the total parcel area and the impervious features on the parcel following equation 3 10. Irrigable area = 0 ; if TA E A (1. 75 ) < 0 Irrigable area = TA EA (1. 75 ) ; if 95 th percentile TA < TA EA (1. 75 ) < 0 (3 10) Irrigable area = 95 th percentile TA ; if TA EA (1. 75 ) > 95 th percentile TA Irrigable area application rates can be expect ed to be somewhat lower t han irrigated area application rates as not all of the irrigable area is typically irrigated. Irrigable area is however more standardized and can be more readily applied. Not all parcels within a utility service area irrigate with potable water or at the s ame rate. The percent of account irrigating from the potable water system varies from 0% to 100% depending on the utility (Whitcomb et al. 2005)
120 The most reliable and robust way to determine the percent of accounts that irrigate from the potable water sy stem is to use billing data for the sector being examined. This billing data need not be linked to the parcel attributes. The discriminant function shown in equation 5 9 should be used on the mean and peak use month of billing data expressed as gallons per account per day (gpad) Accounts with a DS less than 1.42 will be the non irrigators. The application rate of the utility may also vary depending on customer affluence age of the homes and presences of in ground irrigation. It has been found that newer homes tend to have a higher reported value and are far more likely to have in ground irrigation. Results for the study area show that indoor water use can be estimated as a steady state rgime with an average daily flow of 70 gallons per person overall. I rrigation is a transient rgime with an average peak month use of 684 gallons per day per account and an annual average use of 320 gallons per day per account over the period of observation. The median irrigable area for homes using the potable system for irrigation is 10,383 square feet. Thus, the expected water use coefficients for this dataset are: an annual average of 1. 5 inches per month per square foot of irrigable area and a peak rate of 3. 1 inches per month per square foot of irrigable area. The oc currence of sprinklers systems in homes began to increase 2007. In ground irrigation homes make up 27% of the cumulative housing stock. This cumulative percentage i s expected to increase if in ground irrigation remains popular For the it is estimated that at least 44% of accounts with in ground irrigation do not use water from the potable system for irrigation but have substitute sources such as: wastewater reuse, private wells, and/or surface water sources. It is also estimated that about
121 2 0 % of the accounts without in ground irrigation systems do irrigate with potable water. The resulting weighted percentage of accounts using po table water for irrigation is 3 0 %. Florida utilities may use t he discriminant equations from this study to estimate the percentage and usage of accounts using potable water for irrigation in their utility The most accuracy can be achieved when the followi ng conditions are met for the utility : Average people per house is 2.5, Average ET 0 is 37 inches per year, Average lot size is about 15,000 sq. ft., and No irrigation restrictions are in place. T he tendency of a customer to use the potable water for irri gation can be calculated based on the discriminant score. The relationship of discriminant score to irrigation water use in GRU gives the ability to calculate peak month and annual average savings because of lowering the discriminant score. Additionally t he offset can be calculated if the irrigation use is multiplied by its respective offset percentage, 30% for unmetered and 50% for metered alternative source. Thus the generalized equation s 5 9 and 5 10 may be used to apply the results of this st udy to an y utility in Florida. This analysis method can be utilized at any level of aggregation from micro to macro. At the micro level a single parcel may b e classified as indoor only or in door and outdoor. At the macro level a utility wide discriminant score and respective peak and average irrigation use may be calculated and the irrigable area application rate may even be calculated if the parcels within the utility service area boundary are known from the FDOR database. The applications and conservation planning potential for utilizing the FDOR database and billing data characterization and clustering is extensive.
122 Future work is needed to develop method s that more rigorously define the percent of homes that irrigate in both the pre and post 1985 groups. Method s to define the parcels connected to the potable water system within the service area boundary need to be developed on a statewide level. These studies will led to more accurate conclusion s about the FDOR parcel s that are being used to create these estimat es
123 APPENDIX A ANNUAL TRENDS OF PARCEL SQUARE FOOTAGE IN THE ALACHUA COUNTY STUDY AREA Effective year built Median parcel square footage Average parcel square footage Count parcels Pre 1960 10,586 28,552 701 1960 11,774 18,158 290 1961 10,200 12,801 181 1962 10,004 15,069 228 1963 10,499 16,822 193 1964 10,807 14,239 348 1965 11,186 17,250 545 1966 12,581 16,736 327 1967 13,353 19,809 353 1968 10,161 14,662 816 1969 12,440 15,404 733 1970 13,285 20,895 817 1971 12,539 16,849 560 1972 13,661 19 ,547 699 1973 13,660 20,585 730 1974 14,372 22,557 616 1975 13,019 18,852 1 190 1976 12,376 17,355 1 154 1977 12,816 19,327 593 1978 12,600 18,638 905 1979 12,828 16,771 876 1980 14,047 20,145 1 053 1981 13,462 17,556 731 1982 11,450 14,504 520 1983 9,402 13,849 1 178 1984 10,519 13,783 960 1985 12,000 19,627 749 1986 9,684 16,892 539 1987 11,629 18,269 635 1988 10,928 18,752 556 1989 11,931 17,651 567 1990 11,142 17,000 556 1991 10,355 16,004 475 1992 10,519 16,842 553 1993 9,977 17,59 1 561 1994 11,080 18,313 567 1995 10,438 17,656 553
124 1996 9,866 14,079 566 1997 9,963 17,084 602 1998 10,100 16,196 600 1999 9,864 15,948 650 2000 10,012 16,331 606 2001 9,596 16,071 623 2002 9,307 16,686 588 2003 8,850 16,358 531 2004 8,399 12,4 46 668 2005 7,758 12,159 729 2006 8,585 13,315 674 2007 8,641 13,491 465 Overall 11,214 16,997 30,209
125 APPENDIX B ANNUAL TIME SERIES S HOWING BREAKDOWN OF 1, 1.5, 2, 2.5, AND 3 STORY HOMES IN THE ALACHUA COUNTY STUDY AREA Year Built Count of Homes 1 St ory 1.5 Story 2 Story 2.5 Story 3 Story Average Stories Pre 1960 696 96.4% 0.0% 2.3% 0.0% 1.3% 1.05 1960 290 94.1% 0.0% 3.1% 0.0% 2.8% 1.09 1961 181 98.9% 0.0% 0.6% 0.0% 0.6% 1.02 1962 228 97.8% 0.0% 1.3% 0.0% 0.9% 1.03 1963 193 95.9% 0.0% 3.6% 0.0% 0 .5% 1.05 1964 348 95.4% 0.0% 2.0% 0.0% 2.6% 1.07 1965 545 93.4% 0.2% 4.4% 0.0% 2.0% 1.09 1966 327 92.4% 0.0% 2.4% 0.0% 5.2% 1.13 1967 353 90.1% 0.0% 3.4% 0.0% 6.5% 1.16 1968 816 96.0% 0.0% 2.6% 0.0% 1.5% 1.06 1969 733 95.0% 0.1% 3.1% 0.0% 1.8% 1.07 1970 817 88.9% 0.5% 7.2% 0.0% 3.4% 1.14 1971 560 93.6% 0.2% 2.3% 0.0% 3.9% 1.10 1972 699 92.3% 0.1% 2.9% 0.0% 4.7% 1.12 1973 730 90.4% 0.1% 5.9% 0.0% 3.6% 1.13 1974 616 90.6% 0.0% 4.5% 0.0% 4.9% 1.14 1975 1190 90.1% 0.2% 6.2% 0.0% 3.5% 1.13 1976 1154 92.6% 0.3% 4.0% 0.0% 3.0% 1.10 1977 594 88.0% 0.7% 5.1% 0.0% 6.2% 1.18 1978 906 90.5% 0.1% 4.1% 0.0% 5.3% 1.15 1979 876 90.5% 0.2% 3.4% 0.0% 5.8% 1.15 1980 1054 90.5% 0.2% 4.6% 0.0% 4.7% 1.14 1981 731 86.3% 0.0% 7.8% 0.0% 5.9% 1.20 1982 520 76.5% 0. 4% 15.8% 0.0% 7.3% 1.31 1983 1178 74.8% 0.0% 19.1% 0.0% 6.1% 1.31 1984 960 69.0% 0.2% 19.5% 0.0% 11.4% 1.42 1985 749 78.6% 0.0% 10.0% 0.0% 11.3% 1.33 1986 539 77.0% 0.2% 9.8% 0.0% 13.0% 1.36 1987 635 86.5% 0.2% 3.6% 0.0% 9.8% 1.23 1988 556 85.6% 0.2% 10.6% 0.0% 3.6% 1.18 1989 567 86.4% 0.2% 12.9% 0.0% 0.5% 1.14 1990 556 89.7% 0.5% 9.7% 0.0% 0.0% 1.10 1991 475 91.6% 0.0% 8.4% 0.0% 0.0% 1.08 1992 553 91.3% 0.0% 8.5% 0.0% 0.2% 1.09 1993 561 89.2% 0.0% 10.6% 0.0% 0.2% 1.11 1994 567 87.9% 1.4% 10.7% 0.0% 0.0% 1.11 1995 553 83.5% 2.4% 14.1% 0.0% 0.0% 1.15
126 1996 566 85.4% 1.4% 13.0% 0.0% 0.2% 1.14 1997 602 81.2% 4.4% 14.4% 0.0% 0.0% 1.17 1998 600 84.4% 2.9% 12.6% 0.0% 0.2% 1.14 1999 650 81.3% 4.8% 13.5% 0.0% 0.3% 1.17 2000 606 81.8% 5.0% 12.7% 0.3% 0.2% 1.16 2001 623 78.7% 5.6% 15.6% 0.0% 0.2% 1.19 2002 588 77.7% 5.7% 16.4% 0.2% 0.0% 1.20 2003 531 70.7% 11.0% 18.1% 0.0% 0.2% 1.24 2004 668 81.2% 9.3% 8.7% 0.0% 0.8% 1.15 2005 729 77.1% 8.8% 14.0% 0.0% 0.1% 1.19 2006 675 75.0% 7.0% 17.8% 0.0% 0.3 % 1.22 2007 466 68.5% 12.5% 19.0% 0.0% 0.0% 1.25 Overall 30910 85.9% 1.7% 9.0% 0.0% 3.3% 1.17
127 LIST OF REFERENCES Alachua County Property Appraisers (2009). 2008 GIS and Oracle Data Files < http://acpafl.org/ > a c c essed on December 2009 Gaines ville, FL. Addinsoft. (2009). XLSTAT, Statistical software for MS Excel: k means clustering application. < http://www.xlstat.com/en/support/tutorials/cluster2.htm > a c c essed on December 2009 Allen, R. and Enviro nmental and Water Resources institute (U.S.). Task Committee on Standardization of Reference Evapotranspiration. (2005). The ASCE standardized reference evapotranspiration equation American Society of Civil Engineers, Reston, Va. Andr ade, A., and Scott, K (2002). Effective use of reclaimed water demon strated to offset water demand < http://www.swfwmd.state.fl.us/files/database/site_file_sets/118/reclaimed offset docs.pdf > South West Florida Water Management District, Brooksville, FL Aquacraft Inc (20 05). Water and Energy Savings from High Efficiency Fixtures and Appliances in Single F amily Homes. Boulder, C O Baum, M. (2005). Residential Irrigation Water Use in the Central Florida Ridge University of Florida Gainesville, F L Beecher, J. Chesnutt, T ., and Pekelney, D (2001). Socioeconomic impacts of water conservation AWWA Research Foundation and American Water Works Association, Denver, CO. Burton & Associates. (2008). Study to Determine Potable Quality Water Offset From Reuse SJRWMD, Palatka, F L Chesnutt, T., Pekelne y, D., and Maureen, E. (2004). Evaluation of the Landscape Per formance Certification Program Encinitas, C A DeOreo, W., Whitcomb, J., Mayer, P., Henderson, J., Gleick, P., Wolfe, G., Nelson, J. O., and Gauley, B. (2008). California Single Family Water Use Efficiency Study: Draft Report Aquacraft, Inc., Water Engineering and Management, Boulder, CO. Dukes, M. (2009). LEED for Homes: Explanation of the Landscape Irrigation Budget Calculation for Florida < http://edis.ifas.ufl.edu/AE441 > a c c essed on December 2009 University of Florida Cooperative Extension Service, Institute of Food and Agriculture Sciences, EDIS, Gainesville, FL. Dukes, M. Miller, G ., and Hanley, M. (2006). Residential irrigation efficiency asses sment monitoring final report St. Johns River Water Management District, Palatka, FL Dziegielewki, B., and Opitz, E. (2002). Chapter 5 : Water Demand Analysis Urban Water Supply Handbook McGraw Hill, NY
128 Dziegielewski, B., Opitz, E. M., Kiefer, J. C., and Baumann, D. D. (1993). Evaluation of Urban Water Conservation Programs: A Procedures Manual California Urban Water Agencies, Sacramento, California. Dzurik, A. (2003). Water Resources Planning Rowman & Littlefield Publishers, Inc., Lanham, MD. Everi tt, B., and Dunn, G. (2001). Applied M ultivariate D ata A nalysis Arnold, London. Florida Agricultural Weather Network. (2009). FAWN: Report Generator. < http://fawn.ifas.ufl.edu/data/reports/ > a c c essed on December 2009 Universi ty of Florida Cooperative Extension Service, Institute of Food and Agriculture Sciences, EDIS, Gainesville, FL. F lorida Department of Environmental Protection (2008a). Basic Facility Reports. Drinking Water Data, Florida Department of Environmental Protec tion, < http://www.dep.state.fl.us/water/drinkingwater/download.htm > a c c essed on December 2009 Tallahassee FL. Florida Department of Environmental Protection. (2008b). Drinking Water Flow Data. Drinking Wa ter Data, Florida Department of Environmental Protection, < http://www.dep.state.fl.us/water/drinkingwater/download.htm > a c c essed on December 2009 Tallahassee, FL. Florida Department of Environmental Protec tion (2008 c). Guidance on Per Capita Water Use: Uniform Definitions and Performance Me asures March 3, 2008. Florida Department of Environmental Protection, Tallahassee, FL. Florida Department of Environmental Protection (2009). 2007 Reuse Inventory Flo rida Departmen t of Environmental Protection, < http://www.dep.state.fl.us/water/reuse/inventory.htm > a c c essed on December 2009 Tallahassee F L Florida Department of Revenue. (2008). Name Address Legal (N.A. L.) FILE Data Record Layout f or 2008 Comma Delimited Format. < http://dor.myflorida.com/dor/property/ > a c c essed on December 2009 Tallahassee FL. Florida Department of Revenue. (2009). 2009 Name Address Legal Files. < http://dor.myflorida.com/dor/property/ > a c c essed on December 2009 Tallahassee, FL. Florida. Dept. of Community Affairs. Bureau of Local Planning., and Florida. Dept. of Community Affairs. (1989). A guide to local government comprehensive p lanning data sources : a directory to assist local governments in preparing comprehensive plans pursuant to chapter 163, Florida Statutes, and rule chapter 9J 5, Florida Administrative Code Tallahassee, F L Friedman, K. (2009). "Evaluation of Indoor Urban Water Use and Water Loss Management as Conservation Options in Florida," Thesis (ME), University of Florida, Gainesville, FL.
129 Gaviln, P., Lorite, I., Tornero Regional calibration of Hargreaves equation for estimating refer ence ET in a semiarid environment. Agricultural Water Management 81(3), 257 281. Gonzalez Y., and Yingling, J. (2008). How to Quickly Calculate Required and Optional Population Served Estimates using data found in Dis trict and Utility Demographics. Sout hwest Florida Water Management District, Planning Department, Brooksville, FL. G ainesville Regional Utilities (2008). "Calculating Your Residential Wastewater Bill." Gainesville, FL. Haley, M. Dukes, M. and Miller, G. (2007). Residential Irrigation Wat er Use in Central Florida Journal of Irrigation and Drainage Engineering 133(5), 427 434. Han, J., and Kamber, M. (2006). Data mining: concepts and techniques Elsevier ; Mor gan Kaufmann Amsterdam Netherlands ; San Francisco, CA. Hargreaves, G ., and Al len, R. (2003). History and Evaluation of Hargreav es Evapotranspiration Equation. Journal of Irrigation and Drainage Engineering 129(1), 53 63. Hazen and Sawyer Inc. (2003). Potable Water Conservation Best Management Practices for the Tampa Bay Region: F i nal Report for Tampa Bay Water. < http://development.tampabaywater.org/documents/conservation/PotableWaterConservati onBMPs2004.pdf > a c c essed on December 2009 Clearwater FL Heaney, J., Pitt, R. and R. Field, Eds. 2000. Innovative Urban Wet Weather Flow Management Systems. EPA 600 R 99 029, < http://www.epa.gov/ORD/NRMRL/pubs/600r99029/600R99 029prelim.pdf >. Cincinnati, OH. Jensen, M., Burman, R.., Allen, R., and American Society of Civil Engineers. Committee on Irrigation Water Requirements. (1990). Evapotranspiration an d irrigation water requirements : a manual The American Society of Civil Engineers New York N Y. Kirkham, M. (2005). Principles of soil and plant water relations Elsevier Academic Press, Amsterdam ; Boston. Lee, J., and Heaney, J. (2003). "Estimation of Urban Imperviousness and its Impacts on Storm Water Systems." Journal of Water Resources Planning and Management 129(5), 419 426. Lee, J., Heaney, J.., and Lai, F. (2005). "Optimization of Integrated Urban Wet Weather Control Strategies." Journal of Water Resources Planning and Management 131(4), 307 315. Lewis, K., Carriker, R., Marella, R., University of Florida. Food and Resource Economics Dept., and St. Johns River Water Management District (Fla.). (1981). Analysis of residential demand of water in the St. Johns River Water Management District Palatka, FL.
130 Marella, R., (2 004). Water withdrawals, use, discharge, and trends in Florida, 2000 U nited States Geological Survey, Tallahassee, FL Mayer, P., and Aquacraft. (2005). "Tampa Bay Retrofit Study Trace Wizard Database." Aquacraft Inc. Boulder, CO. Mayer, P., DeOreo, W., H ayden, M., Davis, R., Caldwell, E., Mill er, T., and Bickel, P. (2009). Evaluation of California Weather based "Smart" Irrigation Controller Programs < http://www.aquacraft.com/Download_Reports/Evaluation_of_California_Smart_Controll er_Programs_ _Final_Report.pdf > a c c essed on December 2009 Boulder, CO. Mayer, P. and DeOreo, W (1999). Residential End Uses of Water American Water Works Association Research Foundation, Denver, C O Mays, L. (1996). Water resources handbook McGraw Hill, New York. Mays, L. (2002). Urban water supply handbook McGraw Hill, New York. Milesi, C., Running S Elvidge C. Dietz J. Tuttle B. Nemani R modeling Environmental Management 36(3), 426 438. Northwest Florida Water Management District (Fla. ). (2007). Regional water supply plan Region V, Franklin and Gulf Counties Northwest Florida Water M anagement District, Havana, Fla. Oliver, M. (2009). Personal Communication : Alachua County Property Appraiser's Ex ecutive Director of Valuation. Gainesville, FL. Peng, Z and Tsou, M (2003). Internet GIS : distributed geographic information services for t he Internet and wireless networks Wiley, Hoboken, N.J. Ragsdale, C. (2004). Spreadsheet modeling & decision analysis: a practical introduction to management science Thomson/South Western, Mason, Ohio. Sample, D. and Heaney, J. (2006). "Integrated Manage ment of Irrigation and Urban Storm Water Infiltration." Journal of Water Resources Planning and Management 132(5), 362 373. Scott, D. (1979). "On Optimal and Data Based Histograms." Biometrika 66(3), 605 610. Shaw, D. and Maidment, D. (1987). "Interventi on Analysis of Water Use Restrictions, Austin, Texas." Journal of the American Water Resources Association 23(6), 1037 1046. Singh, V., Jain, S., and Tyagi, A. (2007). Risk and reliability analysis : a handbook for civil and environmental engineers ASCE Press/American Society of Civil Engineers, Reston, Va.
131 SJRWMD. (2004). 2004 Land Use GIS Data Layer: lu_sjrwmd_200401 < http://www.fgdl.org/ > a c c essed on December 2009 FGDL, Florida Geographic Data Library, University of Florida, Gainesville, FL Smajstrla, A. (1997). "Basic irrigation scheduling in Florida." University of Florida Cooperative Extension Service, Institute of Food a nd Agriculture Sciences, EDIS, Gainesville, FL Stadjuhar, L. (1997). "Outdoor residential water use," M.S. Thesis, De pt. of Civil, Environmental, and Architectural Engineering, University of Colorado, Boulder. Estimating the crop coefficient of bahiagrass using lysimeters and the F AO 56 Penman M.S. Thesis, Dept. of Agricultural an d Biological Engineering University of Florida, Gainesville. US Census Bureau. (2000). "Florida County Census Block GIS Data." < http://www.fgdl.org/ > a c c essed on December 2009 Florida geographic data library, GeoPlan Center, University of Flori da, Gainesville, FL. US EPA Water Sense. (2008). "WaterSense Water Budget Approach and Tool." < http://www.epa.gov/watersense/specs/homes.htm > a c c essed on December 2009 Washington, D.C Vickers, A. (2001). Handbook of wa ter use and conservation Waterplow Press, Amherst, Mass. Whitcomb, J. (2006). "Evaluation of Irrigation Restrictions in East Central Florida." St. Johns River Wat er Management District, Palatka, FL. Whitcomb, J. ( 2005) "Florida water rates evaluation of single family homes. Final Report to Southwest Fl orida Water Management District and St. Johns River Water Management District. < http://libr ary.conservefloridawater.org/publications/15413502.pdf > a c c essed on December 2009 St. Johns River Water Management District, Palatka, FL; South West Florida Water Management District, Brooksville, FL. Xu, R., and Wunsch, D. (2009). "Clustering." John Wiley & Sons, Inc., Piscatawa y, N.J.. Zhou, W., and Troy, A. (2008). "An object oriented approach for analyzing and characterizing urban landscape at the parcel level." Int. J. Remote Sens. 29(11), 3119 3135.
BIOGRAPHICAL SKETCH John Palenchar was born in Stuart, Florida in 1973. He is married to Jessica Palenchar and ha s three sons: Basil, Edon, and Ezikai. He began his academic c areer in the year 2000 attending Manatee Community College in Bradenton, Florida where he graduated with an Associate of Arts in Engineering in December of 2003. He began his studies at the University of Florida the following year and graduated in December of 2007 with a Bachelor of Science in Environmental Engineering. He immediately began work on his Master of Engineering degree at the University of Flori da under Professor James P. Heaney. He c urrently work s with the C onserve Florida Water Clearinghouse and ha s with Gainesville Regional Utilities.