Short-Term Forecasting Analysis for Municipal Water Demand

摘要

Short-term water demand forecasts inform decisions regarding budgeting, rate design, water supply system operations, and effective implementation of conservation policies. This study develops a linear transfer function (LTF) forecasting model for El Paso, Tex., a growing city located in the desert Southwest region of the United States. The model was used to generate monthly frequency out-of-sample simulations of water demand for periods when actual demand was known. To measure the accuracy of the LTF projections against viable alternatives, a set of benchmark forecasts was also developed. Both descriptive accuracy metrics and formal statistical tests were used to analyze predictive performance. The LTF model outperformed the alternatives in predicting demand per customer but fell a little short in projecting growth in the customer base. Changes in climatic and economic conditions were found to affect consumption per customer more rapidly than did changes in water rates.