Stochastic Forecast of Water Losses

Abstract

For the purpose of irrigation and water resources planning, it is important to know the water loss for a drainage basin. Unfortunately individual measurements of elements of the total water loss are not realistic, at least for the large watersheds studied in this research. Based on the water budget approach, annual water loss series are formulated in this paper. A modeling technique that includes the homogeneity test of data and the best model selection is developed to fit the water loss series by a stochastic process. The results of this study reveal the existence of data nonhomogeneities in the annual water loss series from the Ohio River Basin, which requires adjustments before the mode! fitting by a stochastic process. The selected best model by the criterion of the parsimony of parameter was successfully used to forecast the regional water losses based on the proposed procedure.