Stochastic Modeling and Forecasting of Reservoir Inflows: A Case Study in the Philippines
AbstractThis paper reports the general findings of a research project for stochastic modeling and forecasting of monthly and five-day inflows to the multi-purpose Angat reservoir. The stochastic models investigated were the pure-runoff autoregressive-moving-average (ARMA) types, both of seasonal and nonseasonal forms, and either with or without state estimation techniques of Kalman filtering; and rainfall-runoff ARMA-type of ARMAX (ARMA with exogenous input). The research demonstrated the applicability of the best-selected models for forecasting dry-season low flows and wet-season moderate flows. Recommendations for possible model improvements were also made.