Development of a formula to determine outdoor residential water consumption in College Station, Texas

Abstract

This thesis reports the findings of a telephone survey, public tax records, and water bills of 233 randomly selected single family detached residences, built between 1992 and 1994 in College Station, Texas. Weather information consisting of average daily temperature, daily precipitation, and daily evaporation was also necessary for analysis of gallons of water used. The purpose of this study was to (1) develop a marketing tool that builders could use to determine the water saving features for a particular area to increase sales and lead to possible mortgage reductions, and (2) help cities and developers size water lines appropriately for projected water needs. The COMBEAS computer program and various statistical tests were used to report to findings of the study. No known study has been produced that has analyzed water usage using the COMBEAS regression program and analyzed all of the variables contained in this study. Using the COMBEAS program, comparing gallons to temperature, a base load was determined that remains constant throughout the year. Any watering above this base load was attributed to temperature related (outdoor) watering. Twenty three variables, arrived at by prior research and related to water usage were then tested for significance against the amount of water attributable to outdoor watering. Of these variables, 11 were found to be significant using forward stepwise regression. Multi-colinearity tests were then conducted using the Peal-son Product Moment correlation. After eliminating all but one of those variables in each group that were highly related, 6 variables remained, including non-baseload rainfall and evaporation, yard area, existence of a sprinkler system and/or pool, and the predominant variety of grass. Using these six variables as independent variables, and the temperature dependent watering as the dependent variable, the group was then tested using best subset regression. From these results, those variables making up the highest R2 combination with p-values of less than .05 were then analyzed using multiple linear regression, producing a formula that would most accurately predict outdoor water usage for College Station, Texas and areas with similar climates and populations.