The URL can be used to link to this page

Browse Search

Home My WebLink About851238.tiff DEPARTMENT OF THE ARMY W¢ OMAHA DISTRICT CORPS OF ENGINEERS 6014 U.S. Post Office and Courthouse Omaha, Nebraska 68102 -. REPLY TO ATTENTION OF: February 8, 1985 Planning Division . V.74 PEa i 5 1985 ' f J Dear EIS Participant: A copy of Draft Technical Appendix 2, Future Water Demands, is enclosed for your review. This appendix considers past water usage in the Denver area, historic population trends, future population forecasts, and several water supply demand forecasting models that were developed specif— ically for the Denver region. Estimates of water supply needs for the Denver metropolitan area for a future 50—year period are presented in this appendix. In its final form, this appendix will reflect the water demand analysis for the Metropolitan Denver Water Supply Environmental Impact Statement (EIS). The future water demand estimates presented will be used in the evaluation of alternative water supply and the no federal action scenarios. The appendix will not be made final until interested members of the public, special interest groups, and governmental agencies have had the opportunity to review it and to present their comments to the Omaha District Corps of Engineers. Comments may be presented in writing directly to the Omaha District Engineer or orally at one of the three scheduled public meetings listed below. To be reflected in the EIS, comments must be received by the Omaha District Corps of Engineers office no later than April 23, 1985. Three public meetings will be held at the following times and places. 2:00 p.m., March 6, 1985 East Slope Regency Hotel (I-25 and 38th Avenue) Denver, Colorado 7:00 p.m. , March 6, 1985 East Slope Regency Hotel (I-25 and 38th Avenue) Denver, Colorado 7:00 p.m. , March 7, 1985 West Slope Lake Dillon Lodge Frisco, Colorado 851238 4 —2— Written comments should be sent to: District Engineer U.S. Army Corps of Engineers 6014 U.S. Post Office & Courthouse ATTN: MROPD (TASK 2) Omaha, Nebraska 68102-4978 Questions concerning the report or the review process may be sent to the District Engineer or to Mr. Gene Sturm who can be contacted at (402) 221-4629. Sincerely, gichard D. Gorton Chief, Environmental Analysis Branch U.S. Army Corps of Engineers Omaha District Enclosure SYSTEMWIDE/SITE-SPECIFIC ENVIRONMENTAL IMPACT STATEMENT METROPOLITAN DENVER WATER SUPPLY APPENDIX 2 SYLLABUS This technical appendix presents water demand forecasts for the EIS demand area annually through the year 2035. The study area includes most of Adams, Arapahoe, Denver and Jefferson Counties and parts of Douglas County and Boulder County. The water demand analyses were - initially completed in the Summer of 1983; additional research was performed related to certain key issues through the Fall of 1984. Historical water demand and related data were collected at both the water supplier and household level. Alternative water demand models and growth forecasts for the EIS demand area were developed. Water demand forecasts for public suppliers were developed for each model and each population growth series. Water demand for large self-supplied industrial water users is not included in this forecast. Historical water consumption data from 85 water suppliers for the period 1974 through 1982 were collected through the 1983 Regional Water Study Questionnaire and related efforts. Historical information describing economic and demographic conditions and other measures or factors which might influence water demand were collected. Use factors, defined as water demand by customer type, were developed. They are: Seasonal Water Demand Summer Use Winter Use Customer Type Use Factor Factor Factor (gallons/day) (gallons/day) (gallons/day) Single-family households (metered) 478 per household 258 220 Single-family households (flat-rate) 630 per household 400 230 Multifamily households 217 per household 37 180 Commercial/Industrial 45 per employee NA NA Public 14 per capita NA NA The water demand models, which project water demand at the water supplier level, are based on the historical data base and the use factors. The pooled water demand model relates total water demand per household to income, lot size, percentage of single-family dwelling, population per household, employment by place of work, marginal price of water, the existence of water restrictions and the number of days of measurable precipitation. The use factor water demand model applies the use factors to each individual customer type except for the single- family metered sector. This sector is projected on the basis of a regression model, developed at the household level, incorporating income, lot size, price and persons per household. Projections of population, households, and employment plus other factors in the models are required to implement the water demand models. After an extensive evaluation, the Denver Regional Council of Govern- ments projections of population, households and employment for the demand area and by water supplier to the year 2000 were adopted for the EIS. A modification in the in-migration assumptions was made to DRCOG projections beyond the year 2000, yielding three population growth series. Series 1 was selected as the most reasonable because it best represents a convergence of Denver metropolitan area growth rates with national growth rates over the long term. Population within the EIS demand area is projected to increase from 1.4 million persons in 1980 to 2.2 million persons by the year 2000 and to 3.5 million persons by the year 2035. Other forecasts include moderately increasing incomes, a trend toward smaller lots, and an increase in multifamily housing units. For the unconstrained water demand forecasts, it has been assumed that throughout the forecasting period that weather conditions will be the average of the historical record since 1947 (normalized weather conditions) , severe watering restrictions will not be imposed, and water rates will remain at the 1982 level. Conservation programs currently in place, such as evapotranspiration, are assumed to have the same effect on future customers as they have on present customers. The use factor model is selected as the central planning forecast tool because disaggregation by customer type is achieved. Disaggregation is viewed as crucial to the subsequent conservation analysis. Based on this model, water demand, unadjusted for natural retrofit, is projected to increase from 314,000 acre-feet to 703,000 AF by 2035: Water Demand Forecast for EIS Year Demand Area (acre-feet) 1980 314,000 1990 414,000 2000 515,000 2010 598,000 2035 703,000 TABLE OF CONTENTS Item Page CHAPTER 1 INTRODUCTION GENERAL 1 REPORT ORGANIZATION 2 STUDY OBJECTIVE AND SCOPE 3 DEFINITION OF THE EIS DEMAND AREA 5 i TABLE OF CONTENTS (Continued) Item hat STUDY METHODOLOGY 8 EVOLUTION OF THE WATER DEMAND FORECASTING TASK 8 COORDINATION WITH THE DENVER REGIONAL COUNCIL OF GOVERNMENTS (DRCOG) 9 COORDINATION WITH THE DENVER WATER DEPARTMENT AND OTHER WATER SUPPLIERS 10 GENERAL METHODOLOGY AND LIMITATIONS 10 GLOSSARY OF TERMS 13 CHAPTER 2 THE NATURE OF WATER DEMAND IN THE EIS DEMAND AREA INTRODUCTION 20 AN OVERVIEW OF WATER DEMAND WITHIN THE DEMAND AREA 22 DISTRIBUTION OF WATER DEMAND BY MAJOR SUPPLIERS 25 WATER DEMAND BY CONSUMER TYPE AND USE FACTORS 31 SEASONAL VARIATION IN WATER DEMAND 38 SEASONAL VARIATIONS IN TOTAL WATER DEMAND 38 SEASONAL VARIATION IN RESIDENTIAL WATER DEMAND 41 SELF-SUPPLIED WATER CONSUMERS 42 ii TABLE OF CONTENTS (Continued) Item Page CHAPTER THE HISTORICAL DATA BASE FOR WATER DEMAND MODELING INTRODUCTION 44 COLLECTION OF WATER DEMAND DATA AT THE SUPPLIER LEVEL 45 COLLECTION OF INDEPENDENT VARIABLES DATA 46 ADJUSTMENTS TO THE HISTORICAL DATA BASE 57 DESCRIPTION OF THE FINAL HISTORICAL DATA BASE 58 ANALYSIS OF WATER USE FACTORS 62 CHAPTER 4 POOLED WATER DEMAND FORECASTING MODEL INTRODUCTION 106 DEMAND MODELING PROCEDURES 107 SELECTION OF THE WATER DEMAND MODELS 108 THE POOLED WATER DEMAND MODEL 111 MODEL RELIABILITY 115 POOLED MODEL EVALUATION 126 ALTERNATIVE MODEL SPECIFICATION 128 iii TABLE OF CONTENTS (Continued) Item Paae CHAPTER 5 USE FACTOR WATER DEMAND FORECASTING MODEL INTRODUCTION 133 MODEL DEVELOPMENT 134 USE FACTORS 135 SOCIOECONOMIC ANALYSIS 135 IDENTIFICATION OF SOCIOECONOMIC VARIABLES 137 MODEL SELECTION 146 INITIAL FORMULATION 146 REFINEMENTS 147 MODEL PERFORMANCE AND LIMITATIONS 152 MODEL RELIABILITY 152 IMPLIED ELASTICITIES 153 ADJUSTMENTS FOR CONSERVATION MEASURES IN PLACE 155 LIMITATIONS 156 CHAPTER 6 ECONOMIC AND DEMOGRAPHIC PROJECTIONS FOR THE DENVER METROPOLITAN AREA INTRODUCTION 158 POPULATION, EMPLOYMENT AND HOUSEHOLD FORECASTS FOR THE EIS DEMAND AREA 161 iv TABLE OF CONTENTS (Continued) Item Page DENVER METROPOLITAN AREA FORECASTS 161 DESCRIPTION OF THE FORECASTING METHODOLOGY 169 DRCOG FORECASTS FOR 1980-2000 169 FORECASTING METHODOLOGY FOR 2000-2035 PROJECTIONS 185 EVALUATION OF THE POPULATION, EMPLOYMENT AND HOUSEHOLDS FORECAST COMPONENTS 192 POPULATION FORECASTS 192 EMPLOYMENT FORECASTS 203 OTHER FORECASTING ASSUMPTIONS 210 COMPARISON WITH OTHER FORECASTS 213 NATIONAL FORECASTS 213 COLORADO FORECASTS 214 DENVER REGION FORECASTS 219 CONCLUSIONS ABOUT THE ECONOMIC AND DEMOGRAPHIC FORECASTS 224 CHAPTER 7 SOCIOECONOMIC VARIABLE FORECASTS INTRODUCTION 227 FORECASTS OF MEDIAN HOUSEHOLD INCOME 228 HISTORIC TRENDS 228 INCOME PROJECTIONS 230 r V TABLE OF CONTENTS (Continued) Item Page SF LOT SIZE 236 SF AND MF HOUSEHOLDS 239 EIS DEMAND AREA FORECASTS 240 COMPARISONS WITH NATIONAL FORECASTS 241 SERVICE SECTOR AND NONSERVICE SECTOR EMPLOYEES 243 NUMBER OF DAYS OF MEASURABLE PRECIPITATION 244 MARGINAL PRICE 245 UNMETERED SINGLE-FAMILY DWELLING 245 PRESENCE OF THIRD DAY, THREE HOUR OUTDOOR WATERING RESTRICTIONS 246 CONSISTENCY OF ECONOMIC AND DEMOGRAPHIC FORECASTS 246 DRCOG PROJECTIONS AND ASSUMPTIONS 248 OTHER PROJECTIONS 250 CHAPTER 8 WATER DEMAND FORECASTS FOR THE EIS DEMAND AREA INTRODUCTION 251 POOLED MODEL WATER DEMAND FORECASTS 253 FORECASTING PROCEDURES 253 SUMMARY OF FORECASTS FROM THE POOLED WATER DEMAND MODEL 255 USE FACTOR WATER DEMAND MODEL FORECASTS 259 vi TABLE OF CONTENTS (Continued) Item Page FORECASTING PROCEDURES 259 SUMMARY OF FORECASTS FROM THE USE FACTOR MODEL 260 FORECASTS BY SUPPLIER 262 CRITICAL ASPECTS OF USE FACTOR MODEL FORECASTS 263 SELF-SUPPLIED WATER DEMAND 265 PROJECTIONS OF SELF-SUPPLIED WATER DEMAND 265 NATURALLY OCCURRING RETROFIT 266 COMPARISON OF POOLED & USE FACTOR MODELS 268 BACKCASTING CAPABILITY 270 SELECTION OF USE FACTOR MODEL 272 vii TABLE OF CONTENTS (Continued) LIST OF TABLES No. Title Page 1 WATER SUPPLIERS IN THE EIS DEMAND AREA 7 2 GLOSSARY OF TERMS 14 3 WATER DEMAND PER CAPITA AND PER HOUSEHOLD FOR THE DEMAND AREA, 1974 THROUGH 1982 24 4 WATER DEMAND BY CONSUMERS IN THE 10 LARGEST SUPPLIERS OF WATER WITHIN THE EIS DEMAND AREA, 1979 THROUGH 1982 26 5 AVERAGE DAILY WATER CONSUMPTION PER HOUSEHOLD WITHIN THE 10 LARGEST WATER SUPPLIERS IN THE EIS DEMAND AREA, 1979 THROUGH 1982 27 6 1982 WATER DEMAND FOR MAJOR WATER SUPPLIER CATEGORIES WITHIN THE DENVER SERVICE AREA 30 7 AN ESTIMATED BREAKDOWN OF 1982 WATER DEMAND BY CONSUMER TYPE, WITHIN THE EIS DEMAND AREA 31 8 THE PROPORTION OF SF AND OTHER DEMAND TO TOTAL WATER DEMAND FOR THE 10 LARGEST SUPPLIERS OF WATER IN THE EIS DEMAND AREA, 1982 33 9 AN ESTIMATED BREAKDOWN OF WATER DEMAND BY DETAILED CONSUMING SECTOR FOR SELECTED WATER SUPPLIERS, 1982 34 10 WATER USE FACTORS FOR THE EIS DEMAND AREA, 1974 Through 1982 36 viii TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title East 11 MAJOR WATER CONSUMERS USING DSA SUPPLIED WATER 37 12 SEASONAL WATER DEMAND AMONG RESIDENTIAL CONSUMING SECTORS IN THE EIS DEMAND AREA, 1974 to 1982 µ1 13 HISTORICAL WATER DEMAND PATTERNS OF SELF- SUPPLIED INDUSTRIAL COMPANIES IN THE DEMAND AREA 42 14 THE FINAL STATUS OF THE 1983 REGIONAL WATER STUDY QUESTIONNAIRE µ7 15 INDEPENDENT VARIABLES USED IN 23 ACADEMIC STUDIES OF WATER CONSUMPTION µ9 16 INDEPENDENT VARIABLES TESTED IN THE EIS WATER DEMAND MODELS 50 17 SELECTED USE FACTORS IN THE EIS DEMAND AREA OUTSIDE THE CITY AND COUNTY OF DENVER 64 18 SELECTED USE FACTORS FROM THE CITY AND COUNTY OF DENVER 65 19 AVERAGE DEMOGRAPHIC DATA FOR THE 1974 TO 1982 PERIOD, EXCLUDING 1977 66 20 SYSTEMWIDE DEMAND WEIGHTED USE FACTORS 67 21 SYSTEM LOSSES FOR SELECTED EIS DEMAND AREA SUPPLIERS 69 22 RESIDENTIAL USE FACTORS FROM THE METER/JOHNS HOPKINS DATA (g.h.d.) q1 ix TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 23 ANNUAL USE FACTORS FROM THREE-INCH METERING DATA 72 24 INDOOR WATER USE FACTORS (g.h.d.) FROM SECONDARY SERVICES 76 25 RESIDENTIAL USE FACTORS FROM THE RATIONAL APPROACH, g.h.d. 78 26 SUMMARY OF EDF DEMAND DATA FOR SINGLE-FAMILY HOUSEHOLDS (METERED) (MARCH 1976 - FEBRUARY 1977) 80 27 SUMMARY OF ENTERCOM DEMAND DATA FOR METERED DISTRICTS (JANUARY 1983 AND JULY 1984) 82 28 SUMMARY OF 1982 EDF AND ENTERCOM DEMAND AND SOCIOECONOMIC DATA FOR SINGLE-FAMILY METERED HOUSEHOLDS 86 29 WATER DEMAND FOR SF (METERED) AND MF HOUSEHOLDS IN AURORA (1982-83) 87 30 WATER DEMAND AND EMPLOYMENT FOR INVERNESS AND GREENWOOD PLAZA, 1974-1982 88 31 CONVERSATION PROGRAMS 89 32 SUMMER WEATHER DATA FOR STAPLETON AIRPORT DENVER, COLORADO 91 33 LONG TERM IRRIGATION REQUIREMENTS (INCHES/SUMMER) 92 34 WATER USE (GAL/HH/DAY) 95 35 CHARACTERISTICS OF ALTERNATIVE DEPENDENT VARIABLES 109 x TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 36 THE POOLED WATER DEMAND MODEL 112 37 DESCRIPTIVE STATISTICS OF THE POOLED EIS WATER DEMAND MODEL 113 38 ELASTICITIES FOR INDEPENDENT VARIABLES 116 39 SUM OF SQUARES ANALYSIS OF THE POOLED WATER DEMAND MODEL 117 40 BACKCASTING RESULTS FOR THE POOLED WATER DEMAND MODEL, 1974-1982 118 41 SIMPLE CORRELATION COEFFICIENTS OF THE AGGREGATE WATER DEMAND MODEL 120 42 GOLDFELD-QUANT TEST FOR HETEROSCEDASTICITY 122 43 CALCULATION OF T-STATISTICS FOR THE AGGREGATE WATER DEMAND MODEL 124 44 ADJUSTMENT IN WATER DEMAND FORECAST TO REFLECT NATURAL RETROFIT AND METERING 127 45 THE ALTERNATIVE POOLED DEMAND WATER MODEL 129 46 DESCRIPTIVE STATISTICS OF THE ALTERNATIVE POOLED WATER DEMAND MODEL 130 47 T-STATISTICS AND ELASTICITIES FOR THE ALTERNATIVE POOLED WATER DEMAND MODEL 132 48 REVIEW OF USE FACTORS 136 49 SOCIOECONOMIC VARIABLES - WATER USE CORRELATIONS 1982 EDF AND ENTERCOM DATABASE (DISAGGREGATED) 138 50 SOCIOECONOMIC VARIABLES - WATER USE CORRELATIONS 1982 EDF AND ENTERCOM DATABASE (AGGREGATED) 139 xi TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 51 SOCIOECONOMIC VARIABLES - WATER USE CORRELATIONS 1982 EDF AND ENTERCOM AND 1982 HISTORICAL SUPPLIER LEVEL DATABASE 141 52 OLS MODEL 144 53 HOUSEHOLD INCOME ADJUSTMENT FOR SFM HOMES 149 54 HOUSEHOLD SIZE ADJUSTMENT FOR SINGLE-FAMILY METERED HOMES 150 55 OLS COEFFICIENT ADJUSTMENTS 151 56 MODEL APPLICATION TO HISTORICAL DATA 154 57 CONSERVATION PROGRAMS 156 58 POPULATION FORECASTS FOR THE DENVER METROPOLITAN AREA 1980-2035 162 59 EMPLOYMENT FORECASTS FOR THE DENVER METROPOLITAN AREA 1980-2035 163 60 HOUSEHOLDS FORECASTS FOR THE DENVER METROPOLITAN AREA 1980-2035 165 61 POPULATION GROWTH FORECASTS FOR THE DENVER METROPOLITAN AREA, 1980-2035 166 62 EMPLOYMENT GROWTH FORECASTS FOR THE DENVER METROPOLITAN AREA, 1980-2035 167 63 HOUSEHOLDS GROWTH FORECASTS FOR THE DENVER METROPOLITAN AREA, 1980-2035 168 xii TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 64 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER DISTRICTS IN THE EIS DEMAND AREA, 1980 170 65 YEAR 2000 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER IN THE EIS DEMAND AREA 172 66 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER IN THE EIS DEMAND AREA, 2010, SERIES 1 PROJECTIONS 174 67 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER IN THE EIS DEMAND AREA, 2010, SERIES 2 PROJECTIONS 175 68 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER IN THE EIS DEMAND AREA, 2010, SERIES 3 PROJECTIONS 176 69 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER IN THE EIS DEMAND AREA, 2035, SERIES 1 PROJECTIONS 177 70 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER IN THE EIS DEMAND AREA, 2035, SERIES 2 PROJECTIONS 178 71 POPULATION, HOUSEHOLDS AND EMPLOYMENT BY WATER SUPPLIER IN THE EIS DEMAND AREA, 2035, SERIES 3 PROJECTIONS 179 xiii TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 72 REGRESSION STATISTICS FOR DRCOG EMPLOYMENT FORECASTS 181 73 DENVER METROPOLITAN AREA AS A SHARE OF U.S. POPULATION 192 74 DENVER METROPOLITAN AREA GROWTH AS A SHARE OF U.S. POPULATION GROWTH 1950-2035 194 75 NET IN-MIGRATION RATES INTO THE DENVER METROPOLITAN AREA, 1940-1982 195 76 AVERAGE ANNUAL NET MIGRATION TO THE DENVER METROPOLITAN AREA, 1940-2035 196 77 NET MIGRANTS TO THE DENVER METROPOLITAN AREA PER 10,000 U.S. POPULATION, AGES 20-34, 1940-2035 197 78 COMPARISON OF DRCOG FORECASTS AND ESTIMATES OF 1983 POPULATION FOR THE DENVER METROPOLITAN AREA 198 79 AGE DISTRIBUTION OF DENVER METROPOLITAN AREA AND UNITED STATES POPULATION 1980 AND 2000 201 80 AGE DISTRIBUTION OF DENVER REGION POPULATION 1980, 2000 AND 2035 202 81 UNITED STATES AND DENVER METROPOLITAN AREA PERCENTAGE OF POPULATION, 16 YEARS AND OVER IN THE LABOR FORCE, 1980-1995 203 xiv TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 82 HISTORICAL EMPLOYMENT AND DRCOG FORECASTS FOR THE DENVER METROPOLITAN AREA, 1980-2000 204 83 ESTIMATED 1980 to 1983 EMPLOYMENT GROWTH RATES FOR THE DENVER-BOULDER LMA 207 84 PERSONS PER HOUSEHOLD FOR THE DENVER METROPOLITAN AREA AND THE U.S. , 1940-2010 209 85 OTHER DRCOG FORECAST ASSUMPTIONS FOR THE DENVER METROPOLITAN AREA, 1980-2000 211 86 POPULATION FORECASTS FOR THE UNITED STATES, 1980-2030 214 87 EMPLOYMENT FORECASTS FOR COLORADO, 1960-2000 216 88 POPULATION AND EMPLOYMENT FOR THE STATE OF COLORADO, 1980-2010 217 89 POPULATION FORECASTS FOR THE DENVER METROPOLITAN AREA, 1980-2000 221 90 U.S. MEDIAN HOUSEHOLD INCOME IN CONSTANT DOLLARS, 1967 THROUGH 1982 229 91 REAL PER CAPITA PERSONAL INCOME FOR THE UNITED STATES, 1982-1984 231 92 ALTERNATIVE REAL HOUSEHOLD INCOME GROWTH FORECASTS FOR SELECTED AREAS, 1980-2035 232 xv TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 93 EMPLOYMENT PER HOUSEHOLD FOR THE DENVER METROPOLITAN AREA, 1980-2035 233 94 MEDIAN HOUSEHOLD INCOME PROJECTIONS FOR THE EIS DEMAND AREA 235 95 AVERAGE LOT SIZE OF NEW SINGLE-FAMILY UNITS WITHIN THE EIS DEMAND AREA 238 96 AVERAGE LOT SIZE OF NEW AND EXISTING SINGLE- FAMILY UNITS WITHIN THE EIS DEMAND AREA 238 97 AVERAGE UNIT MIX OF NEW HOUSING UNITS CONSTRUCTED, 1980-2035, FOR THE EIS DEMAND AREA 240 98 DISTRIBUTION MIX OF HOUSING UNITS BY UNIT TYPE FOR EIS DEMAND AREA 241 99 HOUSEHOLDS RESIDING IN SF AND MF DWELLINGS FOR THE EIS DEMAND AREA 241 100 AVERAGE UNIT MIX OF NEW HOUSING UNITS CONSTRUCTED, 1980-2000 FOR THE U.S. 242 101 PROJECTIONS BY PLACE OF WORK FOR THE EIS DEMAND AREA 244 102 UNMETERED SINGLE-FAMILY DWELLINGS IN THE EIS DEMAND AREA 246 103 INTERNAL CONSISTENCY OF ECONOMIC AND DEMOGRAPHIC FORECASTS 247 104 CHILDREN PER HOUSEHOLD AND CHILDREN PER EMPLOYEE, 1980 AND 2000 248 xvi TABLE OF CONTENTS (Continued) LIST OF TABLES (Continued) No. Title Page 105 PUBLICLY SUPPLIED WATER DEMAND IN THE EIS DEMAND AREA, 1980 TO 2035, BASED UPON THE POOLED MODEL 255 106 UNADJUSTED WATER DEMAND FORECASTS BY SUPPLIER GROUP, SHARE OF TOTAL PUBLICLY SUPPLIED WATER DEMAND 258 107 PUBLICLY SUPPLIED WATER DEMAND IN THE EIS DEMAND AREA, 1980 TO 2035, BASED UPON THE USE FACTOR MODEL 261 108 DISTRIBUTION OF FUTURE WATER DEMAND BY USER CLASS 262 109 SENSITIVITY ANALYSES OF SERIES 1 DATA 264 110 FORECAST OF SELF SUPPLIED WATER DEMAND OF MAJOR INDUSTRIAL CONSUMERS (m.g.) 266 111 ASSUMPTIONS FOR CALCULATING WATER SAVINGS FROM NATURAL RETROFIT 267 112 ACRE-FEET ADJUSTMENTS FOR NATURAL RETROFIT 268 113 PERFORMANCE COMPARISON OF USE FACTOR MODEL AND POOLED MODEL ON HISTORICAL DATA 270 114 COMPARISON OF ELASTICITIES FOR USE FACTOR MODEL AND POOLED MODEL 271 115 UNCONSTRAINED WATER DEMAND FORECASTS 274 r xvii TABLE OF CONTENTS (Continued) LIST OF FIGURES No. Title Page 1 WATER SUPPLIERS IN THE EIS DEMAND AREA 6 2 TOTAL WATER DEMAND FOR PUBLIC WATER SUPPLIERS IN THE EIS DEMAND AREA 1974-1982 23 3 MONTHLY AVERAGE TREATED WATER FOR THE DENVER WATER DEPARTMENT, 1971-1980 39 4 SFM VS. IRRIGATION (DCC) 97 5 DEPARTURE TREND (SFM IN DCC) 98 6 SFM VS. IRRIGATION (OUTSIDE DCC) 100 7 DEPARTURE TREND (SFM OUTSIDE DCC) 101 8 SFF VS. IRRIGATION 102 9 DEPARTURE TREND (SFF) 103 10 PAST AND PROJECTED POPULATION FOR THE DENVER REGION, 1930 TO 2035 160 11 ALTERNATIVE POPULATION FORECASTS FOR COLORADO, 1980-2030 218 12 ALTERNATIVE POPULATION FORECASTS FOR THE DENVER METROPOLITAN AREA, 1980-2035 220 13 WATER DEMAND FORECASTS FOR THE EIS DEMAND AREA, 1980-2035 252 xviii TABLE OF CONTENTS (Continued) EXHIBITS Item Page 2-A WATER DEMAND FORECASTS BY WATER SUPPLIER 275 2-B POOLED MODEL RESIDUALS AND INCOME ADJUSTMENT FACTORS 281 2-C ALTERNATIVE POOLED MODEL PROJECTIONS 284 REFERENCES CITED P85 mix CHAPTER 1 INTRODUCTION CHAPTER 1 INTRODUCTION GENERAL This technical appendix is one of several being prepared as part of the Metropolitan Denver Water Supply Systemwide-Site-Specific Environmental Impact Statement (EIS). This technical appendix describes and provides the basis for the water demand forecasts which are to be utilized in the EIS. The forecast of water demand is related to other EIS tasks includ- ing identification of existing supplies and potential future supply sources. The differences between future water demands and the existing Appendix 2 1 water supply aids in the determination of the requirements of and timing for development of new water sources. REPORT ORGANIZATION This technical appendix is organized in the following sequence: . The Syllabus presents a brief synopsis of the report findings. . Chapter 1 - The introduction presents a general description of the study, its purposes, its methodology, and pertinent background _ material. . Chapter 2 - The Nature of Water Demand in the EIS Demand Area provides a description of recent water demand patterns from 1974 through 1982 in the Denver metropolitan area. Total water demand throughout the demand area by major supplier, demand by consuming sector, and seasonal variations in water demand are identified. . Chapter 3 - The Historical Data Base For Water Demand Modeling identifies the historical data collected for the purposes of developing a water demand model for the Denver metropolitan area. The water demand data, the use factor analysis and the independent variables incorporated into the modeling process are discussed. . Chapter 4 - The Pooled Water Demand Forecasting Model identifies and describes the pooled water demand model developed for the EIS. Modeling procedures, reliability, elasticities, and the adjustments and refinements that were made to increase model accuracy are addressed. Appendix 2 2 . Chapter 5 - The Use Factor Water Demand Model identifies and describes an alternative water demand model for the EIS demand area. This chapter addresses the same topics found in Chapter Y as each relates to the use factor model. . Chapter 6 - The Economic and Demographic Projections for the Denver Metropolitan Area, provides an identification, analysis and evaluation of various projections of population, households, and employment for the Denver metropolitan area. One of three series of economic and demographic projections is selected and justified. An evaluation of the Denver Regional Council of Governments' (DRCOG) • economic and demographic forecasts is included in this chapter. . Chapter 7 - The Socioeconomic Variable Forecasts evaluates and selects the other independent variable projections or assumptions which are used to develop the EIS water demand forecast. . Chapter 8 - The Water Demand Forecasts for the EIS Demand Area presents the water demand forecasts from the various water demand fore- casting models and the three series of independent variable projections. Sensitivity analyses are conducted and then a single forecast is selected. STUDY OBJECTIVE AND SCOPE The objective of this EIS component is the derivation of water demand forecasts for EIS demand area consumers. The end product of this Appendix 2 3 effort is a forecast of unconstrained annual water demand on an annual basis through the year 2035. The water demand projection is unconstrained by the influences of water availability or other major discontinuities or restrictions. The term "unconstrained" represents a hypothetical circumstance where water prices and the level of conservation activities are held constant at their present levels. Weather influences are assumed to be equal to long-term averages. The effects of changing water prices on water demand will be addressed once future water supply scenarios are estab- lished in a later EIS task. The opportunities for saving additional water through conservation is also addressed in a separate subsequent EIS task. For the purposes of this EIS, conservation is viewed as a water supply option. In other words, the water demand planning forecast from Task 2 reflects a continuation of existing conditions. The scope of this task consists of a review of existing water demand patterns in the Denver metropolitan area, the development of practical, appropriate forecasting techniques, the preparation of reasonable projections for the forecasting models, and the application of forecasting techniques to produce water demand forecasts. Two functionally distinct water demand models were developed for considera- tion in this task. Projections of Denver metropolitan area growth were obtained, compared and thoroughly evaluated for their usefulness to this EIS. Modifications were performed on the population forecasts provided by the DRCOG; these new forecasts were incorporated into independent variable projections which drive the water demand forecasting models. A sensitivity analysis of the water demand forecasts was also conducted. Appendix 2 4 . DEFINITION OF THE EIS DEMAND AREA The EIS demand area is defined as the outer boundaries of Denver metropolitan area water suppliers which became signatories to the Denver Metropolitan Water Agreement in late 1982. A small number of households and businesses which supply their own water are included in the EIS demand area. The location of the demand area as part of the six county Denver region, or the Denver metropolitan area, is shown in figure 1 . The demand area currently accounts for about nine-tenths of the population of the six county region. All of Denver County and portions of Adams, Arapahoe, Douglas, and Jefferson Counties lie within the EIS demand area. A portion of Weld County, which includes the Town of Erie, is also included in the demand area. The urban areas of Boulder and Longmont in Boulder County are excluded from the EIS demand area because they are presently not participating in the Denver Metropolitan Water Agreement. The majority of the water service available in the EIS demand area is through public water suppliers. Public water districts are empowered to set tap fees and user charges or other levies which they deem neces- sary to provide water service within their respective jurisdictions. The public water suppliers identified within the EIS demand area are listed in table 1. The Denver Water Department's service area (DSA) is listed as a single supplier although it serves a large number of addi- tional distributors. EIS demand area water suppliers range from Florence Gardens which serves 50 persons to the DSA with a population of over 800,000 people. Within the boundaries of some water suppliers, there are a small number of subdistricts or subdivisions. For instance, Appendix 2 5 The , I ~j. ¢j . --• I o �( .1!c.\` v .rtes-- ._• i\ M1� laws. %, tea. i �`.' i - ` �t .i - t i r ,_„-... .,� r,. } - J • . A ,_ r . j ,l` 'Lf /, -Y - w�1. • �,r i M.O. r' rt e_ I 6 •rra�:� Ei°~ry _ aww �� - i \ -I. • • f" 4 _ i • S 0 on nun.M4 MO inns swum, "Li., , f�� ,---,' M •••••r r r.•••a•• D 'LLR Mr • all AVM s • ' k '� M -- .SUP a i IMAM Mal••S OM i M r•Nn.•• a .n••••r • .• rMAY M W.A.. � m. _ • Gaya Oa• u.-..• n-VW AM �..�.// Y : 60 Sal MIS AO a ISI.• a alai.MA IS al ISNIa MP a NMps Ora a l+. r —I—w k .// �! / c I .••••••••••Mal•••••1 .. OS/ 1 i •.-••11r••rr r n..rrr w.r.w•r.. ' anal••r ' •.•-—It ar r r•r•• a ....▪ .'•:••►•••••.:l M ••r.••••••.••••Ca"• WIDE EIS SNOWS 041 •. n•.0 w.•i •••• Itn.O•O.NAM 1011511 tats SUP•IY war Wan a • r Ma r r ••••••••••••••••••••rNM•w W.nunr W SAS MO Iliall Mall MO M°�'" Ih• „�.•„ WATER SUPPLIERS '^ • 'UMW Sala MO ON allial.~ Ma•r ' • 'n.. •�� THE SYSTEMWIDE •▪ rr /O111E tit onus EIS DEMAND AREA a Ollia•• •• r •1•••• n• Mal_Bala r Flee I Appendix 2 6 r. Table 1 Water Suppliers in the EIS Demand Area 1/— Arapahoe W&S Arvada Hazeltine Heights W&S Aurora Hi-Land Acres WA Beverly Hills MWA Hillcrest Villgge2/ Brighton Hilltop Acres Brook Forest WD Holly Mutual WD Broomfield Idledale WD Castle Pines MD Inverness W&S Castle Rock Lafayette Chaparral WAS Lincoln Park West MD Charlou Park WD Louisville Consolidated Mutual Louviers MS Co. Cottonwood WAS Maple Grove WA Crestview W&S Mission Viejo (Highlands Ranch) Denver Water Department Service Morrison Area Mt. Carbon W&S Denver SE Sub. W&S North Table Mountain W&S Dolly-O-Denver W&S Northglenn Eastlake W&S Orchard Hills WD East Cherry Creek Valley W&S Parker W&S East Valley W&S Roxborough Park MD Englewood Sedalia WD Erie Silver Heights W&S Evergreen MD South Adams County W&S Florence Gardens WD Stonegate MD Forest Hills MD Thornton Genessee WAS Thunderbird W&S Glendale View Ridge WA Golden Weisner Estates WS Greenwood Plaza Westminster Willows WD 1/ The following abbreviations are used in this table: W&S = water and sanitation; MWA = mutual water association; WD = water district; MD = metropolitan district; WA = water association. 2/ Hillcrest is a mobile home park and Hilltop is a nine-home single- family development. Water demand for these suppliers is accounted for on an individual household basis with Balance of Area. In gen- eral, the balance of area or unserved area represents unincorporated parts of the demand area. Source: Denver Regional Council of Governments and Denver Water Department, 1983. Appendix 2 7 within the Cottonwood Water and Sanitation District, the Sierra Vista subdivision and the Travois subdivision resell water to end users. In addition to the public water suppliers, there are a small number of industrial companies and individual households which provide their own water supplies. These are referred to as self-supplied consumers. The present and projected water demand for these self-supplied consumers has been identified but receives limited emphasis because supplies are assumed to be secure, according to interviews with these consumers. STUDY METHODOLOGY EVOLUTION OF THE WATER DEMAND FORECASTING TASK In order to fully understand the technical appendix, it is important to recognize the evolution of the research effort and the interim decisions which were reached at critical junctures. The task was initiated in December 1982 and originally scheduled for completion in July 1983. Following completion of the initial draft of the Task 2 Technical Appendix, a number of questions were raised about the results. The model did not provide water demand forecasts disaggregated or separated by consumer category. Concerns over certain statistical aspects of the analysis were also raised. The DRCOG economic and demographic projections used in the original water demand model were seriously questioned. After a period of reevaluation, it was decided that DRCOG population forecasts beyond year 2000 would be revised down- ward, and that a second modeling effort would be undertaken with the revised population forecasts. Appendix 2 8 The additional research and analysis related to water demand fore- casts began in the spring of 1984. Additional data related to household level water demands were collected and a water demand model based on water use factors was derived. The DRCOG population forecasts were modified, which necessitated changes in other economic and demographic components of the water demand models. Periodic reviews of interim research findings were conducted by the EIS Coordinating Committee, a group representing parties •interested in Denver metropolitan area water development issues. This document incorporates the new analyses as well as pertinent elements of the initial draft of the Task 2 Technical Appendix. COORDINATION WITH THE DENVER REGIONAL COUNCIL OF GOVERNMENTS The Denver Regional Council of Governments (DRCOG) made an important contribution to the development of water demand forecasts for the EIS demand area by directing the 1983 Regional Water Study Questionnaire. This survey was the primary data collection effort among Denver metropolitan area water suppliers upon which the demand model was based. DRCOG also prepared a map of the existing and future service area boundaries of the water suppliers within the demand area. DRCOG developed regional forecasts of population, households and employment for the years 1980, 1990, 2000, 2010 and 2035 and allocations of these measures to water suppliers for the years 1980 and 2000. These forecasts were carefully analyzed and evaluated as part of Task 2. Forecasts to the year 2000 were found to be acceptable and were adopted for the purposes of this EIS. Among others, one assumption dealing with the level of future in-migration was found to be unacceptable. Upon request, DRCOG agreed to develop new population forecasts with three alternative levels of in-migration. • Appendix 2 9 COORDINATION WITH THE DENVER WATER DEPARTMENT AND OTHER WATER SUPPLIERS The primary contribution of the Denver Water Department (DWD) and other Denver metropolitan area water suppliers to the demand forecasting task was their participation in the 1983 Regional Water Survey Question- naire. The pertinent information sought was historical water demand by consumer category to the greatest level of detail available. The majority of the water suppliers responded to this data request to the fullest extent possible. GENERAL METHODOLOGY AND LIMITATIONS RESEARCH METHODOLOGY Initial research consisted of the examination of published studies of water demand patterns and a review of historical water demand records of the water suppliers to understand the nature of water demand in the Denver metropolitan area. From this step, it was determined that recent, comprehensive information about past or present water demand for the Denver metropolitan area would need to be gathered as part of the Task 2 effort. It was also evident that a clear trend in total water demand within the study area was indiscernible during the 1970's. Water demand forecasting methodologies currently or previously used by water suppliers and others in the Denver metropolitan area were identified and evaluated. An extensive review of demand forecasting approaches which had been used throughout the United States was also conducted. Through this research, a general type of statistical analysis, termed regression analysis, was selected for examination of historical water demand and projection of future water demand. Appendix 2 10 r Analysis then proceeded to the testing of alternative.model forms with the available data and the formulation of the most suitable water demand forecasting models for the EIS. Regression analysis was employed in deriving the demand models. Availability of certain historical water demand data such as data dissaggregated by type of customers, was a con- straint on the form of the water demand forecasting model. Disaggrega- tion was finally achieved by developing use factors, or demand per unit per day, and projecting these use factors into the future. Projections of a number of economic and demographic variables were required to implement the water demand forecasting model. Data were gathered from DRCOG, from published information pertaining to the Denver metropolitan area, and from about 40 municipal and county planning departments in the region. LIMITATIONS The historical water demand data placed certain limitations on the analysis. Records maintained by the water suppliers typically cover brief recent periods, and this is especially true of household-level records. Consumer type (sector) breakdowns are not consistently defined by the suppliers. For example, a number of water suppliers could not distinguish between water demand by MF units versus commercial consumers. The limitations were overcome somewhat by the demand modeling pro- cedures. For instance, two different household data bases were merged, increasing the sample size and providing historical insight. The 1974 through 1982 data are averaged for the use factor modeling effort to avoid problems which would result from the small number of years repre- sented in the data base. In response to the problems with consumer sector definitions and record-keeping among water suppliers, the pooled Appendix 2 11 modeling effort adopted total water use per household as the focus of the demand forecasting. The projections of economic and demographic factors which comprise the independent variables in the water demand forecasts are subject to limitations inherent in any assumptions about the future. Regardless of the amount of research or analysis performed, questions will always remain about the underlying assumptions behind population forecasts of the Denver metropolitan area. Issues such as the area's attractiveness as perceived by businesses and individuals throughout the United States and the influence of the area's quality of life upon future population growth do not lend themselves to absolute or definitive analysis. The adopted forecasts are based upon the historical experience and a careful consideration of the underlying growth influences. Further, population forecasting is an ever-evolving process, pro- ducing new projections periodically. DRCOG developed New Policy fore- casts for the year 2010 publicized in the fall of 1984. These forecasts were not considered for the water demand projections because full docu- mentation and water supplier level population forecasts were unavailable for the timely completion of this technical appendix. As appropriate, subsequent consideration of these or other alternative forecasts can be reserved for later time. Given imperfect water demand forecasting techniques and growth forecasts, the accuracy of the water demand forecasts is probably greatest in the near future and may decrease over time. Forecasts to the year 2010 are likely to be more accurate than those for the year 2035, because the assumptions behind the forecasts have a greater oppor- tunity of deviation over the long term as compared with the near term. In addition, the relationships of the independent variables to water Appendix 2 12 demand will change over time. A deviation from the 1974 through 1982 experience will likely increase over the long term, that is beyond the year 2010. Sensitivity analyses and range estimates have been provided to place the variability of a future outcome in perspective with predic- tions set forth in this appendix. GLOSSARY OF TINS Table 2 provides definitions of selected, possibly unfamiliar terms which are used throughout this technical appendix. Particular attention has been given to statistical terms and those which are specific to the Denver metropolitan area. Appendix 2 13 Table 2. Glossary of Terms Term Definition Acre-foot of water 1 acre-foot of water equals 325,900 gallons or .3259 million gallons (m.g.) of water. Calibration Calibration, as used in this report, refers to the methods used to estimate a value for water demand per capita, per household or per acre. Cohort-survival model This population forecasting technique is based on the characteristics of specific age groups of males and females in the local area. People in different age groups have different mortality and fertility pat- terns (e.g., the mortality rate is high for elderly individuals, the fertility rate is highest for younger women). The model also takes into account aging of the population over time. The cohort-survival model, by itself, does not forecast in-migration and out-migration. Correlation coefficient, R2 This is a measure of the explanatory power of a regression equation. It varies from a low of 0.0 (meaning the model explains no variation in a dependent variable such as water consumption per household) to a high of 1.0 (meaning that the model explains all of the variation in the dependent vari- able). A model with an R2 of .50 explains one-half of the variation in the dependent variable. Cross-sectional analysis This type of regression analysis examines data from many different individuals, busi- nesses, or regions, for example, at one point in time. The regression model attempts to identify those characteristics associated with each observation (e.g. , household or water district) which are re- lated to the dependent variable. Appendix 2 14 Denver Water Department Service In this study, the DSA includes all areas Area (DSA) exclusively receiving water from the DWD. This includes the City and County of Denver, all total service and read and bill districts, and most master metered districts. Crestview, Broomfield, Consolidated Mutual, South Adams County, North Table Mountain, and Arvada are not included in this definition of the DSA. - Econometric model This is a regression model including economic factors often used for forecasting or policy analysis. ET Program ET stands for evapotranspiration, publicly known as a measure of the amount of water used by lawns. The ET conservation program instituted by the DWD in 1981 recommends the amount of water needed by a typical Kentucky bluegrass lawn in the Denver area to retain a certain level of lawn quality. The water requirements are reported on a daily basis through newspapers and the Denver television news programs. Every third day, three hour Many area water districts, including all of maximum watering restriction the DSA, instituted an outdoor watering programs restriction program during the summer of 1977. Households were designated as "circles," "diamonds," or "squares" depending on their street address. Outdoor watering was allowed for one group on one day on a rotating basis. Calendars charting the circle, diamond, and square watering days were mailed to local residents, and the group allowed to water was reported each day in the local news- papers and on television newscasts. Water- ing was allowed for a maximum of three hours. The program was enforced through a staff of monitors and through violation re- ports by neighbors. Warnings were given and fines were imposed for violation of the program. Appendix 2 15 Every third day watering This program was instituted in the summer restriction program months from 1978 through 1982 for many area water districts including all of the DSA. The program is identical to the every third day, three-hour-maximum program except there was no limit on hours of outdoor watering. Gallons per capita per day Calculated as total annual water demand (g.c.d.) within a water district divided by the population served within the district, and then divided by 365 days. Gallons per household Total annual water demand within a water per day (g.h.d.) supplier divided by total households served within the supplier and then divided by 365 days. Impervious area An area of land which does not absorb moisture. Interaction term This type of independent variable is cal- culated by multiplying one variable by another. The percent of SF homes in an area times the SF lot size is one example. The value of this variable for a water dis- trict with 90-percent SF households and 0.2 acre lots is 90 times .20 which equals 18. The value for a predominantly MF district with large SF lots might be 27 by 1.0 which equals 27. If the variable were significant, an explanation would be that a high average SF lot size won't increase water consumption if it is mostly MF development. Likewise, predominantly SF districts would not have high consumption if average single lot size is small. Marginal price The marginal price of water is the incre- ment paid for the last unit of water used. For example, the marginal price for an unmetered household is $0 because, after paying a lump sum monthly fee, the house- hold can use any amount of water at no fur- ther cost. Another household could be sub- ject to a rate structure in which a $10 flat monthly fee is charged and $1 is Appendix 2 16 charged for every 1,000 gallons of water above a 2,000 gallon allowance. Since almost all households use more than 2,000 gallons a month, the marginal price for this household would be $1. Master metered water districts These districts purchase their treated water supplies from the DWD. Each district sets its own rates and bills its own cus- tomers. Customers are required to be metered. Examples include Bancroft-Clover, Cherry Creek Village, Ken-Caryl, and Wheat Ridge. Observation An observation is a single data point in a statistical analysis such as a regression model. Water demand in one water district in 1976 is one observation, demand in that same district in 1977 is another observation, and water demand in a different district in 1976 is a third observation. Pooled time series, cross This is a regression model using a data sectional analysis base consisting of many different individuals, businesses, or regions over time. An example is water demand by households for each year from 1970 to 1980. This analysis would identify factors explaining differences in demand between household and over time. Publicly supplied water demand This is water provided by water dis- tributors who are selling to the general public. Public water consumers These consumers include public buildings and parks and any other public metered use. Public uses which are unmetered such as use of water for fire fighting are not included in the public water consumption estimates in this report. Read and bill water districts The DWD provides treated water, reads water meters, and bills the customers within these districts. Appendix 2 17 Regression analysis This statistical technique mathematically identifies those factors, known as indepen- dent variables, that are associated with the dependent variable. Water use, for example, may increase in years with little rainfall. Regression analysis of water use over 10 years might find an association between high water use and low rainfall. The technique would then express this relationship in mathematical terms. Self-supplied water consumers Water consumers who supply their own water by ground water wells or from surface water. Straight line interpolation A method of estimating values of an item of interest for years in which data are not available. For example, if employment equals 200 in 1980 and 160 in 1976, the difference in employment is 40 and the difference in time is 4 years. A straight line interpolation would assume that the annual change in employment between 1976 and 1980 is 40 divided by 4 equals 10 employees. The estimated values for 1977, 1978, and 1979 would be 170, 180, and 190, respectively. This technique also can be used to estimate future values such as water consumption in 1995 if projections were only made for the years 1990 and 2000. System losses System losses refer to the difference between the amount of water distributed by suppliers and the metered amount at the end use level. Also termed "unaccounted for water," system losses often include use of water for fire protection, construction, and sometimes unmetered public uses such as parks, as well as leaks in the distribution system. System losses also include any under registration of water use by water meters at the end use level. Time series analysis This form of regression analysis examines data over many years. The regression model would work to identify those factors which explain variation in the dependent variable over time. For example, water use in each year might be related to rainfall in a year. Appendix 2 18 Trend extrapolation The simplest form of this forecasting tech- nique is the drawing of a straight line be- tween data points for two years, and extending the line into the future to fore- cast demand. Another technique is to simply look at a time trend and make proj- ections based on visual inspection of the data. Total service water districts These water districts have assigned the full responsibility as a water district to the DWD. Customers are required to be metered and are generally charged higher water rates than the City and County of Denver. Appendix 2 19 '^ CHAPTER 2 THE NATURE OF WATER DEMAND IN THE EIS DEMAND ►REA CHAPTER 2 THE NATURE OF WATER DEMAND IN THE EIS DEMAND AREA INTRODUCTION This chapter briefly describes current water demand patterns within the EIS demand area as defined in the introduction of this technical appendix. The first portion of this chapter provides an overview of past and present water demand for the entire demand area, including total water demand as well as per capita and per household demand. Water demand as reported by major water suppliers in the Denver metro- politan area is also set forth. The second portion of this chapter addresses water demand by consumer type such as residential or major industrial users. Seasonal variation in water demand is described in Appendix 2 20 terms of SF and other water demand categories. The final element of this chapter examines self-supplied water consumers in the Denver metro- politan area. Self-supplied water demand is defined as water demand which is satisfied by a company's or a household's own supplies, such as a well. Almost all of the water demand information presented in this chapter was obtained from surveys of water suppliers in the Denver metropolitan area conducted by DRCOG. The DRCOG's 1983 Regional Water Study Questionnaire is discussed further in Chapter 4. In addition, water demand was estimated for a small number of water suppliers which did not respond to the formal survey and a small number of self-supplied households. Previous water demand studies for the Denver metropolitan area have limited application in this chapter because the geographic areas and the basis for measuring water demand differ substantially from this analy- sis. The water demand data presented in this chapter have been uni- formly translated to the amount of water delivered to the end user. In other words, unaccounted for water, system losses or any other dif- ferences from water diversion have been removed from the water demand estimates. Unaccounted for water is addressed separately from the water demand estimates. No distinction or accounting for consumptive use versus return flow has been attempted. A full accounting of water requirements in addition to end user requirements has been made in the EIS itself. Appendix 2 21 AN OVERVIEW OF WATER DEMAND WITHIN THE DEMAND AREA Recent historical trends in the amount of water delivered to end users in the demand area are illustrated in figure 2. These estimates are based upon the results of the 1983 Regional Water Study Questionnaire because some of the EIS demand area water distributors could not provide data for each year. The complexity of water use patterns in the EIS demand area is evident in comparing total water demand with demographic growth during the 1974 through 1982 period. Water demand ranged from about 255,000 acre-feet during 1977 " to 316,000 acre-feet in 1980. Population increased about 38 percent while total demand fluctuated in a narrow range, increasing slightly. This phenomenon is due to wide fluctuations in per capita and per household demand since 1974. Daily water use trends on a per capita and per household basis are presented in table 3. Total demand area water use per capita varied from 167 gallons per day in 1977 to 231 gallons per day in 1974. Total water demand per household ranged from 468 gallons per day in 1977 to 582 gallons per day in 1974. Appendix 2 22 4-7 cc 10 o O C f- G) 3 L 0 Y C (13 >siss‘. \ 0 GI L. Cll H O. O. ,.a c = CG .44 N \ d1 wr 3 C, 44 o W W 6 \\ \ e • E � D \ N I n v N W W > N \\ \ a i v GJ v, m v = w > r N IN Its to al [O •-I 3Gvr O I. \ Y u-a. q +' N \ r c..7 Ill S- 17.1 E cn a: o c O U1— \\1O 0 OO \\ \\\�\ N "4- in In N N 0 0 0 0 0 in O N _ (spuosnogl) Joa,, Jad lead—any Appendix 2 23 Table 3 Water Demand Per Capita and Per Household For the Demand Area 1974 Through 1982 Water Demand Water Demand Per Capita Per Household Year Per Day Per Day (g.c.d.) (g.c.d.) 1974 231 582 1975 185 1976 183 523 515 1977 167 468 1978 194 532 1979 179 483 1980 197 1981 183 522 479 1982 181 472 Source: U.S. Bureau of the Census, 1980; DRCOG, 1982; DRCOG, 1983; Water Providers Survey, 1983. The per unit water demand fluctuations are attributable to many factors which are believed to influence water demand patterns. For example, the relatively high g.c.d. in 1974 might be partially explained by the fact that this was a dry year, with only 31 days of measurable precipitation between May and September. The relatively low g.c.d. in 1982 might be partially attributable to the wet summer with 57 days of measurable summer precipitation. Three day, three-hour water restrictions probably reduced g.c.d. in 1977. Superimposed on these influences are fluctuating household income levels, a declining proportion of new SF dwellings, a reduction in lot sizes, and increasing water rates. The water demand models developed for the EIS demand area explore the relationship of the Denver metropolitan area water demand to these and other independent influences. Appendix 2 24 DISTRIBUTION OF WATER DEMAND BT MAJOR SUPPLIERS The 10 largest water suppliers within the demand area accounted for 93.5 percent of the total water demand in 1982. Similar data were not available from all of these suppliers for the years from 1974 through 1982. Estimates of water demand within the major water districts in the demand area from 1979 through 1982 are provided in table 4. The DSA accounted for 64 percent of the total demand in 1982, a reduction from approximately 69 percent in 1979. Of the 10 major water suppliers, Aurora experienced the most rapid increase in total water demand from 1979 through 1982, about 2,000 m.g. or a 26 percent gain. Average daily water demand per household for the 10 major suppliers from 1979 through 1982 is presented in table 5. Each supplier shows an increase in water demand per household between 1979 and 1980, followed by lower water demand per household in 1981 and 1982. This might be explained by dryer weather in 1980 as compared with 1981 or 1982 or various conservation measures in evidence during the latter years. The range of daily water demand per household among suppliers in 1982 is 315 gallons per day (g.p.d.) for Westminster to 536 g.p.d. for the DSA. This range is indicative of many factors, including the large industrial and commercial base, flat-rate customers and relatively low prices evident in the DSA and absent in Westminster. Each water supplier in the EIS demand area exhibits a different mix of influences of water demand, such as income levels or SF household lot size. Differences in water consumption per household among water suppliers and over time is the focus of the water demand forecasting effort described in later chapters of this technical appendix. Appendix 2 25 Table 4 Water Demand by Consumers in the 10 Largest Suppliers of Water Within the EIS Demand Area, 1979 Through 1982 Millions of Gallons (MG) Water District 1979 1980 1981 1982 Denver Service Area 1/ 60,400 66,600 62,500 62,900 Aurora 7,700 ?/ 9,023 8,008 9,700 Arvada 4,331 4,797 4,949 4,566 Consolidated Mutual 3,242 3,690 3,568 3,328 Westminster 5/ 2,707 3,217 3,179 3,094 Englewood 2,388 2,770 2,638 2,500 ?/ Thornton 3,428 3,852 2,394 2,498 Northglenn -- 4i -- 4� 1 ,539 1 ,582 South Adams County 1 ,144 1,321 1 ,327 1 ,252 Broomfield 1,096 1,343 1 ,139 1,128 Total of 10 Largest Suppliers 86,400 96,600 91 ,200 92,500 Other Public Water Suppliers 5,900 6.500 6,600 6,600 Total Water Demand (MG) 92,300 103,100 97,800 99, 100 Total Water Demand (AF) 5/ 283,200 316,400 300,100 304,400 t/ Does not include raw water sales to districts or treated water sales to Broomfield, Consolidated Mutual, Crestview or South Adams County. These districts also have other water sources and resell the DWD water to consumers in their districts. This adjustment is made to avoid double counting. V Estimate based on other years' data for this district and annual trends among all districts. 5/ Includes water sales to Shaw Heights and Federal Heights. 1 Included in Thornton in 1979 and 1980. 5/ An acre-foot (AF) of water is equal to 325,900 gallons. Source: Water Providers Survey, 1983. Appendix 2 26 Table 5 Average Daily Water Consumption per Household Within the 10 Largest Water Suppliers in the EIS Demand Area, 1979 Through 1982 Average Daily Water Consumption per Household (Gallons) Water District 1979 1980 1981 1982 Denver Service Area ' 544 586 543 536 Aurora 424 ?/ 473 372 . 416 Arvada 408 446 455 417 Consolidated Mutual 394 444 428 396 Westminster 21 387 407 351 315 Englewood 496 574 543 499 ?/ Thornton 348 386 358 368 Northglenn -- 4/ -- 4i 440 448 South Adams County 388 437 391 357 Broomfield 442 523 403 393 ?/ Does not include raw water sales to districts or treated water sales to Broomfield, Consolidated Mutual, Crestview or South Adams County. These districts also have other water sources and resell the DWD water to consumers in their districts. This adjustment is made to avoid double counting. ?/ Estimate based on previous year's data for this district and annual trends among all districts. 2/ Includes water sales to Shaw Heights and Federal Heights. 4i Included in Thornton in 1979 and 1980. Source: Water Providers Survey, 1983 and Annual Household Estimates by Water District, 1983. Appendix 2 27 The DSA is defined for the purposes of this report as those water suppliers which receive all of their water supplies from the DWD. Those districts which purchase part of their treated water needs or raw water needs from the DWD are accounted for as separate districts in this report. The DWD is responsible for ensuring a dependable supply of water to the residents of Denver County by charter and to others through contractual commitments. The DWD which provides water for the DSA is by far the largest water supplier in the demand area in terms of total water delivered to end users. There are four major categories of water districts within the DSA depending upon the type of service contract each may have with the DWD. . The City and County of Denver is a single entity which is served by the DWD. . Total service water districts have assigned the full responsi- bility as a water district to the DWD. Total service district customers are required to be metered and are generally charged higher water rates than those within the City and County of Denver. The financial and operational data for the total service districts are comingled with one another so that a single total service district is indistinguishable. . Read and bill districts are somewhat more autonomous than total service districts, and they take care of their own system. The DWD pro- vides the water, reads the water meters, and bills the customers within the read and bill districts. Activities for each read and bill district are individually accounted for by the DWD. Read and bill water district customers are required to have water meters and are charged higher rates than the total service district customers. Appendix 2 28 . Master-metered water districts are the most autonomous category of the DSA. Master-metered water districts purchase all of their treated water supplies from the DWD, but otherwise operate as independent water districts. Each master-metered district sets its own water rate structure and bills its own customers. Customers within master-metered water districts generally have individual water meters, and their rates are usually higher than the other categories of DSA dis- tricts. In addition, the DWD regularly sells treated water supplies to four districts in the demand area: Consolidated Mutual, Broomfield, Crestview and South Adams County. These differ from other master- metered districts, primarily in that they also obtain water from sources other than the DWD. The DWD also sells raw water supplies to Arvada and the North Table Mountain Water District. Since these six water dis- tricts also obtain water from sources other than the DWD and to avoid double counting, each is treated independently in the subsequent water consumption and demand analysis. The consumption patterns of the four major categories of water districts within the DSA are detailed in table 6. The City and County of Denver accounted for 68 percent of total DSA water demand in 1982. Water demand per household varies considerably among the four categories of districts in 1982. These differences are attributable to economic and demographic factors and water rates, among others. Read and bill districts have higher personal income levels than the other DSA supplier groups. The City of Denver has smaller lot sizes and lower income levels counter balanced by a larger commercial base and unmetered households. These influences are examined in the ensuing demand analysis in Chapters 4 and 5. Appendix 2 29 Table 6 1982 Water Demand for Major Water Supplier Categories Within the Denver Service Area 1/ 1982 Total Water Demand in millions of gallons (MG) City and County of Denver 42,900 Total Service Districts 5,300 Read and Bill Districts 7,100 Master-Metered Districts 7,600 Total for DSA 62,900 Total for DSA (acre-feet) 2/ 193,000 Total Water Consumptions per Household (gallons per day) City and County of Denver 549 Total Service Districts 522 Read and Bill Districts 639 Master-Metered Districts 383 1/ Does not include raw water sales to districts or treated water sales to Broomfield, Consolidated Mutual or Crest- view. These districts also have other water sources, and resell the DWD water to consumers in their districts. This adjustment is made to avoid double counting. 2/ One acre-foot equals 325,900 gallons. Source: Water Providers Survey and supplemental data sub- mitted by the Denver Water Department, 1983; and sur- veys of master-metered water districts, 1983. It is recognized that the above 1982 data are estimates, whereas actual results will be forthcoming in the 1983 Annual Report of the Denver Water Department. Appendix 2 30 WATER DEMAND BY CONSUMER TYPE AND USE FACTORS An estimate of total water demand for broad consuming sectors pro- vides further insight into water demand patterns in the Denver metro- politan area. Sectors include single-family (SF) dwellings, multifamily (MF) dwellings (apartments, condominiums, and so forth) and mobile homes, commercial users, industrial users, and public consumers. A breakdown by consuming sector has been estimated for the entire demand area and for representative individual suppliers to indicate the degree of variation among individual water suppliers. Finally, individual water consumers have been identified, and their proportion of the total water demand has been estimated. The distribution of water demand by consumer type is order of magnitude as determined from responses from the 1983 Regional Water Study Questionnaire. An estimate of 1982 water demand by consuming sector among the water suppliers in the demand area is provided in table 7. Table 7 Estimated Breakdown of 1982 Water Demand by Consumer Type, Within the EIS Demand Area Percent of Total Consuming Sector Water Demand SF 65 MF and Mobile Home Parks 14 Commercial and Industrial 16 Public 5 Total 100 Source: 1983 Regional Water Study Questionnaire. Appendix 2 31 Water demand among SF dwellings or residences, including both indoor and outdoor use, comprised almost two-thirds of the water demand in the EIS demand area during 1982. MF and mobile home water use represented a combined 14 percent of water demand during 1982; water use for these subcomponents of the residential sector were not separated because they cannot easily be distinguished by most water suppliers in the Denver metropolitan area. Most of the commercial and industrial water demand, which together comprised 16 percent of the total, is believed to be in the commercial sector. Certain large industrial water consumers provide a portion of their own water supplies. These supplies have not been included in these calculations. The commercial sector includes mostly retail, services, and office activities. Accounting for five percent of the total water 1982 demand, the public sector consists of parks and use in other publicly owned structures. Among the various water suppliers in the demand area, there is a wide range in the proportion of water demand by consumer category. One indication of this range is the percentage of SF to total water demand among the 10 largest suppliers of water in the demand area, estimated from the 1983 Regional Water Study Questionnaire. As indicated in table 8, SF water demand ranges from 58 percent to 89 percent of total water demand for the 10 largest suppliers of water. r Appendix 2 32 Table 8 The Proportion of SF and Other Demand to Total Water Demand for the 10 Largest Suppliers of Water in the EIS Demand Area, 1982 Water Supplier SF All Other Total (%) ($) (%) Denver Service Area 60 40 100 Aurora 61 39 100 Arvada 1/ 89 11 100 Consolidated Mutual 65 35 100 Westminster 63 37 100 Englewood (1981) 72 28 100 Thornton 69 31 100 Northglenn 75 25 100 South Adams County 59 41 100 Broomfield 58 42 100 1/ Includes townhouses in SF demand. . Source: 1983 Regional Water Study Questionnaire. A further breakdown of historical non-SF water demand by consuming sector is available in a consistent format for only a small number of water suppliers in the demand area. Of the 10 largest water suppliers, only five suppliers offered a detailed estimate of non-SF water use in the 1983 Regional Water Study Questionnaire as presented in table 9. Use factor estimates (water demand by customer type, stated in g.p.d.) have been derived from several data bases and information sources and combined to derive a point estimate for a particular sector. This process considers the origin of each factor, what the supporting data base represents, the sample size of the data source, and the conditions that would cause an estimate derived from one source to vary from a generalized estimate. The use factors were developed as an average of the 1974 through 1982 historical data base of water demand (a period approaching average long term precipitation) and related measures available for the EIS demand area. The estimate for each sector is Appendix 2 33 x �m o 0 0 0 O 4.)i O 0 0 0 0 F .O1 O O 0 0 . m a 4 I 1 •— N I I es 1 I I ° A. L *+ o ^+ r7• 0 c K �I ,I m .0 co a. m a Co C el .d 7 al ea c C.3• 0 el 1st 111 M O I I O • N 'O "1t. I 1 .-1 0 O N a. K NI NI A .mcn l N 7-- r CO in a 0. 0 ° .° d 0 07C w- e g . m L .--I m la Qa'411 40 El I 0 I I en 1 O F a) m o a S 1 ma m w U 1 sic �'M m Z o (...E �. 'd o m N m en S. m CO a a1 .-1 . 1a d Ogl S Let 'a en in CO .1 S. 43 7 C CO CO3 L m • C C > • 0) 4 m L +-1 .C M m W C. .I • 44 L O +1 O L /1 .0 0) W .m1 0. . el 0 +' c 4 V a 4-1 C S O 0. ...1 c o t. O B m 43 d] 7 O 'C W 01 CO CIS o co N 0 .4 .Od CO m m m m 'p° c V 03 V d .°i - °i M CO c O N .- C per_, C C C 0 +1 O 9 .C eg MI C C C U >a 43 m CO a a' O L. 0 W C 0 5 Co CO �I NI MI O Appendix 2 34 made under the constraint that the resultant predicted composite demand of all suppliers equals 262 m.g. , the total annual water demand averaged over the 1974 through 1982 period. The basis for these use factors is thoroughly discussed in subsequent chapters of this technical appendix. The use factors applicable to the EIS demand area over the 1974 through 1982 period are presented in table 10. The differences in SF flat-rate versus metered use factors are attributable to different rate structures and socioeconomic characteristics of flat-rate customers, largely in Denver and Englewood, as compared with metered SF customers. MF dwellings require less water primarily because they typically do not have individual lawns. There are a small number of very large individual water consumers in the demand area. In the 1983 Regional Water Study Questionnaire, water suppliers were asked to identify major customers and recent water demand data. The DSA had eight water customers with 1980 demand of over 100 m.g., as shown in table 11. The Public Service Company of Colorado is the largest with 624 m.g. (less than 2,000 acre-feet). This demand is in addition to the Public Service Company's requirements of its own water supplies. Appendix 2 35 i� Table 10 Water Use Factors for the EIS Demand Area, 1974 Through 1982 SEIS Total Use Regional Percent Number Factor Demand of Consuming Sector of Units (gals/unit/day) (MGD) of Total Single family households 226,600 478 102 39 (metered) housing units Single family households 96,900 630 60 23 (flat rate) housing units Multifamily households 191 ,400 217 45 17 housing units Commercial and industrial 703,000 45 34 13 (employment) employees Public (population) 1,380,000 14 21 8 population Appendix 2 36 Table 11 Major Water Customers Using DSA Supplied Water 1980 Demand of DSA Customer Supplied Water (m.g.) Public Service Company of Colorado (four power plants) 624 Lowry Air Force Base 597 Gates Tire and Rubber Company 311 Samsonite P98 Federal Center 231 Martin Marietta 225 Metro Sewer 119 Rocky Mountain Arsenal 112 Total for Eight Largest Consumers 2,517 Total in Acre-Feet 7,723 Source: DWD, 1983 The largest eight consumers in the DSA accounted for less than 8,000 acre-feet or about one-tenth of the 1980 DSA non-SF demand and about four percent of the total 1980 DSA water demand. Western Electric Company, with 1980 water consumption of 59 million gallons, was the only other large water consumer identified in the questionnaire. Western Electric is supplied by the City of Westminster. Based upon available data, large individual water consumers do not account for a large portion of water demand in the demand area. Fur- ther, there appears to be no pattern of economic activity or particular industry which dominates the Denver metropolitan area's water demand. Appendix 2 37 SEASONAL VARIATION IN WATER DEMAND Water demand in the Denver metropolitan area exhibits highly seasonal variation during a normal year. Demand is greater during the summer months, primarily because of outdoor lawn watering. The degree of seasonal variation in water demand is also dependent upon summer rainfall and temperature levels. Seasonal variation in total SF and non-SF water demand is discussed separately below. The seasonal component of the Denver metropolitan area's water demand is most pronounced in the SF sector, although seasonal variations are also evident in the non-SF consuming sectors. Monthly water demand data were not obtained from the majority of water suppliers in the demand area. Monthly water demand information by consuming category is even more sparse. As a result, insights about seasonal water use must be gained from a few water suppliers with com- patible record keeping systems and a large, diversified customer base to produce meaningful results. The DWD, Aurora, and Englewood were selected to provided insights about seasonal variation in water demand. For this purpose, owing to their diversity, they are considered generally representative of the demand area. SEASONAL VARIATIONS IN TOTAL WATER DEMAND Monthly water demand as a proportion of total annual demand from DWD treatment plants and averaged over the 10 year period from 1971 through 1980, is indicated in figure 3. These data, as reported in the DWD 1981 Annual Report, provide the best available insight into seasonal Appendix 2 38 cji.! • -& N :,.; N r, n \�\�`\�\\� O a c:i \\N \\\, . .\\\ N Si 49 w Ch I-I PG ii w w 0' i r \3 t;1! \\\ CU 43 0. Ca N \ ta L cu Lt. ci \\ N L .' Q {r1.1 7 n \� I t I I I I I I I I I i ILei 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 . . . . . . . . . . . . . . . . . . o n t0 6 d M N — O O 00 n tD v1 d M N O asn Ja}oM fonuuy to}ol ;o }uaoiad Appendix 2 39 variation in total water demand, for an extended period of time. Peak monthly water demands occur during the months of June, July, and August; approximately 43 percent of the total water demand occurs during these summer months, according to the DWD's figures. The months of May and September are also important outdoor irrigation months. The five month May-through-September period accounted for more than 62 percent of annual water demand during the 1971 through 1980 period. A modest amount of irrigation water demand also appears to occur during April and October. The remaining months--January, February, March, November and December--represent a low period of demand during which little or no outdoor watering occurs. The average amount of water treated monthly at DWD plants for these five traditionally low demand months over the 10 year period approximates 3.3 billion gallons. If the average water demand of 3.3 billion gallons for the months in which outdoor watering is minimal is subtracted from the remaining seven months of treated water production, an estimate of outdoor lawn irrigation or other seasonal, summer-related demand may be derived. Based upon the DSA treated water production data between the years 1971 and 1980, lawn watering and other summer-related demand is estimated to be 42 percent of the total water demand. Other personal or summer-related demand would include increased tourism, construction activity, or other sources of seasonal demand. Although single-year analyses of the seasonal water demand phenomonon can be misleading, Aurora's 1982 monthly consumption data provides further insights into the magnitude of monthly water demand variation (unpublished data from the City of Aurora's Utility Department files, 1983). Aurora water demand during June, July, and August was almost 45 percent of the 1982 water demand. The May through September period accounted for 66 percent of the 1982 demand, and the April r Appendix 2 40 through October period represented 79 percent. Based upon the same technique used to calculate outdoor irrigation or other summer-related water demand for the DSA, an estimated one-half of Aurora's total 1982 water consumption was devoted to these seasonal purposes. SEASONAL VARIATION IN RESIDENTIAL WATER DEMAND As part of the previously discussed use factor analysis, an examination of summer and winter water demand was performed for each of the residential sectors: SF metered (SFM) , SF flat-rate (SFF) , and MF. A diverse number of data sources were used in this analysis, and these will be discussed in detail in subsequent chapters of this appendix. The results of the use factor analysis which relates. to seasonal water use is presented in table 12. Table 12 Seasonal Water Demand Among Residential Consuming Sectors in the EIS Demand Area 1974 through 1982 Sector Season Factor (g.h.d.) SF households Winter 220 (metered) (SFM) Summer 258 Total 478 SF households Winter 230 (flat-rate) (SFF) Summer 400 Total 630 MF households (MF) Winter 180 Summer Total 217 Appendix 2 41 Winter use is derived first as a base, and then subtracted from the total water use factor to produce a summer use rate. (Difference in sector water use in the summer might reflect different lawn sizes or socioeconomic characteristics as well as differences in rate structure.) SELF-SUPPLIED WATER CONSUMERS Self-supplied water consumers include those businesses and house- holds which provide and consume their own water supplies. Individual companies which divert water for their own needs comprised more than 32.6 billion gallons or 100,000 acre-feet of water demand in 1980, as indicated in table 13. Table 13 Historical Water Demand Patterns of Self-Supplied Industrial Companies in the Demand Area Millions of Gallons 1970 1975 1980 Public Service Company of Colorado 12,300 23,000 18,000 Adolph Coors Company 8,700 14, 100 14,200 Gates Rubber Company 880 880 300 Cooley Gravel Company 120 120 120 Total 22,000 38,100 32,620 Total in Acre-Feet 67,500 116,900 100,100 Sources: Personal communication with selected company officials, 1983; COE, 1978; DRCOG, 1983. Appendix 2 42 Public Service Company, Coors, and Gates also purchase additional water from public water suppliers in the demand area. There are a small number of residences and businesses inside the demand area which depend upon shallow ground water wells. Based upon a review of the number of shallow ground water wells in the demand area and DRCOG's estimate of the number of households not served by public water districts, an estimated 10, 100 self-supplied households or busi- nesses exist in the demand area. This figure also includes a small num- ber of mobile home parks within public water supplier boundaries which rely upon ground water wells. Water demand for these residences and businesses is estimated to be 2,900 m.g. or almost 8,900 acre-feet in 1980. Appendix 2 43 CHAPTER 3 THE HISTORICAL DATA BASE FOR WATER DEMAND IODELING r CHAPTER 3 THE HISTORICAL DATA BASE FOR WATER DEMAND MODELING INTRODUCTION This chapter describes the historical data base used to develop water demand forecasts for the EIS demand area. The selection of cer- tain types of data is substantiated. The data collection process is described, and the results are presented. The time frame for the his- torical data base was established as the 1974 through 1982 period because this was the period for which water demand data were considered to be the best available. The historical data series encompass water demand data, economic and demographic characteristics relating to water use, weather data, Appendix 2 44 water rate information, and conservation measures in place. These data were compiled for each water supplier in the demand area. Certain data base adjustments, necessary to exclude water demand observations unrepresentative of water demand patterns in the region, are described and justified. Ranges, means and standard deviations of key variables are examined. Subsequent to the original data collection effort, a second data collection effort focusing on water use by consuming sector and seasonal demand patterns was initiated. This use factor analysis is described at the end of this chapter. The original data collection effort was conducted at the water supplier level, whereas a major element of the use factor analysis entailed gathering household level data suitable for use in water demand forecasting. COLLECTION OF WATER DEMAND DATA AT THE SUPPLIER LEVEL A survey was conducted among the water suppliers in the EIS demand area to obtain water demand data and other information which would be useful for the water demand forecasting effort. Recognizing that the selection of a water demand forecasting approach and its eventual suc- cess would be heavily dependent upon the quality of the historical water demand data base, considerable emphasis was placed on this data gather- ing activity. The survey was conducted under the direction of DRCOG with the intent that it would be used for an update of DRCOG's Regional Water Study as well as for the EIS water demand forecasts. The 1983 Regional Water Study Questionnaire was prepared and mailed to Denver Appendix 2 45 area water suppliers in early December 1982. Followup efforts, including personal interviews, were performed to verify responses and facilitate the provision of water demand data for the EIS. Each response was examined for consistency and accuracy. Responses were actively sought until 11 April 1983. Additional responses after that date were considered unlikely. The questionnaire requested detailed data on water demand, the number of customers, and revenue by consuming sector for the 1974 through 1982 period. Based on test surveys, the maximum level of detail was sought. The response to the survey is summarized in table 14. Although responses were received from all of the large water suppliers, a substantial proportion of the respondents were not able to report water use by type of consumer. In addition, the EIS demand area sup- pliers could rarely supply water demand data for all of the years requested. Based on personal communications with the water suppliers that offered incomplete responses, it is believed that additional water demand data were either impractical to gather or unobtainable. COLLECTION OF INDEPENDENT VARIABLES DATA Independent variables, also known as explanatory variables, are those factors or outside influences which might have an effect on water demand levels. The selection of the independent variables to be tested in the analysis was based upon a review of explanatory variables used in Appendix 2 46 4.1'3 _ , 00j0 . © _ 4.1 as • ! i% ' . a ate ! N ` �} - � 5 �2 ) Com ■ a 2 . ! |2!$ X. 0L I . _ , (�2 aa '`i o. It 0 • SE .0O3 0 2 • _ | - ■ | • ) a l. § ■ k��\, fa 2 •'f !] .2a2i -4 / f| . ° k �§� \ ! a & • ! | O. ) " h .4 0 �k %! & N a . .- it a4 ` `e •a. .cU 8. ! ! • ■ Va i � ° I ) o 0 ;\ sa ° I.I. a L. a ■ t go 0 a !�•\ i_ O � \! w • �.4 is \ !▪ \; | ; � | f% � �2 \ ! | X* 4 •Viz as 0 Append ay 47 previous water demand models throughout the United States, plus addi- tional variables which were thought to be relevant to the Denver metro- politan area. The independent variables which have been used in a representative sample of other water demand models and the number of studies in which they were applied are listed in table 15. The independent variables most often applied include price variables, weather variables, household income, persons per household, housing unit value, lot size and time. Almost all of these variables, or similar measures, were tested in the development of the EIS water model. Certain variables, such as housing unit value, were not considered in the EIS model because accurate estimates or projections could not be derived. The variables that were initially considered for the water demand model are listed in table 16. This list of variables encompasses a broad range of influences which might possibly affect water consumption in the Denver metropolitan area, for which data are available, and which can be measured quantitatively and objectively from reliable sources. At the same time, independent variables must be amenable to future assumptions or projections in order to be considered in the water demand model. For each of the major variable categories, such as price or weather, a number of related variables were tested to ensure that the most relevant measures might be selected through the regression analy- sis. Once the specific independent variables were selected for con- sideration, the next step was the compilation, review, and organization of historical data. An extensive effort was undertaken to gather data for each independent variable for each water district in the Denver metropolitan area for the period of 1974 through 1982. Appendix 2 48 Table 15 Independent Variables Used in 23 Academic Studies of Water Consumption Frequency Independent Variable of Use Demographic Characteristics: Persons per households 8 Adults per household 1 Children per household 1 Age of head of household 1 Percent of population under 18 1 Education of head of household 1 Race 1 Income Measures: Household or per capita income 11 Housing unit value 4 Retail sales per capita 1 Regional cost of living index 1 Housing Unit Characteristics: Water using fixtures 2 Conservation devices 1 Rooms per person 1 Percent of homes with hot water 1 Percent of homes with plumbing 1 Average water pressure 1 Age of dwelling unit 2 Outdoor Watering: Lot size 3 Outdoor watering index 1 Non-Residential Use: Employees per account 1 Industrial production per capita 1 Price of Water: Marginal price 15 Average price or average bill 7 Weather: Rainfall 11 Temperature 5 Evapotranspiration 4 Other: Time (Years, Months, etc.) 3 Source: Gottlieb (1963), Gardner and Schick (1964), Howe and Linaweaver (1967), Turnovsky (1969) , Way (1970), Grima (1972) , Young (1973) , Morgan (1974) , Clark and Goddard (1974), Berry and Bonem (1974) , Hogarty and McKay (1975), Camp (1978), Gibbs (1978) , Danielson (1979), Foster and Beattie (1979) , Hughes and Gross (1977) , Agthe and Billings (1980), Carver and Boland (1980) , Morris and Jones (1980) , Beattie and Foster (1980) , Gallegher et al. (1981) , Hanke and deMare (1982) and Billings (1982). Appendix 2 49 Table 16 Independent Variables Tested in the EIS Water Demand Models Demographic Characteristics: Population Households Persons per household Population in households per household 1— / Median age of population Percent of population under 19 Percent of population 60 or older Median Household Income Housing Unit Characteristics: Percent of total housing units which are single family dwellings Percent of units built before 1950 Percent of units built after 1970 Percent of dwelling units rented Lot Size of Single Family Dwellings Economic Activity: Total employment by place of work Service sector employment Retail sector employment Other employment Non-service sector employment Price of Water: Marginal price Average monthly bill Weather: Annual precipitation April-September precipitation May-September precipitation Precipitation up to one inch in one day (May-Sept.) Number of days of measurable precipitation (May-Sept.) Average daily temperature (May-Sept.) Cooling degree days Conservation Measures: ET conservation program Every third day 3 hour maximum watering restrictions Every third day watering restrictions 11 Excludes persons residing in group gr p quarters. Appendix 2 50 DEMOGRAPHIC CHARACTERISTICS The basis for the population and household estimates for each water supplier by year was the 1980 Census data for census tracts. DRCOG used this information to estimate the population and number of households for each water supplier. The DRCOG estimates of 1980 population and households were compared with information reported in the 1983 Regional Water Study Questionnaire and with data obtained from local government planning officials. Certain DRCOG estimates were revised based upon this review. Population and household estimates were derived for the years 1974 through 1979 and 1981 through 1982 based upon information from a number of sources. Population and households in 1970 were calculated for each water supplier by allocating 1970 census tracts to water suppliers. Straight-line interpolation of 1970 and 1980 population and households generated an initial estimate of population and households by year. Data available from local planning officials were also analyzed. Annual number of building permits were studied as were the annual number of water taps reported in the 1983 Water Providers Survey. The best available information was applied to make annual population and house- hold estimates for each supplier by year. Each of the age-related demographic characteristics indicated in table 16 was estimated for each water supplier using a simple straight- line extrapolation between 1970 and 1980 Census values. Appendix 2 51 MEDIAN HOUSEHOLD INCOME Median household income for the Denver metropolitan area water sup- pliers was estimated from the Census tract allocations for 1970 and 1980. Interpolations for the years 1974 through 1979 were based on annual income growth rates as reported by the U.S. Bureau of Economic Analysis for each county in the Denver metropolitan area. Constant real incomes were assumed for the years 1980 through 1982. Annual median household income figures were translated into constant 1980 dollars using the Consumer Price Index (Survey of Current Business, selected years) . HOUSING UNIT CHARACTERISTICS The percent of total housing units served by a water supplier which were SF detached dwellings was estimated based upon data from the 1983 Water Providers Survey, the DWD 1978 Land Use Inventory, personal communication with municipal and county planning officials, and building permit data. These sources were used because the 1980 Census did not include data on SF units appropriate for this analysis. Other data, such as age and percent of dwelling units rented, were derived from Census data for 1970 and 1980 and interpolated on the basis of household growth. SF LOT SIZE Average SF lot size for the purposes of demand modeling is defined as the number of acres per SF unit in a water supplier's service area. The measure includes non-irrigated areas suchas the dwelling and drive- Appendix 2 52 ways but does not include roads or alleys. This variable was estimated and verified for each water supplier through a number of different sources including the DWD 1978 Land Use Inventory, an examination of aerial photographs of the Denver metropolitan area, and personal com- munications with local governments, assessor offices, and planning departments throughout the demand area. More than 40 municipal and county planning officials were contacted to derive estimates of lot size. Lot size was assigned a zero in those districts such as mountain communities where little or no lawn watering is conducted or allowed. Average lot sizes are single-point estimates from data focusing on the year 1980. Additional lot size estimates by supplier for other years were considered impossible to obtain on a valid basis. Values are assumed to be constant for the 1974 through 1982 period for each water supplier. This assumption is reasonable according to personal communications with local assessors and planning officials in the Denver metropolitan area. ECONOMIC aunnTY Various employment measures, including total employment and employ- ment by economic sector, are indications of nonresidential water use for water suppliers. Employment by sector is important because different economic sectors might exhibit different water demand patterns. Employ- ment data are by place of work rather than by place of residence. These data were obtained from DRC0G surveys of Denver metropolitan area employers in 1976 and 1980. The DRC0G estimates by census tract were allocated to water districts. Interpolations for the 1974 through 1982 period were based on a knowledge of the area and the 1976 through 1980 trends. Appendix 2 53 PRICE OF WATER The effects of water prices on demand patterns have been the sub- ject of a number of studies throughout the United States, and widely varying opinions concerning the influence of price on water demand are evident. Two different price variables were considered: marginal price and the average monthly bill. Water rate structures were obtained from the 1983 Regional Water Study Questionnaire and annual surveys of water rate structures provided by the DWD. The EIS price variables reflect 1982 dollars, adjusted by the U.S. Consumer Price Index. Marginal price and average monthly bill were calculated on the basis of average monthly demand per SF dwelling or residential account as reported in the 1983 Regional Water Study Questionnaire. A single figure of monthly water demand per SF account was used. The rate structure of each water supplier was compared to the average demand to ascertain the marginal price. This was done for each year from 1974 through 1982. Water rates for SF residential customers in the summer season were selected for the marginal price analysis. The average monthly bill for a SF residential customer was also calculated using the monthly water demand for a SF residence and the residential rate structure for that water district in that year. Flat-Rate Versus Metered Households. A flat-rate customer has a marginal price of water equal to zero because there is no charge for the last 1,000 gallons consumed. The average bill for unmetered households Appendix 2 54 was calculated on the basis of the average household for the water supplier (e.g., number of rooms in the house, number of bathrooms, size of lot, and so forth). An additional price variable was calculated by transforming marginal price into a weighted average for water suppliers with a mix of metered and flat-rate residential customers. For example, a district might have a price per 1,000 gallons of $1 for metered customers, and 60 percent of residential customers might be metered. The weighted average marginal price used in this example would be 60 percent times $1.00 plus 40 percent times $0, or $0.60. WEATHER VARIAR1.Rg A substantial number of weather-related variables were examined in formulating the water demand model. Five different forms of rainfall measures were tested along with three different measures of temperature. The range of weather variables which have previously been found by the DWD or others to have a significant effect on water demand was con- sidered for the analysis. Through regression analytical techniques, the precipitation or temperature variables which best explained variations in water consumption would be selected in the model formulation effort. Weather data were obtained from the National Weather Service, National Oceanic and Atmospheric Administration, for the weather station at Stapleton International Airport. Weather data were assumed to apply equally to all water suppliers in the Denver metropolitan area for a given year, because published historical weather data for each portion of the Denver metropolitan area were unavailable. CONSERVATION MEASURES The conservation measures selected for examination in the EIS water demand analysis were programs or measures which could be objectively Appendix 2 55 identified for each water district. Hence, public education programs are not amenable to a statistical-type regression analysis. Three con- servation programs were identified for the 1974 through 1982 period which could be objectively identified for a number of water suppliers. The ET conservation program, the every third day, three-hour watering restriction, and the every third day watering restriction were selected for evaluation of associated effect on water demand. Dummy variables were used to denote the presence or absence of a conservation program for a water supplier in a particular year. A dummy variable is simply a yes-no response which refers to some particular characteristic of an independent variable. If a conservation program was in practice by that water supplier in that year, the number 1 was assigned as the value of the variable; a zero was assigned if the pro- gram was not in force for that water supplier in that year. MODIFICATIONS FOR THE USE FACTOR MODEL Two additional steps were taken to modify the data base for use in the use factor model. First, selected economic and demographic charac- teristics were estimated for the unmetered and metered portions of the City and County of Denver and Englewood. The observations representing average characteristics of unmetered and metered households were replaced in the data set by a set of observations for unmetered and metered households within each city. The 1983 Regional Water Study Questionnaire was a key factor in formulating these estimates. The second modification was to average each water demand variable and certain economic and demographic variables over time for the indi- vidual water suppliers in the data base. The year of 1977, a period of every third day, three-hour watering restrictions, was not included in this averaging. All other years were averaged to obtain a single value for each district for each variable. If data were only reported for one year, each economic and demographic variable in that year would be attributed to that water supplier. Appendix 2 56 ADJUSTMENTS TO THE HISTORICAL DATA BASE The water demand models developed for the EIS were designed to be tools for projecting future water demand in the EIS demand area. As such, water suppliers whose customers exhibit fundamentally different water demand patterns from more typical EIS demand area water consumers have been excluded from the data base. In other cases, observations in a single year were rejected if they appeared to be very inconsistent with data from other years. Classification of supplier observations as "outliers" is made on the basis of outdoor water use patterns and employment base. Generally, an observation was considered to be an outlier if it was two standard deviations or more from the mean value. OUTDOOR WATER USE/LOT SIZE Because outdoor water use is a major component of total water demand in the region, those water suppliers exhibiting outdoor water use patterns fundamentally different from typical use in the demand area were excluded from the data base. For example, households in water districts with very large lots were considered unrepresentative of typical outdoor use. In order to exclude these districts, those areas with average SF lot size of one acre or more were removed from the data base. Also excluded were households in mountain areas which often leave most of their lot area in native vegetation and tend to have little or no outdoor use. In addition, concentrations of households which significantly rely upon well water and ditch water for lawn irrigation were identified. Water suppliers fitting any of these three criteria were excluded from the data base. Appendix 2 57 EMPLOYMENT BASE Solely commercial or industrial areas or water suppliers serving primarily nonresidential customers were also considered unrepresentative of typical EIS demand area water demand. For example, 15 employees work within the North Pecos Water District for every one household residing within this district. DESCRIPTION OF THE FINAL HISTORICAL DATA BASE WATER DEMAND Water demand at the end use level ranges from 2.92 m.g. for Panorama Park in 1977 to 47,499 m.g. for the City and County of Denver in 1974. Water demand was divided by the number of households in order to compare water demand patterns of large and small suppliers. For example, the number of households varied from 26 (102 people) in Panorama Park to 202,300 (525,000 people) in the City and County of Denver in 1974. Demand per household per day in these two supply areas was 308 gallons for Panorama Park in 1977 and 643 gallons for the City and County of Denver in 1974. Average g.h.d. is 492.3 for those districts in the data base. The standard deviation is 148.9. Demand ranges from 241 g.h.d. for Glendale in 1974 to 968 g.h.d. in Bow-Mar in 1974. INDEPENDENT VARIABLES Several independent variables were identified as key influences on water demand in the EIS demand area. The historical data for these variables are briefly described. Appendix 2 58 Persons Per Household. Household size averaged 2.98 persons among the various water districts during the 1974 through 1982 period. This measure of household size excludes persons living in group quarters. The standard deviation was 0.49 persons per household. Household size ranged from 1 .46 persons per household in Glendale in 1980 to 4.02 persons per household in the Holly Mutual supply area in 1982. Water demand on a household basis is expected to be higher in areas with greater numbers of persons living in each household. Median Household Income. Real median household income averaged $23,900 (in terms of 1980 dollars) among the different water districts during the 1974 through 1982 period. The standard deviation was $6,600. Income ranged from $9,304 in Glendale in 1978 to $75,000 in Holly Mutual in 1982. Higher incomes were expected to be associated with higher water demand, particularly greater outdoor water use. Percent SF. Water districts ranged from almost entirely MF units to entirely SF dwellings. Single units averaged 76.4 percent of the dwellings among the different water suppliers during the 1974 through 1982 period. The standard deviation was 22.6 percent. Since outdoor water use is greater for SF dwellings, water demand is expected to be higher in areas which consist predominantly of SF dwellings. Estimation of the use factors also required number of households living in SF units and households residing in MF or other dwellings. These figures ranged upward to 115,000 SF households and 98,800 MF households in the City and County of Denver in 1982. Appendix 2 59 SF Lot Size. The mean SF lot size is 0.281 acre per unit among the districts in the data base for 1980. This represents about a 12,000- square-foot lot. The standard deviation is 0.141 acre per unit (6, 100 square feet). The SF lot size (nonoutliers) ranges from 0.137 acre per unit (6,000 square feet) in Thornton to lots approaching one acre in size in Bow-Mar. Service Employment Per Household. Service employment per household is calculated by dividing the number of service sector employees working at establishments located within a water supply area by the number of households residing within that area. This scaling is done in order that the relative employment bases among suppliers could be considered. For example, Wheatridge with 2,950 service employees and 5,385 house- holds has a greater relative service employment base (0.55 service employees per household) than the City and County of Denver with 85,200 service employees and 213,920 households (0.40 service employees per household) . Service employment averages 0.164 employees per household among all suppliers over the historical time period. The standard deviation is 0.150. Nonservice Employment Per Household. Nonservice employment per household is calculated by dividing employment other than service sector employment of establishments within a water supply area by the number of households within that area. Entirely residential suppliers have a zero nonservice sector employment per household. The average for all sup- pliers over time in the data base is 0.994 nonservice employees per household. The standard deviation is 0.774. The use factor model also requires total employment in each supply area. This ranges from zero in entirely residential areas to 454,000 workers in the City and County of Denver in 1982. Appendix 2 60 Marginal Price of Water. The marginal price of water (in 1982 dollars) was $2.34 in Northglenn in 1982 and $0.27 in Sedalia in 1981. The marginal price was $0.00 in Holly Mutual and Silver Heights, where all households are unmetered and thus pay no charge related to the amount of water required. Parker households are metered but as of 1982 were not charged a price per thousand gallons demanded. The marginal price for suppliers such as Englewood and the City and County of Denver is a weighted average of a zero marginal price for unmetered households and the appropriate price per thousand gallons for metered SF customers. Further analysis in the use factor model subdivides these two suppliers into unmetered and metered customers. Number of Days of Measurable Precipitation. Summer rainfall directly affects the amount of lawn irrigation in a given year. Although many measures of summer rainfall are possible, the number of days which had at least 0.01 inches of precipitation from May through September was found to best relate to water demand patterns. Days of measurable precipitation varied from 31 days in 1974 to 57 days in 1982. The average number of days of precipitation during the summer months for the entire data base is 44.0 (about one out of four days over the period) with a standard deviation of 7.8. Relatively "dry" years included 1974 and 1978. "Wet" years are 1975 and 1982. Precipitation for other years is closer to the long term average of 44 days of measurable precipitation since 1947. Presence of Third Day, Three Hour Outdoor Water Restrictions. Every third day, three hour maximum outdoor watering restrictions were instituted in 1977 over all of the Denver Service Area and most other area water districts. The variable takes a value of 1.0 for those dis- tricts in that year. The variable has a value of 0.0 for districts without the restriction in 1977. Appendix 2 61 ANALYSIS OF WATER USE FACTORS An important part of the historical data base development effort was the derivation of water use factors. These use factors refer to water demand by specific customer categories, on a seasonal basis where possible. The use factors are a key element of the use factor water demand model. The consuming sectors and units of measure addressed are: SFM or single family metered (g.h.d.); SFF or single family flat-rate (g.h.d.); MF or multifamily (g.h.d.); E or commercial and industrial (gallons per employee per day); and P or public (g.c.d.). Water use factors are estimated on the basis of several information sources. An analysis was made of all the factors for a particular sector to derive a single factor. An analysis of each source as well as temporal and climatic influences was completed. USE FACTORS FROM THE 1983 REGIONAL WATER STUDY QUESTIONNAIRE The goal from reviewing this data source was to estimate the water demand for all sectors in the EIS demand area and to calculate the associated system losses. To provide an independent estimate of the water demand, the results of the 1983 Regional Water Study Questionnaire were analyzed for each water supplier. The historical data base described earlier in this chapter was used to provide demographic information. Methodology. Data from the Denver metropolitan area water suppliers were used to estimate historical daily water demand for each of the years 1974 through 1982. The City and County of Denver was analyzed separately because Denver has a different use pattern (higher industrial and commercial use) and a higher percentage of flat-rate homes than the surrounding suburbs. In addition, disaggregation of Appendix 2 62 commercial and MF data was not possible for the City and County of Denver. For some suppliers, SF households included duplex residences. In other estimates of water use, duplex residents were considered MF households. Mobile homes were considered MF households in all analyses. Tables 17 and 18 depict year by year, sector by sector use for outside the City and County of Denver and within, respectively. To provide a frame of reference for tables 17 and 18, the average historical data base (1974 through 1982, excluding 1977) is shown in table 19. The household numbers and total water demand data were also com- pared to the production and system loss data given in the 1983 Regional Water Study Questionnaire to verify that the data did represent actual end use demand numbers. Five of the 15 districts used for the SFM household estimates provided data for every year from 1974 through 1982. Calculations for the other demand sectors had similar occurrences of missing data. Depending on the year, between five and 13 districts were incorporated into the yearly estimate of demand. Because of the large variation in the characteristics of each provider's service area, a weighted average was used to calculate the yearly demand estimates. Based on engineering judgment and follow-up interviews with sup- pliers, a unit weighted average of SF household demand (tables 18 and 19) was used in estimating systemwide use factors for SFM and SFF house- holds. Use factors for MF households, P and E were estimated by averag- ing demand over the years to get use sector. These uses were then weighted by total demand to obtain the systemwide percentage. Knowing the number of units in each category, a systemwide use factor was then calculated. Estimated factors and discussion. Table 18 and 19 summarize the resulting yearly demand data. The average yearly demand varied from 422 to 531 g.h.d. for SFM households outside the City and County of Denver Appendix 2 63 Table 17 Selected Use Factors in the EIS Demand Area Outside the City and County of Denver SFMI/ SFF2/ El/ Pp/ Use Use Use Use Year Units Factor Units Factor Units Factor Units Factor 1974 63,900 531 -- -- 23,200 47.3 81 ,300 5.3 1975 ' 66,100 470 -- -- 25,600 47.8 81 ,100 6.0 1976 68,600 469 8,500 627 27,500 46.9 81 ,400 7.3 1978 88,400 508 8,470 709 47,400 50.5 137,000 7.6 1979 97,900 437 8,470 501 57,800 48.7 160,000 8. 1 1980 102,900 489 8,460 604 110,000 60.9 168,000 9.5 1981 108,000 422 -- -- 132,000 69.4 385,000 7.6 1982 108,000 422 -- -- 132,000 69.4 385,000 7.6 Range 63,900- 422- 8,460- 501- 23,200- 46.9- 81,100- 5.3- 108,000 531 8,500 709 132,000 69.4 385,000 9.5 1/ Denver Read and Bill, Denver Total Serve, Broomfield, Consolidated Mutual, Crestview, East Cherry Creek Valley, Erie, Hi-Land Acres, Northglenn, North Table Mountain, Parker, Sedalia, Thornton, West- -^^ minter, Willows. 2/ Englewood. 1/ Broomfield, Consolidated Mutual, Crestview, East Cherry Creek Val- ley, Erie, Louisville, North Table Mountain, Parker, Thornton, West- minster, Willows. Note: Gallons per unit per day is the unit of use factor measure. Insuf- ficient multifamily data available for summarization. Source: 1983 Regional Water Study Questionnaire. Appendix 2 64 Table 18 Selected Use Factors from the City and County of Denver SFM SFF P Use Use Use Year Units1/ Factor2/ Units1/ Factors2/ Units1/ Factors2/ 1974 22,700 613 88,700 735 525,500 22.0 1975 23,200 552 88,400 684 518,800 19.8 1976 23,600 536 88,200 665 511 ,000 21.5 1978 25,000 611 88,000 677 498,600 24.1 1979 25,900 519 87,800 604 496,600 19.9 1980 27,100 580 87,600 636 492,400 24.4 1981 27,200 538 87,300 557 494,900 23.5 1982 28,000 501 87,200 507 496,800 21.6 Range 22,700- 501- 87,200- 507- 492,400- 19.8- 28,000 613 88,700 735 525,500 24.4 1/ Units: SFM = single family households (metered) ; SFF = single family households (flat); P - public 2/ Gallons/per unit/per day. Note: Disaggregation of E and MF was not possible. Source: 1983 Regional Water Study Questionnaire. Appendix 2 65 Table 19 Average Demographic Data for the 1974 to 1982 Period, Excluding 1977 Inside Outside DC&C DC&C Total Single family households 25,400 201,200 226,600 (metered) (11%) (89%) Single family households 87,900 9,000 96,900 (flat rate) (91%) (9%) Commercial and industrial 405,000 298,000 703,000 (58%) (42%) Public 504,000 876,000 1 ,380,000 (37%) (63%) Source: Historical data base. Appendix 2 66 and from 501 to 613 g.h.d. inside the City and County of Denver; from 501 to 709 g.h.d. for flat-rate customers in Englewood and from 507 to 735 g.h.d. for flat-rate customers in the City and County of Denver. A unit weighted average of 478 g.h.d. for SFM and 630 g.h.d. for SFF was calculated. The tables also have ranges for E as well as for P use. The E ranges are from 46.9 to 69.4 gallons/employee/day and represent providers outside the City and County of Denver; the corresponding City and County of Denver data include MF data and could not be disaggregated. The P use ranges are 5.3 to 9.5 gallons/capita/day outside the City and County of Denver and 19.8 to 24.4 gallons/capita/day inside Denver. Table 20 presents the demand weighted systemwide use factors. The fraction of demand was obtained by weighting sector use by total demand. A use factor was calculated based on the number of units. Table 20 Systemwide Demand Weighted Use Factors Fraction Sector of DemandL Use Factor (gal/unit/day) SF households (metered) (SFM) .39 451 SF households (flat-rate) (SFF) .23 622 MF households (MF) .17 233 Commercial & industrial (E) .13 48 Public (P) .08 15 ?/ Total Area Water Demand (Q) = 262 million gallons per day. Source: 1983 Regional Water Study Questionnaire. System losses. System losses affect demand forecasting in the EIS demand area at two junctures. The first is that in addition to meeting Appendix 2 67 projected demand, water providers must size facilities to accommodate projected losses. Second, in Denver and Englewood, significant numbers of residences are unmetered which requires technical judgment to allo- cate produced water to either unmetered residences or losses. Loss estimates were average (weighted by total demand) for the EIS demand area. The most recent estimates of losses as shown in the Water Providers Survey and subsequent communication with DWD, Aurora; Thornton and Highlands Ranch are presented in table 21. Losses range from 5.5 percent in Englewood to 21.8 percent in Glendale in recent years, with a systemwide average of 7.9 percent. Some inconsistencies in the data exist. For example, the City of Aurora includes water use associated with construction in losses while the City and County of Denver does not. THREE-INCH METER/JOHNS HOPKINS USE FACTORS The objectives of this analysis were to estimate water use rates for SFM, SFF and MF households, and to extrapolate the use rates for flat-rate households to make an estimate of the Denver Water Department (DWD) system loss. The DWD estimates a system loss of six percent of production. Water demand and selected demographic and socioeconomic characteristics are from the Three-Inch Meter Study and Johns Hopkins Study. Methodology. The Three-Inch Meter Study (DWD, 1980, 1981, 1982) and John Hopkins Study (1982), collected weekly demand data for specific residential areas in the City and County of Denver for three years (1980 to 1982). The data were aggregated by season to compare the demand from SFM, 3FF and MF households and to compare seasonal variations within each sector. Appendix 2 68 Table 21 System Losses for Selected EIS Demand Area Suppliers Percent Supplier System Applicable Demand Water Supplier Loss Year (MGD)L/ Aurora?/ 11 1982 24.4 Arvada 11 1981 12.1 Broomfield 11.8 1981 32. Consolidated Mutual 7.8 1981 9.3 Crestview 11 1981 2.1 Denver5/ 6.04/ 1982 152.1 East Cherry Creek Valley 8.2 1981 0.3 Englewood 5.54/ 1981 7.1 Glendale 21.8 5/ 0.6 Golden 8.0 5/ 2.4 South Adams County 7.2 1981 3.4 Thornton 19.3 1982 6.7 Westminster 5.6 1981 6.7 Willows 11.6 1981 1.2 Weighted Averageb/ 7.9 -- __ 1/ Eight year average. 2/ Includes construction. 5/ Denver includes City and County, Read and Bill and Total Service, but not Master Metered. J Estimated, provider has unmetered households. 5/ Not available. 6/ Weighted by demand. Source: 1983 Regional Water Study Questionnarie; Ketellapper et al. , 1984. Appendix 2 69 Seasonal use was divided into winter (December, January and February) and summer (March to November). Dividing summer use by lot size gave estimates of the differences in lawn watering rates between SFM and SFF households. Certain areas in the studies were not repre- sentative of any of the three sectors and were omitted from this analy- sis. The only exception was the retention of the apartment area (including nonresidential consumption) since it was the only area that was representative of MF households in either study. Estimated factors and discussion. To interpret these data, two factors should be noted. First, the summer of 1980, 1981 and 1982 experienced slightly higher maximum temperatures and less rainfall than is typical. Second, water priced remained stable without adjusting for inflation over the period. Both of these factors influence demand to some degree. The average annual use factors for the areas are presented in table 22. The SF households exhibited greater demand than did the MF house- holds. Flat-rate customers did not display significantly greater water demand than metered customers, in terms of average total daily demand. In fact, adjusted for average City and County of Denver lot sizes, metered customers demanded more water, particularly in the summer. How- ever, the effect of meters does appear when calculating the ratio of use to lot size. Flat-rate customers typically show ratios 20 to 30 percent greater than metered customers. Total demand for metered customers tends to be higher because of their larger lot and household sizes. The ratio of summer demand to lot size was used to extrapolate the summer use rates to the average lot sizes in the City and County of Denver (6,534 ft2 for flat-rate and 8,712 ft2 for metered households) . Appendix 2 70 oso 1 .n AM • " ' AAAx A *..w2nn e3ene.33 80322 :Ig a AO ccc s cO C ^ nn _ nn ww w -__ w r w .N C. •0 y • wN•. r.. "... • 333iiii ve333 N N - C e S : Isaze2« t_8__ X a wa o - N • C r •• QA w w N i; AZ SAX w N •. N \ S.. W W £ is iie!ieg 8 i— i 3333 W L . i .n.Aisi.X -Nwn- n_ E E ti . . L. ; Iaw7�'ds+N• Y;wtn- n 41 U R LL W 3 wN '° n S 44 44 = w w a C ! ; N . 3 a . C A wo.—.ZNo • a N�nnN a • :w_wrw� wwn_.w • ! I r • 4 i • 1 A 1 gi A 242 :c t •• 413A"l • s •* 33, 2 ins : " : I •1 . I i • a 1211.7.14 " 5 i Is iki - wss 2 �• wt n, al n1 r1 •..1 Si Appendix 2 71 The flat-rate use factors are derived from a measurement of 455 households. These values seem stable and should represent the summer use. The metered use factors are derived from a smaller number of house- holds (68). The metered winter use factors agree with the flat-rate winter use factors. The flat-rate summer use factors show an applica- tion ratio of 125 percent of metered ratios. But, the application of this ratio results in gallons per household values that are very high; this may be due to the small sample size, socioeconomic differences that set the metered areas apart from the typical setting, or the slightly higher (5 percent) irrigation requirements for the summers of 1980, 1981, and 1982. As a summary of the analysis, the annual g.h.d. use factors for City and County of Denver lot sizes are shown in table 23. Table 23 Annual Use Factors From Three-Inch Metering Data Number of Winter Summer Total Households Metered 207 348?/ 555 68 Flat-rate 231 333 564 455 1/ Out of range. System losses. The flat-rate total use factors for 1981 and 1982 are used as a basis of extrapolation to check the DWD loss estimate of six percent as shown in their annual reports for those calendar years. Since winter estimates in table 22 cross calendar years--from December Appendix 2 72 through February--the appropriate winter estimates are weighted in the following manner: . Flat-Rate Use (1981) = 2/3(256)+366+1/3(223) = 611; . Flat-Rate Use (1982) = 2/3(233)+281+1/3(2,4) = 501. The following extrapolation uses data from DWD Annual Reports: . 1981 (millions of gallons) a = production 71,152 b = metered 49,075 e = flat-rate (a-b): 22,077 d = flat-rate household usage: 19,480. 365(day/year)x611(g.h.d.)x87,349(flat-rate households) 1,000,000 e = loss (c-d) f = loss percentage: (e+a)x100=3.6%. . 1982-repeating the 1981 computations using 501 g.h.d. and 87, 151 flat-rate units yields: a = 68,421 b = 48, 117 c = 20,304 d = 15,937 e = 4,367 f = 6.4% Appendix 2 73 Thus, a loss estimate, that is conditional upon measurements of water use of 455 metered households, varies between 3.6 percent and 6.4 percent depending upon the year. This is consistent with the six per- cent used in the DWD annual reports. ffirtercem survey overlap. In addition, a preliminary screening of the potential overlap between the Entercom survey households and the 3 inch meter areas was completed. In the absence of the addresses of the 3 inch meter households, Entercom survey households in the proximity (within a four or five block radius) of a 3 inch meter intersection were identified. A total of 54 addresses matched this criterion, of which only 12 could be positively (according to survey data) identified as SFF households. An additional 12 could be positively identified as MF households. Given this result, further seasonal analysis of these households using Entercom and 3 inch meter data were not undertaken for the fol- lowing two reasons: . The small number of identified households would be reduced to an even smaller set once it was determined if a household was in a 3 inch meter area and not just near one. . Individual household data for SFF and Imo' households do not exist in either data set and would have to be inferred. RATIONAL USE FACTORS To provide an estimate of the demand for water, national water use estimates were used and compiled for the various water consuming activi- ties of an average household. These data were derived from water resources encyclopedias, textbooks, and pamphlets. The results of the Household User Survey (Entercom 1983) were used to confirm some of the r Appendix 2 74 data, such as bathing estimates, size of family and ownership of swimming pools. Due to the geographical variation in irrigation requirements of lawns, outdoor use data specific to the Denver region were used, specifically the Colorado State University study (Danielson, et al. , 1979). The objective of this analysis was to estimate the SF and MF water use factors and calibrate it to other estimates. Methodology. To estimate the household demand for water, indoor and outdoor water use estimates were calculated. The indoor water use was estimated for the following indoor activities: bathing, toilets, laundry, drinking, housecleaning and kitchen activities. The average water use for each of these activities was then compiled and listed in table 24. The assumptions made concerning each activity and the resultant values are also given in the table. Data from the Household User Survey were used to determine that there are 2.87, 2.44 and 2.32 persons per household in SFM, SFF and MF households, respectively. The use factors used for bathing and showers were also verified with the Household User Survey. The estimation of method assumed the lower range of water use factors to reflect the use of some water saving devices in the EIS demand area. It was also assumed that there were no restrictions on the use of water, such as the three hour, three day restrictions that were in effect in the metroplitan area during 1977. In the EIS demand area, the summer outdoor water can be a major part of the total household water demand. The demand for lawn watering is based on lawn irrigation requirements and lot size. The lawn irriga- tion requirements were from estimations calcualted by Danielson et al. , (1979) that were based on potential evapotranspiration and rainfall for Appendix 2 75 Table 24 Indoor Water Use Factors (GHD) from Secondary Services tegory Raw Data Source Assumptions SFlti/ SFr-' HP- Ca - Toilets 6 gal/flush (1) 6 gal/flush x 69 59 56 33 gcd (2) 4 trips/day/capita 25 lied (6) watersaving (2,4) toilet-3.5 gal/flush _ Showers and 7.5 gal/min (1) 0.75 shower or 81 69 65 Bathing 5 min/shower (2) bath/day/capita x 24-29 gal/shower (5) 5 min/shower x 30 gcd (2) 7.5 gal/min 25 g/bathtub Dishwashing 15 gal/run (1) 0.33 load/day/ 16 14 13 17-25 gal/run (2) capita x 17 gal/run watersaving dish- (2) washer-12 gal/run Laundry 40 gal or more/load (1) 1 load/week/capita 15 13 8 46 gal/load; 46 gal/load (2) 80% of single 3 gcd (2) households & 50% watersaving model- of multi-family 'a" 23 gal/load households do laundry in their houses. Drinking 4 gcd (2) 0.5 gcd 1 1 1 0.5 gcd (3) Housecleaning 2 gcd (2) 2 gcd 6 5 5 Handwashing 1.5 gcd (1) 1.5 gcd 4 4 4 Toothbrushing 1.5 gcd (1) 1.5 gcd 4 4 4 Shaving 4 gcd (1) 1 person shaving/ 4 4 4 family x 4 gcd Leaks 15 gal/day/slow drip (1) 1 slow drip/hh x 15 15 15 15 gal/day/slow drip Garbage 5 gal/min x .5 min/ 4 4 2 Disposal run x 2 runs/day(5F) .5 min/run x 1 run/ day (MF); 80% of households have gar- bage disposals Miscellaneous 1 1 1 Total 220 193 176 11SFM.2.87 persons/household; SFF.2.44 persons/household; MFs2.32 persons/household. Source: (1) AWWA, 1975; (2) VSWCB, 1977; (3) AWWA; (4) Sharpe, 1981; (5) Metcalf and Eddy, Inc., 1979; (6) Flack, 198 . Appendix 2 76 Denver. Their calculated irrigation requirements ranged from four inches/season for lawn quality ratings of 40 percent of maximum, to 19 inches/season for 60 percent of maximum to 27 inches/season for 80 per- cent of maximum (100 percent implies a lush, perfectly green lawn and 0 percent implies a completely brown lawn) . Nineteen inches/season was used since it most closely corresponded to the empiric quality of, lawn (59 to 72 percent) demanded by metered residents of Northglenn, Colorado (Danielson et al., 1979). The average lot size of a SF household in the Denver area is 6,690 ft2 for flat-rate customers and 10,400 ft2 for metered customers. The average lot size of MF households with some irrigable area, such as duplexes and townhouses, was assumed to be 3,000 ft2. In all cases, 75 percent of the lot size was assumed to be irrigable. Furthermore, based on the Households User Survey, 94 percent of SF homes and 61 percent of MF homes irrigate. The following equation was then used to calculate the demand for water: Q = c x R x L where: Q = annual water demand for irrigation (g.h.d.), R = irrigation requirements for six month irrigation season (inches), L = average irrigable lot size (ft2) using 75 percent of total size, and c = unit conversion factor (0.623 gal/ft2/inch) . Estimated factors and discussion. Indoor water use was derived using best engineering judgment in applying national use estimates and activity rates to the EIS demand area. These assumptions are shown in table 24. The metered outdoor water use estimation provided in table 25 is based on lawn watering requirements reflecting the desire to have a 60 percent lawn without waste because of sensitivity to costs. Outdoor demand by flat-rate customers is determined by the same desire for an Appendix 2 77 attractive lawn, with adjusted lot sizes. Therefore, these flat-rate estimates are more applicable to metered customers, and could be low for flat-rate customers who are indifferent to price, want a 100 percent lawn and are insensitive to waste. Table 25 Residential Use Factors From the Rational Approach, g.h.d. SF SF MF (metered) (metered) Indoor use 220 193 178 Outdoor use Grass watering 238 153 45 Car washing 1 3 3 3 Total 461 349 226 1/ Car washing estimate based on 10 gal/min for a 1/2" garden hose, 10 min/car wash, 10 car washes/year x 1 car/family. Source: National Water Use Estimates. END USE FACTORS This data source provided estimated use factor for winter and sum- mer SFM demand. Data for the EDF data base were collected over a 26 month period beginning January 1976. However, only March 1976 through February 1977 data were used because of summer watering restrictions in effect in 1977. The historical data base was also used for computing weighted averages. Methodology. The consumption data for each household were divided into annual and winter (December through February) use and averaged for each household in a supplier's district. These winter values along with es- Appendix 2 78 the annual values were then used to calculate an average weighted by the total number of Denver metropolitan area SFM households. Estimated Factors and Discussion. Table 26 displays the use factor calculations by supplier and for the EIS demand area as a whole. Since suppliers included in the EDF data represent 82 percent of the 3FM households in the metropolitan area, the totals should be representative if the households themselves are representative of their respective dis- tricts. Winter use represents 49 percent of total water use for these households, leaving 51 percent for summer user. It should be noted that rainfall for 1976 (8.66 inches) was only slightly less than the long- term average (8.80 inches) for the five month summer season (May- September). ERTERCOM SURFS! USE FACTORS Objective. The purpose of this use factor analysis was to gain insight into seasonal water use. Further, a comparison with the 1976 EDF analysis can be made. Data Sources. The sample of households consisted of the 1983 Household User Survey respondents. Water providers in the EIS demand area provided billing information for these respondents. The DWD identified the appropriate water supplier for the dwellings included in the Household Users Survey sample. Entercom sent each water supplier a list of addresses requesting the water bills for each address for two billing periods: one including 15 January 1983 and one includ- ing 15 July 1982. The number of days in the billing period (30 or 60 days) was also reported. Appendix 2 79 Table 26 1/ Summary of EDF Demand Data for Single Family Households (Metered) (March 1976 - February 1977) Average Average Winter4/ Annual Water Number of Number of Demand- Demand District' Households Observations- (GHD) (GHD) Arvada 22,4985, 78 223 431 Aurora 31,6985 79 230 431 Bancroft-Clover 6,203- 21 238 442 Cherry Creek Valley 641 21 294 662 Cherry Creek Village 475 29 265 614 Consolidated Mutual 16,451 64 199 402 Crestview 3,624 29 209 418 Denver (C6C) 23,603 70 225 528 Denver (1168) 16,878 59 209 • 490 Denver (T.S.) 15,929 57 220 485 Englewood 660 31 256 478 Green Mountain 5, 19S5, 20 280 517 Kelton Heights 200 ', 15 187 253 Northglenn 5,896', 33 256 416 Thornton 9,63 6/ 17 212 384 Westminster 9,859- 33 188 377 Totals and Weighted Averages 169,448 656 222 450 1/1976 experienced rainfall during the summer (May-September) slightly less than the long-term average. 1/Boulder was excluded from this analysis as were the flat rate portions of Denver and Englewood. 31This is the number of surveys with adequate demand data for the period; other data are missing for many of these. 4/Winter is the three-month period December 1976 - February 1977. 5/1980 data. 6'1978 data. 11Estimated. Source: Environmental Defense Fund Database, 1976. Appendix 2 80 Methodology. The available data returned by the water suppliers were reviewed. SFM housing units were separated from MF units by using unit counts if reported or size of bill. Average daily consumption was calculated for these suppliers by dividing total use in the billing period by number of days in the period. Although not requested, some districts also reported demand in thousands of gallons for each address. Several steps were necessary to estimate demand for districts reporting only bills. An average monthly bill for July and January was estimated from the survey respondents. Next, the appropriate rate schedule for each district was identified using DWD Tabulation of Water Rates Metropolitan Area and Surrounding Communities, (DWD, 1982), and rate structures included in the 1983 Regional Water Study Questionnaire. Average demand for each district and each month was estimated from the average bille. Results. Table 27 reports average daily demand per SFM household for January 1983 and July 1982 for water suppliers reporting data for five or more SFM households. A number of suppliers had less than five SFM households in the sample and this sample size was considered inade- quate for the use factor analysis. Exclusion of these districts reduced the study area sample size by only 16 households to a total of 343. In general, the results are weighted toward older, more established resi- dential areas among the demand area's water providers and should be reviewed with the following caveats: . Not all data are for the same period. Some July demand includes June or August average use depending upon the water districts billing periods. In addition, July 1982 and January 1983 are not necessarily representative of seasonal use in other years. Appendix 2 81 Table 27 Summary of Entercom Demand Data for Metered Districts- (January � 1983 and July 1984) Demand Water Districts (gal/hh/day) Reporting Data for Numbe!2/of Number of January July Estimatft Five or More Units SFM- Surveys 1983 1982 Annual- DWD City and County 25,350 46 230 790 - (metered) DWD Read and Bill 19,800 43 220 710 - DWD Total Service 17,530 28 240 760 - Aurora 33,730 68 200 570 - Arvada 23,670 43 200 610 - Consolidated Mutual 16,760 27 290 780 - Westminster 11,610 27 250 670 - Thornton 11,680 13 240 480 - Green Mountain (DWD 5,720 10 210 780 - Master Meter) Bancroft-Clover (DWD 6,530 8 190 780' - Master Meter) Wheat Ridge (DWD 3,290 8 200 430 - Master Meter) Crestview (DWD 3,640 6 140 340 - Master Mater) r Lakehurst (DWD 2,470 6 - 600 - Master Meter) Willows 2,080 5 270 1,000 - Broomfield 2,740 6 210 470 - Totals and Weighted Averages 186,800 343 224 668 446 1/Demand could not be calculated from reported South Adams County data. Districts reporting data for less than five households were excluded. This accounted for only 16 single family households. 2/Average of 1974 to 1976 plus 1978 to 1982. 3/Average of January and July. Source: Entercom Household Survey, 1983. Appendix 2 82 . SF demand may include some MF use such as townhouses or small MF complexes. Three-quarter inch taps were assumed for all units in cal- culating the base service fee and rate structure. . Averages reported for certain districts are based upon data for five to 10 units and might not be representative for the entire dis- trict. . Calculating demand for average bills assumes each household uses the same amount of water in each block of the rate structure. This could lead to minor errors in water use estimates. The latest available rate structures were used. However, no effort was made to resurvey water suppliers. EDF AND ENTERCOM USE FACTORS Objective. The objective of this analysis was to estimate SF households water use factors using additional updated household survey data. Specific goals included determining the winter and summer com- ponents and evaluating causal relationships. Data sources. Three existing data bases formed the basis of this analysis: the Entercom Household Survey (1983) , the Environmental Defense Fund (EDF) , Survey (1977) and the historical data base ( 1974- 1982). The updated Entercom and EDF survey data were merged to form a combined data base. The Entercom Household Survey data were collected during the early part of 1983 under a separate contract with the Denver Board of Water Commissioners. Water consumption for 1982 (monthly and annually) , lot size, impervious area and assessed valuation information were obtained and used in supplementing the original Entercom Survey dta base. Data Appendix 2 83 for 1,149 households (SF and MF) were collected and computerized. Fol- lowing a quality control analysis of the data, 325 observations had usable demand data (reported monthly or annually and an annual demand of less than 500,000 gallons) and represented SF households. Of these 325 observations, 163 observations also had values for the three socio- economic variables of interest; lot size, income and household size. Nine water suppliers were represented. Water districts with less than five observations were excluded from the data base. The EDF household survey were updated with the 1982 water use and assessed lot and households values. Data for 877 households were obtained and quality checked. Usable demand data were found for 646 observations and 425 observations had values for the three socioeconomic variables. Income values were adjusted to 1982 using a GNP deflator. Fourteen water suppliers were represented. As with the Entercom Survey data base, water suppliers with less than five observations were excluded from the data base. The Entercom and EDF household data bases were merged to form a combined data base of 966 observations, representing 19 water suppliers with five or more observations. Five hundred ninety-two observations had values for lot size, income and household size, and represented 15 water suppliers with five or more observations. The data for lot size came from the tax assessors or the sewer authority offices when possible; otherwise, they were taken from the questionnaires. The data for income were taken from the questionnaires; in raw form, each household was designated by an income range. The raw data ranges were transformed to a scaler by using the geometric average of the range end points. The dat for marginal price came from the 1982 BBC compilation. The data for household size came from the question- naire. Appendix 2 84 Methodology. Use factors for winter and summer SF demand were estimated in this analysis and compared to the analysis of EDF demand data for metered districts (March 1976 - February 1977). The merged data base of EDF and Entercom data was used for this analysis. Data for each household were divided into annual and winter water use and averaged water supplier. A weighted average by number of SFM households was then calculated for the EIS demand area. Results are shown in table 28. Socioeconomic variables (lot size, impervious area ratio, marginal price, income and household size) were included in these analyses. Estimated Factors and Discussion. Results of the 1976 EDF analysis and the current 1982 combined EDF/Entercom analysis are comparable. The combined data bases contain information from water suppliers represent- ing over 80 percent of the SFM households in the Denver area. Analysis of the 1976 EDF data estimated a winter use factor of 222 gal/hh/day; 49 percent of the total water use. A winter use factor of 234 gal/hh/day, 55 percent of the total water use, was estimated from the 1982 EDF and Entercom merged data base. SUPPLEMENTAL WATER PROVIDER USE FACTORS Additional use factor data were acquired from individual water sup- pliers. Aurora, Greenwood Plaza and Inverness are presented herein. Aurora. The City of Aurora has provided monthly water demand data for the period of March 1982 through February 1983 disaggregated for SFM and MF households (see table 29). Appendix 2 85 e _ _ O\ a t•-N0;0000 r N P r l f0 N.0 P m.0 CO r� p n10 NV" CO.O o� aPmmot..NOPT rtnNNa. 0P 0%0% COPNCOND m Cam one,fnlvmNmmm v CO vmNPlenNNenN• vfn�aN ,,v f• n y LO = e s2 - .. ■ u > °a y ■ ■o \ O P0101 Pa a r a al.0:Pmm;°tr N 4 F tr C•m CO{rl a N ma m ! '� s ■p 0 1nm.010PIn.0r PNar UN Pt.at,.OO11O u1 t-u t-N m.O t. r CO O V O N as mvH1f� en NaNllas NN rmra menNlf'Ivr rCfl�Nrlf'1N T 'I O CD 0.�. v v v v r1 L P aN C is a ■ 0� P '/ 0 I. r en Ill 0I 0 O V 0 O a m N• 0• N• 01 IA to N 0 r r• OP 2 Ayrr C ■...... ■ \ L 0 a .4. .si 0 O O C °f !t m D ■ a C 4 • T \I N O t- b P N to. t� N m ' 10 fr1 O 0- 0 .C a ■ „O,I � P N P N IO P IO O P P a CD O M ° > 1 O 5 0 ` t ■ O O 0 A ° ij el a C $ ■ w,. 0 000000000000000N§0000000000000e � ■ • 0.11 •NN NO.-.-O aPV I[10mb Nb to0%1. 01Z,ONNNNNOO.NIaf1 S 0 elN O Cel O.C 1 L O.-c0 0000N YINamr CO 0%Pm 010Or NO rtrr P t•P a 03 0% O V ■Q■pV CP .■i O CO o • v .. v v v v -. v v t - Ov = a ■ 5 e= ■ a G9 0 1 ' 0 ■ C■p L el a 20 F "—` i 0 ■o a o P 0 N CO a to m 0 o r co O P t` m o 5 0 1 o ti P •; 500 IPn m m .a0 .o m CO a a IA .O.Q ' m a N N Y Or. a O f. E cl I 0 n p• \ ais '0 6.1 IA km 0 a L Id 0 0 IL : C pp L c = 4 m 1 ' t• P P t•- m 0 .0 ! m en Molls" m > C O .C C 0..1 Oti 0 N N a in N N in N a N N N 0 .1 0 O 2 . ,s 3 �Y N 0 H o a tg 8 o$ J el a �' L. 'Ci'O pa 4i C 8 o a .a0 CO in a a Ift CO IIC 001 N Om1 •- N N IA o 0 Q Jr N OL. C S O V o a to b ' CO N ! O .O t• N ^ a ' 11-•0 0 at N C L C = O C N m N N N P 1 O "I ■ C id 0 = Or P O a. 0 0 - C fO a t m € s . oa sa ' dt P m N N N 1n N CO a N N N P . E Y V el s 0 N 0 N N CO - I'- CO O N P m N tl S O O C to 0 ` C • i o 0 8 i f3 O a O O C • ■ • O w 0 11 - 1 ; a % L GG 1 r Iw ► o .. • • 7 e 0 1 v .0 •■a •• o =■p .i y Ggl e. CO g • Y 3 ° ` ■ Co 9 t 11 L i N 0 .1 ° o s C` a CO 1. a C I O CO r =o I m wp L. I. a a _p tip 4. ; Q CrZO �. ` '' a o0 S. QOo C O i C 1g .■. a 00 ! 1 r m tO 0.1 O G O C titbit'7t I" el eI NI 10 al 1\n� _ 0 Appendix 2 86 Table 29 Water Demand for SF (Metered) and MF Households in Aurora (1982-83) Month 1982-1983 SF MF March 185 152 April 332 232 May 463 259 June 379 271 July 653 392 August 657 344 September 515 326 October 357 143 November 202 151 December 167 127 January 183 175 February 216 137 March-February 359 226 December-February 189 146 Greenwood Plaza. Additional demand data are provided for two sup- pliers, Greenwood Plaza and Inverness. These suppliers are entirely commercial and are used to calculate demand per employee. The data are summarized in table 30. CONSERVATION, WEATHER, AND SHORT-TERM TRENDS Use rates are influenced by a number of factors, two of which are discussed in this section: conservation and weather. Short-term trends in use rates are also presented in this section, with emphasis being placed on the roles of conservation and weather in those trends. • Conservation Programs. Conservation programs in effect during the data period have influenced (to an unknown degree) the numbers in various analyses. The period during which two conservation programs, the every third day voluntary watering restrictions and Evapotranspira- Appendix 2 87 O N of .0 0C( y 0 I IT I. N T C O 0 W O 0 cO 7 CA L Z n In e- • 4) O O C, Z .1 0 ,..I 0 a - -I 44 7• 9 9 C W N C II I. N C 0 co C O u-,u-, CD I co co 4' O. N .I O. L s- N M I .0 M M n 0 43> . . N N N U H L r• N _ 0. N of N A of N .-0 -I - a a 0 o V I I I I I I m en rn N O O M In O C NNN 0 to v 0 C 0 V .1 1O C .i 0 . O 0 U RI co O aON of Ol d 42 d0 CD 0 N I . cO 0 N 4- y . C O If1 O. I ^ ' 0. O ? .0 CI C I. S M N O O O. O fO. F 1-1 D 4- N S IA .0 N N- fD CI L. Ci-i a a° g a O of m 01 C M N CO O Cl e 6:0: p > a V o I $ I 1 I t. I to .'00 en co ti s V 'O 0 m C 10 as m b C g a 0 0 0 L. L1 L. W d 'c CO YI a \ Q a) C) ea C N- — In I t- 0 Co ? M `+ p+ Z v f. .0 Ili S I N- 0. 0 Oa CO P D6 > I- a C7 0 L. N O Cn N 0 C 0 m O GI f. 4 IA .0 N- CO O. O w- N t. CII N- t- N- f- N N m W W 0 N Oa Oa T O. O. O. O. O. O. 0 N . . . . . .r .. y Appendix 2 88 tion (ET) Program and their correspondence with the principle data sources are summarized in Table 31. Table 31 Conservation Programs Evapo- Voluntary transpi- Watering ration EDF Johns Water Year Restrictions Program Survey Hopkins Entercom Providers BBC 1974 X X 1975 X X 1976 x X X 1977 Data not used because of mandatory restrictions 1978 X X X 1979 X X X 1980 X X X X 1981 X X X X X 1982 X X X X X X Although the influence of any particular conservation program is difficult to quantify, comparison of the use factors from the Water Providers Survey in the period 1974-76 with that of 1978-82 shows a decline which can, at least partially, be attributed to the initiation of mandatory restrictions in 1977 and the ensuing assortment of volun- tary and awareness programs afterwards. The Three-Inch/Johns Hopkins and Entercom data appear to reflect this change in behavior. Curiously, the EDF data also seem to reflect post-1977 use patterns even though they pertain to 1976. Weather. This section will place the period representing the demand data for theuse factor analyses in historical perspective and presents an analysis of its effects on use rates. Four weather vari- ables are commonly discussed in this context: Appendix 2 89 . Summer rainfall - volume of rainfall in the months of May through September. . Days of Significant Summer Rainfall - days of rainfall greater than 0.01 inch in the summer. . Effective Summer Rainfall - product of total summer rainfall with the fraction of summer days on which it was observed. . Summer Average Maximum Daily Temperature. Denver has a semi-arid climate that is summarized in table 32 by long-term averages of each of the above weather variables. Also in the table are the eight year averages (1974-1976, 1978-1982) for the data period. This particular eight year period is typical of long-term weather. The number of days of significant rainfall matches exactly with the long-term average, while the other three variables are all within four percent of the long-term averages. Weather conditions that occurred during the summers of record for the various data sources are also shown in table 33. The Water Pro- viders Survey and the historical data base both span the entire eight year period and are therefore typical in terms of weather. The summer portion of the EDF survey was from 1976. This summer experienced less total rainfall, less effective rainfall, and a lower maximum daily tem- perature than the long-term average, with a greater number of days of significant rainfall. All departures from the long-term average are within 10 percent. The Three-Inch/John Hopkins data are from a three year period (1980-1982) showing significantly less summer rainfall (10 percent) and Appendix 2 90 .e I C pC • I a ue ml MXX i MMXXM 1 I N 0 in C O ■I co N 0 44 CA I d CO a I N. • I • u 1. I 9 •a1. 9 • I d eo w • u > 4 M X X i M X X X M I 1 WI O• en O 03 d N I LtCZ 01 N CO C ssl • Is I U I> • m O •■ C u • e ~9 L U I I I .Oa l I I I M I I .NO h CO e N a ai N U aO 14. e'1 14 O 9 47 .1 01 1- .. O 5 Is .a _ C u a I. C L a .O C I. >1 N III 1 I IX X O • - X 1 i e. I% O Is U • r1 to CO O w • u 0 w r to • 9 ..a O Cy SIX e0 CO P- n 00 I. I. I. WW O at al 0. 2 d is 9 u 0 N 7. M T • • .I O... Cr -. d .. 0 .0 N33 in d •.>. 1 C4 2 en 0 IC 0110 N010 J0 O. N ten ( di r.0 r 7 O F CO 1n eO eO 1� e0 CO e. X CO -, S .•a .a O• 6-, ^ L 0. C • F u > .... . t . • 0 - O u • • .. O •+ • .0 .00100. CO CA 7 O • C U .e e. e. O O eVul N. ml CO 01 >V .w w my .4 C ••1 el 04 N - N NegN N .. I w 10 •.a u U • 0OGv U u J = a I. ..4 i a a far.C oe odu e u C is r • 1• wr .. O P. e•1 -. O .OP en en •+ • 0 O 11 end P1 .t en Lt 1n .t .? .o u r 3 • o iii .. Z C n• 0. U II 3 0 L 0 N ••1 1 • N Is 7 9 J • $ N u I > 0• 1. J M .+ co en e1 .0 OdO0N .t O 5 $ co j•• 0.n 0. .0 1n 01 e9 CO .0 ,n CO N 0 .-1 • C U .n 01x0 P- NCO .O030 O = Y .. C .•e •-1 5 F OC •el �.. L y 0 !. 9 u • .•1 SC 04 44. • .. n I. • to .-4 oG C 0 >. U 1. > Da CIO 4r ...1 • w 1ri v • L O. f` a .+ w 0 CG • ff O s m 01 00 el C 8 OO O 9 .1 .. • • .. 2 cc "oak 01 U qq • >. > • u N O • 9 24. 213 X :1 • .. • • •• a07F 1 007 L O uL 1. • • .a 01 • .•a .+ U O U • •• U L d .n .O n O 01 O .•1 N M U 00-.7 I. U • C w • C O. • Y CI a MaCI0% a T 7001 > 3 " a '" 0 ' 'esug ahta vF 0 'J0 Appendix 2 91 Table 33 Long Term Irrigation Requirements(Inches/Summer) Long Term Average 1974 1975 17766 MIL 1978 1979 1980 1981 1982 May 1.96 4.91 1.29 3.04 4.52 0.64 0.62 1.70 0.67 0.71 June 4.28 3.98 3.37 5.09 5.40 4.63 3.27 6.33 5.60 3.07 July 4.73 4.32 3.63 4.42 3.63 6.12 5.72 3.93 5.90 5.49 August 4.08 5.15 3.45 2.88 4.38 5.06 0.00 4.05 4.42 4.53 September 2.65 2.47 3.22 1.78 3.99 3.88 3.70 3.39 3.88 2.28 October 1.14 0.52 1 .95 0.99 1.78 0.79 1.01 2.10 1.42 0.45 Totals?/ 18.84 21.4 16.91 18.21 23.70 21.12 14.33 21.50 21.89 16.53 1/ Excluded from Ordinary Least Squares (OLS). ?/ Totals are used as independent variates in OLS. Note: The average annual requirement (excluding 1977) is 18.98, with a standard deviation of 2.87. Appendix 2 92 slightly less effective rainfall while at the same time experiencing more days of rainfall (greater than 0.01 inch) and higher maximum tem- peratures. Depending on which variable(s) influence demand, weather may be a minor influence on demand in this study. Finally, the Entercom survey data are from 1982 which showed a sig- nificantly higher number of days of rainfall (33 percent) and effective rainfall (20 percent) . Maximum daily summer temperature (down three percent) mitigating these two parameters are total summer rainfall (down two percent) . Again, the ultimate influence of weather depends upon the relative importance of each of these variables, but will only have a minor influence on the use factors. In order to quantify the effects of weather on use rates, the Jensen-Raise method (Jensen and Haise, 1963) was used to calculate sum- mer irrigation requirements for lawns in the Denver area. The Jensen and Raise method is an equation to estimate the evapotranspiration from soil and the following assumptions were made: . Monthly rainfall measured at Stapleton Airport is applicable to the entire demand area. . Typical long-term average monthly solar inputs (Langleys/day) are used. . Monthly air temperature (average daily) from Stapleton Airport applies to the entire demand area. . Six months are used; May through October. . The crop coefficient is 0.9. Appendix 2 93 . Irrigation equals Jensen-Raise evapotranspiration minus the rainfall. The year 1977 was excluded from the analyses but is presented for comparison purposes. Water demand data (g.h.d.) were taken from the 1983 Regional Water Study Questionnaire for: . SFM houses inside the City and County of Denver. . SFM houses outside the City and County of Denver. . SFF houses inside the City and County of Denver. The water demand data were fit using Ordinary Least Squares (OLS) to the summer irrigation requirements data (n_8) . The hypothesis is that SF hosuehold water demand variation is directly related to summer irrigation requirements. The year 1977 was plotted for contrast. The OLS prediction difference was fit again (n=8), using OLS, to the value of each year as a datum. This shows prediction difference related to year. The hypothesis is that the prediction difference is directly related to year, showing a trend. The calculated irrigation requirements (inches/summer) by year are shown in table 33; also shown is the long-term average irrigation requirements. The average of the historic 1974-1982 data period (excluding 1977) is very equal to the long-term average; this indicates the historic period is not atypical from a climatologic standpoint. Table 34 shows the corresponding water demand data. Appendix 2 94 Table 34 Water Use (gal/hh/day) 1974 1975 1976 19771i 1978 1979 1.9142 1981 1982 SFM inside DC&C 613 552 536 511 611 519 580 538 501 SFM outside DCBC 513 470 469 458 508 437 489 449 422 SFF inside DC&C 735 684 665 549 677 604 636 557 507 1/ Excluded from OLS. Source: 1983 Regional Water Study Questionnaire. Appendix 2 95 A two part analysis was performed to quantify any short-term trends which may be evident during the 1974-1982 period (1977 excepted) . The first part was to isolate the effects of weather. This was accomplished by using regression analysis of the demand and irrigation data for each year under the hypothesis that demand is proportional to irrigation requirements. The second part was to use these results to adjust demand for typical weather conditions. The adjusted demand was then regressed on time to yield any short-term trends. Short-term trends. Results of the weather analysis for SFM house- holds inside the City and County of Denver are presented in figure 4, which shows the SFM water demand (D) vs. irrigation (I). The resultant water demand equation is: D = 361 + 10.31 + 3, RD = 0.510 Adjusting demands for weather using this equation yields the data shown in figure 5, which shows the prediction departure (e) versus time (t). The trend equation is: e2 = 547-7t, Re = 0.491. The combined Rc = 0.751. The combined Rg is computed as: R2 = R2 + R2 (1-R2) C D e D • Appendix 2 96 a a ` II IIII 1111 1111 IIII41N E .< 1N r1N m . � N II 1pq - N ma tom a e00 ^ - DC U i ' IN CY 0 W v -I N E Zr10, H o. 4 — 02 p0 H v H ' to T ^ ct : a 0 -4 N z V ^ a H i 2. 0 H V) 1 4 Q > � -I In H ce ; — H to r _i or r N r kill lu .1l.11_i.I. ,_UaIu.., a.l I I l I (9 L In In 1.0 CAVG/HH/1V0) 3sn Appendix 2 97 1111 1111 1111 1 11 1111 1111 CO ta- x M F ll 0 - N0 n i r a N n EF V 0 e 0 0 Fo _ < a C. 2 G~4' H I IL o N CO n V } V 0 W i C° F W to t L H LL N r L Jo) a. 0 In W G ® N I ° n I I I I I I±LLIJA U._llJ_1J11111_L M N N N U)) N mn I I I -- I CAVO/HH/TV$) 3dfllbdd30 NOIl0IO321d Appendix 2 98 Figures 6 and 7 show the water demand versus irrigation and predic- tion departure trend for outside the City and County of Denver. The water use equation is: D = 312 + 8.441 + e, RD = 0.441. The trend equation is: e = 628 - 8.04t, Re = 0.731. The combined R2 = 0.850. Results of the analyses for SFF households inside the City and County of Denver are shown in figures 8 and 9. Figure 8 shows the water use versus irrigation. The water use equation is: D = 488 + 7.631 + e, RD = 0.0874 The prediction departure trend is shown in figure 9. The trend equation is: e = 1801 - 23.05t, R2 = 0.895. Appendix 2 99 a Lf) i N A . IT v 2 N - i 1M N . U N - e ' U N wN " 0 gr a GO -i " on w "° � o °a 1 -- na A . - N i H N z O � im 0 V) v ; — N Z ° O 1 � H H to,c, I— Ors �, —t _N Z t w <') — m p 0 EH — (O rj Q H -1 1-0 CD t - H 0 - - -4 H > - _. M r C 4 N N . WI 1 I I. L111111 1J.1.111.1 .1.1_LU m o LO N 0 N UV) N m U) Lf) 11) LI) t t tit CAVO/HH/1VJ ) 3sn Appendix 2 100 1111 1111 1111 11/11 1111 IIII co — / -4 es — 71 _ N — CD o m U 0 co N V C a W 0 - Mt:a _ co M n H z ' 4) 4 n 0. D co o cc t <Y } co V — e — co V N. W E Z / H - Isi CZ — 1 - ,� LL. I / 4 4D D / 0 _ co Q 0- W o ,.. / A - N i — a / — v N I II liil IIII IIII Iill IIII m co in N U) N cv in U) Q i I t I CAda/HH/ld9) 321f11dYd3a N0IjoIa3?!d Appendix 2 101 a LO g IIII 1111171H 1111 11111N It Z 4• 1N N r. mm mo e w• ~ N . e n . GI x p0 /1 Z • N tY W H -1N z I- "ID � �LI E 1 -H0 Z 4 CO H LY H 0, r- 4O N ^ z • a o ao � H �' r (O E LL. I- U) H N L • a (Y _4 t F -- H M C _I r -1 N - - "I " IIl1 ' ' w.diuiliwlu.uliwli ulnjWui m CO LA CD Ln m in 0 LA 0 In GI^ LA N 0 N LO N 0 N LA N IS) N N. N co CO (O (O Ln LA LA L1) CAbd/HH/1VO) 3sn Appendix 2 102 £Oi Z xrpueddy PREDICTION DEPARTURE CGAL/HH/DAY) I I I I I — — N CO V 01 vU1i m N CD ONi N U1 S UVi OD Ul N CO ! III lilt fill II I - V1 O H V O rn tiD H a rn V O C - - A - rn z v � _ p co n co m r , . C • C = XI - . F+ ~ O : W II I IMilll MI MIIIII Iiil IIII MI 1111 — The combined R2 = 0.904. The values of R2 that are significant given a sample size of eight, can be computed using the Fisher z normalizing transform of Z = 1n HI R that has zero expectation for R=0 with a standard deviation of 1/ n-3 or 1/ 5=0.447 for n = 8. Using a two-tail critical region, the values of R2 should be greater than 0.39 for 10 percent significance and greater than 0.49 for five percent significance. CONCLUSIONS OF THE USE FACTOR ANALYSIS As a result of this analysis, the following conclusions can be made: . The adopted use factor of 478 gal/hh/day for SFM is justified by the 1974 through 1982 data. The weather effect has not significantly biased the value. The value of 478 corresponds to what would be expected for an average climatological year. Using the long term irrigation requirement of 18.9 inches/summer as the independent variable gives 556 gal/hh/year for Denver. SFM (figure 4) and 472 gal/hh/year for SFM outside the city and County of Denver (figure 6) . The percentage of homes in the two categories is 12 percent and 88 percent. Weighting the two use factors by these percentage gives a systemwide use factor of 482; this use factor is within 0.8 percent the adopted use factor of 478. . The adopted use factor of 630 for SFF fits the 1974 through 1982 data. Using the average irrigation requirements of 18.9 inch/summer as the independent variate gives 632 gal/hh/year for the City and County of Denver (figure 8) which is comparable to the adopted use factor. Appendix 2 104 . The trend in the water use data does not adversly affect the adopted use factors for the purpose of representing the 1974 through 1982 BBC data period. The trend graphs (figures 5, 7 and 9) cross zero in the center of the 1974 through 1982 period. The early higher use balances the latter lower use over the BBC data period. The trend graphs also appear to be linear, a fact which allows the offsetting early and late trend departures to cancel each other. . 1977 is a very atypical year in terms of water use. All the plots (figures 4 through 9) show 1977 to have the largest depature from the ordinary least squares. . The data indicate a downward short term trend in water use from seven to eight gal/hh/year for SFM and 23 gal/hh/year for SFF. The departure from the irrigation equations (figures 4, 6 and 8) as plotted as trends (figure 5, 7 and 9) have significant R2 values; the trend lines are clearly shown in the scatter plots. The rates given above are the slopes of the lines fitted to figures 4, 6 and 8. The strongest trend as indicated by R2 is the SFF trend which has an R2 of 0.895. . The SFF use is independent of irrigation requirements. It is widely scattered. It however exhibits a strong short term trend. The scatter is evident as shown on figure 8. The R2 value for irrigation dependency is 0.0874 which is well below the value of 0.39 signifying a 10 percent significance level; that is 0.0874 is statistically insig- nificant. . The significant trends shown in this analysis are short term. There is evidence (DWD, 1984) that for periods longer than 1974 through 1982, the long term trend is more gradual. .-r Appendix 2 105 CHAPTER 4 POOLED WATER DEMAND FORECASTING !MODEL CHAPTER 4 POOLED WATER DEMAND FORECASTING MODEL INTRODUCTION This chapter presents the results of a pooled, time series, cross sectional, multivariate regression model explaining water demand in the EIS demand area. The data base described in Chapter 3 was applied in this analysis. Relationships found in the analysis form the basis for forecasting future water demand, which is set forth in Chapter 8. This chapter examines the structure of the pooled linear model, statistical tests, and its reliability and limitations. An alternative model form which utilizes interaction terms is also presented. Appendix 2 106 DEMAND !MODELING PROCEDURES Variables were considered for inclusion in the model if they met the following criteria: . strong theoretical basis . sensible magnitude andsign of coefficient . contribution to explanatory power of model . relationship with indoor residential, outdoor residential and nonresidential demand Potential dependent and independent variables were merged into a single data base in preparation for the multi-variate regression analy- sis. The regression analysis statistically estimated the amount of variation in the dependent variable, water use per household, for example, attributable to each of the independent variables. The rela- tionship of the independent variables in combination with one another were considered in explaining variations in water demand per household. Several additional modeling procedures were undertaken after the initial set of independent variables were analyzed: . A number of interactive variables were created and added to the list of independent variables. The theory behind the creation of these interactive variables was that certain influences might better explain water use variation in a combined form instead of discreetly. . A number of the independent variables originally considered were transformed into a higher or lower power in order to test for nonlinear Appendix 2 107 ,-- relationships with water demand. Past studies of water demand in Denver and elsewhere indicate the theoretical basis and magnitude of coeffi- cients variables might have. Certain independent variables tended to have the same influence on water demand. For example, the rainfall measures in the data base were all highly correlated. Inclusion of two such variables results in unsatisfactory coefficients for both variables. In these cases, the relative theoretical strengths of each variable were reassessed and the contributions to the explanatory power of the model were examined. In the case of weather, number of days of measurable rainfall exhibited the greatest contribution to the model's explanatory ability and had been previously identified in Denver Water Department analysis to be the best indicator of outdoor watering patterns. SELECTION OF THE WATER DEMAND MODES The results of the analysis of alternative dependent variables, discussed in detail below, are summarized in table 35. An aggregated water demand measure, g.h.d., was selected as the dependent variable in formulating the pooled EIS water demand model following an experimenta- tion process with alternative dependent variables or groups of dependent variables which might together formulate into a water demand forecasting approach. Appendix 2 108 Table 35 Characteristics of Alternative Dependent Variables Disaggregated Models Single Multi- Fully Family Family Partially Aggregated Demand Demand Disaggregated Models Model Per Per Non- Nonsingle Total Single Multi- residential Family Water Family Family Demand Per Water Demand Per Household Household Employee Demand Household Good Quality Data Yes No Yes Yes Yes High Number of Observations?/ Yes No Yes Yes Yes Data Available for Most Major Districts?/ Yes No No Yes Yes Sufficient Explanatory Powerl/ Yes No No No Yes (R2=.64) (R2=.34) (R2=.18) (R2=NA) (R2=.64) Sensible Coefficients for Variables with Theoretical Basis Yes No No No Yes Conclusions Rejected/ Rejected Rejected Rejected Accepted 1/ Only 86 observations among 21 water districts for multifamily model. ?/ Aurora, DWD Read and Bill, DWD Total Service, Northglenn and Golden miss- ing for both multifamily and nonresidential models. In addition, Arvada and Brighton missing for multifamily model and Thornton missing for non- residential model. 1/ R2 for nonresidential model was 0.18. 4/ Since nonsingle family models could not be developed, this approach to forecasting demand was rejected even though the model performed well in terms of forecasting SF demand. Appendix 2 109 DISAGGREGATED AND PARTIALLY DISAGGREGATED MODELS For the purpose of this study, total water demand was disaggregated or partially disaggregated to various consuming sectors to provide, if feasible, a series of disaggregated water demand models. Each of the resulting models was evaluated in terms of the quality of the data from which the model was formulated, the number of observa- tions in the data base, whether or not the observations were representa- tive of the major water supplier service areas, the explanatory power of the model (R2, see glossary) and whether the model contained feasible coefficients for variables which are theoretically believed to have an effect on water demand. The results of these evaluations for each of the alternative model formulations are presented in table 35. FULLY AGGREGATED !MODEL Since disaggregated and partially disaggregated models were all rejected on the basis of the criteria in table 35, total water demand per household (all types of households) per day was examined next (g.h.d.). In total, 254 observations of total water use per household were utilized in the formulation of the EIS water demand model. A single observation consists of water demand per household for a single district for a single year. Total water demand per household is considered a more logical unit of demand than total demand per capita or g.c.d. in the EIS demand area since demand by dwelling unit better reflects lawn watering needs than per capita demand. Lawn watering is household based rather than popula- tion based, and outdoor use, which includes lawn watering, is estimated to account for one-third of the EIS demand area requirements. Appendix 2 110 The total water use per household model form is accepted as a reli- able forecasting tool for the following reasons: . large number of observations with good quality data (254 in final model). . full representation of the largest districts. . maximum number of years of data available, able to capture trends in the 1974 through 1982 period. . inclusion of variables with a theoretical basis, included with sensible coefficients. . all major components of water demand--indoor residential, out- door residential and commercial/industrial/public explained by variables in the model. . high explanatory power, R2 of 0.64. TEE POOLED WATER DEMAND MODEL Based on the above analytical procedures, the aggregate, pooled water demand model for the demand area was derived and is presented in table 36. According to the model, g.h.d. in the Denver metropolitan area can be forecast on the basis of median household income, SF lot size, percent of dwelling units which are SF, employment, weather, water restrictions, and the price of water. The statistical mean and standard deviation for the dependent vari- able and each of the independent variables among the demand area water districts over the 1974 through 1982 period are presented in table 37. Appendix 2 111 Table 36 The Pooled Water Demand Model Total Water Use Per Household = + 8.49 (Median Household Income in Thousands) + 534.0 (Average Single Family Lot Size in Acres) + 0.37 (Percent Single Family Dwellings) + 35.2 (Persons per Household) + 241.7 (Service Sector Employment per Household) + 43.1 (Non-service Sector Employment per Household) - 3.59 (Number of Days of Measurable Precipitation) - 33.4 (Presence of Third Day, Three Hour Restrictions) - 67.3 (Marginal Price) - 158.2 (Intercept Value) Regression Statistics: Number of Observations (n) = 254 Coefficient of Determination (R2) _ .64 Appendix 2 112 Table 37 Descriptive Statistics of the Pooled EIS Water Demand Model Standard Variables/ Meant' Minimum Maximum Deviation Total water demand per household per day in gallons (GHD) 491.2 229.6 968.2 148.2 Median household income in thousands 28.9 9.3 0.92 6.6 Average single family lot size in acres 0.28 0.14 0.92 0.14 Percent single family dwellings 76.3 0.00 100.0 22.7 Persons per household 2.98 1.46 4.58 0.49 Service sector employ- ment per household/ 0.16 0.00 0.66 0.15 Nonservice sector employment per house- hold.a/ 1.00 0.00 4.61 0.76 Number of days of measurable precipi- tation, May to September 44.0 31.0 57.0 7.8 Presence of third day, three hour restric- tions!' 0.071 0.00 1.0 0.257 Marginal price (dollars per 1,000 gallons) 1.12 0.00 2.34 0.44 1/ See Chapter 3 for detailed definition of variables. ?/ Number of observations = 254. 2/ These variables are derived by the total number of service employees working in a water supply area divided by the total number of households in that water supply area. 4/ This is a dummy variable where 0 = absent and 1 = present. Appendix 2 113 The large standard deviation in the water demand variable points out the very different water consuming characteristics of each water supplier. The regression equation explains 64 percent of the variation in g.h.d. among the demand area water suppliers during the historical period ana- lyzed. The interpretation of the pooled model can be revealed through each term and its relationship to g.h.d.: . g.h.d., particularly outdoor residential use, is higher in those districts which are predominantly SF with higher incomes and large lots; . g.h.d., probably indoor residential use, is also higher in those districts which have larger households. . g.h.d. is higher in areas with large employment bases, presum- ably accounting for nonresidential water use. . Areas with a larger service employment base, such as hotels, hospitals or schools, have even higher nonresidential use than areas without this type of employment base. . Water use is less in wet years than dry years. The number of days of measurable precipitation directly relates to outdoor residential use. . The every third day, three hour maximum outdoor watering restrictions imposed in 1977 reduced g.h.d. by an average of 33 gallons. This calculation can be made since this factor is a dummy variable. This factor also directly relates to outdoor residential demand. Appendix 2 114 . Water use per household is lower in water districts which have higher rates per thousands gallons for water. The largest effect of marginal price is believed to be in outdoor residential demand. The above summary presents a general picture of how these independent vari- ables influence water demand. Elasticities for each variable are presented in table 38. The elasticity is the percent change in the dependent variable resulting from a one percent change in the independent variable. The larger the absolute value of the elasticity, the more sensitive the dependent vari- able is to changes in the independent variable. Through this measure, the most critical independent variables can be isolated as those with the highest elasticities. MODEL RELIABILITY There are a number of statistical tests and other means of assess- ing the reliability of a water demand forecasting model. Although no single test is by itself essential, these evaluation criteria can together reflect on the model's usefulness and dependability as a fore- casting tool. SUM OF SQUARES ANALYSIS Sum of squares analysis indicates the ability of the model to explain variations in observed water demand from the historical water demand data base. The total sum of squares of the dependent variable is the sum over all observations of the squared difference between the value of the dependent variable for an observation and the mean value of the dependent variable for all observations. As indicated in table 39, sixty-four percent of the variation in g.h.d. is explained by the water demand model. Appendix 2 115 Table 38 Elasticities for Independent Variables Variable Elasticity 1 Income +0.41 Lot size +0.30 Percent single family +0.06 Persons per household +0.21 Service sector employment per household +0.08 Non-service sector employment per household +0.09 Days of precipitation -0.32 Presence of third day, three hour restrictions -0.012/ Marginal price -0.15 1/ Percent change in GHD if one percent change in independent variable (evaluation at the means). 2i Estimated decreased demand of seven percent (mean) in 1977. Area-wide average decrease might be different. Appendix 2 116 Table 39 Sum of Squares Analysis of • the Pooled Water Demand Model Degrees Mean of freedom square Sum of Squares Explained = 3,532,000 9 392,500 Sum of Squares Error = 2,027.000 244 8,308 Total Sum of Squares = 5,560,000 R2 = 3,532,000 = 0.64 5,560,000 F-Ratio = 322.1129. = 47.24 8,400 The F-ratio is used to test the hypothesis that a linear relation- ship exists between the dependent and independent variables. At the 95 percent confidence level, the relationship is linear if the F-distribu- tion has a value no less than 2.73 for the indicated degrees of freedom. Since the pooled model has an F-ratio of 47.24, this means that the independent variables have a relationship with the dependent variable with a high degree of certainty. Therefore, a linear relationship does exist between the dependent and independent variables. BACKCASTING Estimating historical water use for each district by applying the same techniques as utilized for forecasting is one indication of the soundness of a model. Admittedly, any model should be able to predict very well the data from which it is drawn. Forecasting is accomplished by multiplying values of the variables for a particular district by the model coefficients. Applying these methods backwards, known as back- casting, resulted in estimates of water use within nine percent for each year. Table 40 demonstrates these results. Six of the nine annual Appendix 2 117 Table 40 Backcasting Results for the Pooled Water Demand Model, 1974-1982 Percent Deviation Number Projected Actual from of Water Year Use in MGL/ Use in MG Actual Districts 1974 67,990 71,650 (5.1)% 16 1975 65,920 72,240 (8.7) 21 1976 68,120 72,870 (6.5) 21 1977 68,920 68,900 0.0 23 1978 82,920 84,240 (1.6) 27 1979 81 ,490 77,680 4.9 30 1980 95,370 98,180 (2.9) 38 1981 93,210 91,520 1.8 38 1982 86,930 88.720 (2.0) 40 Total 710,870 726,000 (2.1) ?! Backcast from model. Note: Only districts which reported water consumption in the 1983 Water Pro- viders Survey are included in this analysis. Appendix 2 118 estimates were within five percent of actual use. The backcast for total use from 1974 through 1982 was within 2.1 percent of the demand reported by districts. This suggests that, on average, the model gener- ates accurate backcasts of water demand. MRH.TI OLI1EABI fl Multicolinearity is the statistical term for a high degree of cor- _ relation between two or more independent variables. Multicolinearity often occurs in regression analysis. It can result in less efficient estimation of regression coefficients, and t-statistics utilized for hypothesis testing can be unreliable. Presence of multicolinearity does not bias the coefficients, however. In other words, there is an equal chance of the true coefficient being higher or lower than the model coefficient. Certain steps were taken to ensure proper specification of the model in the face of multicolinearity. A major intervention was to not rely solely on t-statistics as the basis for inclusion or exclusion of an independent variable. Table 41 presents the simple correlation coefficients for each independent variables with the dependent variable and the other indepen- dent variables in this water demand model. The most highly correlated independent variables in the demand model are median household income, percent SF dwellings and persons per household. High income water dis- tricts tend to be predominantly SF, have large lot sizes and relatively large households. Service sector employment per household and nonser- vice sector employment per household are also highly correlated. Appendix 2 119 .0.—'. I .-1 m 1 I I I I I I I I O Lco` I I I I I I I I I o Z■ • fn. a o. o •' 8g :o I I I I I I I I 8 0 8 J L • {, m°i i �. O u. .- ID900 I I I I I I 1 8 o 0 0 c sS t S tt i i I I I I i 8 10 e- 8 o m S M • i a _ to S• G a I 1 I 1 i 8 N O to .01 0 ON I o ��a= o o 0 • L. m = m : e a 'g I I I I g y 9 m ' ! N N • 0= P. • '� ! as. mr o i i 1 $ $ a aN cu o N0 M C S g L. a O O • o v v O O J O m .4 .mi m m .N i I I• M N ! .O M ! y �.I.•a I I O T N N N O O O' %g.s 0- 0 0- 0 0 m a v 5 S 1 § O! m m T P 0 t.9 in t. .pj o O I M N .m a N O O U4 O N - O O 0 v O O O 0 m i = 4.3 ti p p 01 NI N St it y yp ID G 0 8 a �V § a in At T • O P O - m o~ 7 0 .vi >, O O O O O O O o 0 8> == ao I Ia at o ao' av N2 o yt.oa m . mgg g I. 3°. 0 ° 8 am 0 v P Stella .vi C G m m as m O ; I. L. .+ y p .. oqgq vCU ° a m -'0 P. 0 L. a. mL til SYJ avo m.1 1. 5 L. 31. 2a 9g eo L. a .a $I 4" mm. e W = O : 3 Z . C-. I.. 8 M Appendix 2 120 HETEROSCEDASTICITY A standard assumption of multivariate regression analysis is that the mean and the variance of the error terms (residuals) are constant for all values of the independent variables. Heteroscedasticity des- cribes a statistical condition of systematic correlation between resid- uals and values of an independent variable. The implications of heter- scedasticity are less precision in coefficient estimation, but no bias in the coefficients. That is, there is no reason to expect the coeffi- cients to be higher or lower if no heteroscedasticity were present. The presence of this condition also opens up the possibility of unreliable t-statistics. A standard, widely accepted test for the presence of heteroscedas- ticity is the Goldfeld-Quant test. In this test, the data set is split into two equal parts, created by placing the lower values of a certain independent variable in one group and the remaining higher values in the other. Regression equations are estimated for each data set and the results compared. Table 42 presents the results of the Goldfeld-Quant test for heteroscedasticity for each independent variable. Several variables indicate the possibility of heteroscedasticity. SERIAL CORRELATION Serial correlation refers to correlation of the error terms (resid- uals) over time in a time series regression analysis. An example of serial correlation might be an increasing trend in the error term from the initial time period to the ending period of the analysis. A wave pattern of residuals is another indication of serial correlation. The implications of serial correlation are similar to those of heteroscedas- ticity. Coefficients can be less precise but remain unbiased. T-sta- tistics cannot be viewed with the same degree of reliability, however. Appendix 2 121 Table 42 Goldfeld-Quant Test for Heteroscedasticity Sum of Squared Residuals Heterosce- (Millions) dasticity High Low Likely to Independent Variable Values Values Ratiol/ be Present 2/ Gallons per household per day 2,830 1 ,632 1.73 Yes Household income 6,915 7,392 1.06 No Single family lot size 4,595 4,882 1.06 No Percent single family dwellings 7,431 3,542 2.10 Yes Persons per household 9,068 3,636 2.49 Yes Service sector employment per household 8,729 4,368 2.00 Yes Non-service sector employment per household 9,057 4,468 2.03 Yes Number of days of precipi- tation n/a n/a n/a Not known Presence of third day, three hour restrictions n/a n/a n/a Not known Marginal price 8,833 3,889 2.27 Yes ?/ Sum of squared residuals for high values a low values or low values + high values, whichever is greater. • ?/ Hypothesis of no heterscedasiticy rejected at 95 percent confidence level if ratio greater than 1.4. Appendix 2 122 The EIS water demand data set exhibits missing values for some years since not all water districts could report data for all years. Since a statistical test has not been found to examine serial correla- tion for a pooled time series, cross-section data set which has missing values, presence of serial correlation could not be directly tested. An indirect test was performed, however, by temporarily removing all dis- tricts exhibiting missing years of data from the data set and then per- forming test for serial correlation. The Parks as well as the Fuller and Battese methods were applied on this smaller data set consisting of only 16 water suppliers. Variable coefficients did change from the unadjusted coefficients for the 16 water districts, indicating the possibility of serial correlation for this group of districts. However, the significance of this test may be questioned because the 16 water supplier model might not be representa- tive of the 57 water supplier, pooled model. The larger data set (57 water districts) was used for purposes of the EIS in spite of possible serial correlation because possible gains in the efficiency of the coefficients and reliability of t-statistics were less important than the effects of severe data losses inherent in analyzing only those 16 water districts for which data were complete. In other words, the water district based modeling approach is believed to be satisfactory only if a large number of districts which include relatively homogeneous households are contained in the data set. T— STATISTICS In theory, the t-statistic allows the evaluation of the probability that a statistical relationship exists between each independent variable and the dependent variable. Table 43 depicts the t-statistics for each variable of the aggregate model. The coefficients of all variables Appendix 2 123 Table 43 Calculation of T-Statistics for the Aggregate Water Demand Model Standard Variable Coefficient Error Statistie:L/ Median household income in thousands 8.49 1.21 7.00 Average single family lot size in acres 534.0 45.6 11.71 Percent single family dwellings 0.37 0.38 0.96 Persons per household 35.2 18.2 1.93 Service sector employment per household 241 .7 54.5 4.44 Non-service sector employment per household 43.1 9.9 4.34 Number of days of measurable precipitation (3.59) 0.76 4.72 Presence of third day, three hour restrictions (33.4) 22.5 1.48 Marginal price (67.3) 13.9 4.86 ?/ Calculated as coefficient/standard error. The hypothesis of no relation- ship between the independent variable and GHD can be rejected at the 95 percent cofidence level if the t- ratio is greater than 1.97. r Appendix 2 124 except for percent SF dwellings and watering restrictions are statistic- ally significant from zero (e.g. have a relationship with g.h.d.) at the 95 percent confidence level. However, relevance of the t-statistic is minimal in this instance as indicated from the other statistical tests. For the EIS water demand model, the more relevant question is whether a proper theoretical relationship exists between each indepen- dent variable and the dependent variable. That is, does a logical rela- tionship, positive or negative, exist with conceptually meaningful vari- ables, such as rain fall or marginal price? The aggregate should demand model passes this criterion. ADJUSTMENTS TO THE POOLED MODEL 1981-82 RESIDUALS The definition of unconstrained demand requires a demand forecast with conservation measures which are already in place. The ET program was initiated in 1981. The water demand model has been derived using 1981 and 1982 water use patterns, adjusted for weather conditions, to fully incorporate in-place conservation and the effects of the ET pro- gram. These years approximate long-term averages for weather. The method for making this adjustment is to apply the average of the 1981 and 1982 residual for each supplier (water demand unexplained in the model) to the predicted use per household for future time periods. This residual is applied consistently for all future time periods. The residuals for each supplier are presented in exhibit 2-B. This adjustment also accomplishes other objectives. The unique characteristics associated with each district which are not fully cap- tured by the model can be accounted for by using actual usage. Further, Appendix 2 125 potential statistical uncertainties are minimized with this correction factor. NATURALLY OCCURRING METERING The decline in unmetered SF dwelling (Denver and Englewood) has averaged about 250 units per year. (Denver Water Board, City of Englewood annual reports, selected years). Assuming that this trend continues, 14,860 unmetered units will become metered (or lost to the housing stock) by 2035. As a result, annual water savings will increase to 3,100 acre-feet by 2035. Table 44 exhibits these adjustments to the pooled model. INCOME ADJUSTMENT FACTOR Because the pooled model is developed on a per household basis, a scaling of the income variable must be performed. This is necessary to ensure that demand area income growth projections are weighted on a per household basis while ensuring consistent relationships among suppliers and demand area income projections. Exhibit 2-B provides the income adjustment factors. POOLED MODEL EVALUATION The pooled water demand forecasting model exhibits three types of limitations stemming from the nature of the water demand data, measure- ment of independent variables, and statistical characteristics of the equation which affect the reliability of the water demand forecasts. Water demand data from the water suppliers limited the formulation of the model in several respects. Only 16 suppliers out of 57 were able to provide total water demand for the entire 1974 through 1982 period. Water demand by consumer type was also not available from certain sup- Appendix 2 126 Table 44 Adjustment in Pooled Model Water Demand Forecasts to Reflect Natural Metering Additional Housing Units Annual Naturally Water Year Metered Savingl/ (acre-feet) 1980 0 0 1990 2,500 140 2000 5,100 290 2010 7,700 430 2035 15,100 850 1/ Assumes savings of 184 gallons per day per housing unit. Source: EIS Conservation Analysis. Appendix 2 127 pliers. Due to the difficulties of data collection at the individual household or business level and lack of high quality MF and nonresi- dential demand data, an aggregate model of water demand was developed. Only partial disaggregation of the demand forecasts is possible with this model, whereas a fully disaggregated model is preferable. Outdoor SF water demand is explained in part in the model by aver- age SF lot size. Although a better prediction of outdoor water use would have been average turf area, difficulties in accurately estimating these values prevented inclusion of average turf area as an independent variable. The outcome of the statistical tests for serial correlation, heteroscedasticity and multicolinarity do not necessarily represent serious flaws in the pooled water demand model. The coefficients are not biased in either direction. Even more importantly, the independent variables show a very logical, theoretically consistent relationship to the water demand variable. The backcast tracks closely with the his- torical data, especially the most recent years which encompass the bulk of the observations. ALTERNATIVE MODEL SPECIFICATION An alternative model specification was developed utilizing inter- action terms and nonlinear specifications of independent variables. Results for the alternative model are summarized in tables 45 and 46. Coefficients for service sector and nonservice sector employment per household, days of precipitation, watering restrictions and marginal price are very similar to those for the linear model. The relationships between income, lot size, and percent SF variables and water demand are not directly comparable for the complex and linear models. However, Appendix 2 128 Table 45 The Alternative Pooled Demand Water Model Total Water Use Per Household = + 28.8 (Median Household Income in Thousands s Lot Size) + 1.21 (Percent Single Family Dwellings • Lot Size) . - 393.6 (Lot Size2) + 213.0 (Service Sector Employees per Household) + 44.9 (Non-service Sector Employees per Household) + 41.2 (Persons per Household - 3.51 (Number of Days of Measurable Precipitation) - 33.6 (Presence of Third Day, Three Hour Restrictions) - 57.8 (Marginal Price) + 317.5 (Intercept Value) Regression Statistics: Number of observations (N) = 254 Coefficient of determination (R2) _ .63 Appendix 2 129 Hello