Regression models in water supply forecasting

This paper presents a review of some of the most common index or regression methods for making seasonal water supply forecasts. It outlines a strategy for adapting the multiple regression technique to the problem of real or near real-time forecasting and concludes that the regression technique will remain a viable tool of the water supply forecaster because: 1. It is conceptually simple and widely accepted; 2. Confidence intervals about the estimate may be easily calculated; 3. It requires a minimum of computational facilities; 4. The forecast estimates are available instantaneously; 5. The forecast errors are acceptably small.