Time Patterns of Water Use in Six Texas Cities

Abstract

Time series of monthly water use data from 6 Texas cities are examined. The time variation of the data may be divided into long memory (on average, 70% of total variation), short memory (15%) and residual error (15%). Long-memory results from trend and seasonal variation. In each city, trend is analyzed by stepwise regression of mean annual water use on population, number of water connections, household income and water prices, in which population is consistently the most significant explanatory variable for trends in water use from year to year. Short-term memory results from autocorrelation and climatic correlation. Climate correlation is strong in the three cities examined in the semi arid High Plains of Texas where if one inch more than mean monthly rainfall or pan evaporation occurs, it results on average in a 10 gpcd (38 Lpcd) change in mean monthly water use (decrease for rainfall, increase for evaporation). In three cities in humid East Texas the response of water use to weather variations is weaker. Confidence limits on modeled water use adequately envelop the response of water use to the droughts experienced by these cities from 1961-1978.