Flood Forecasting Using the Discrete Kalman Filter

This study evaluates the usefulness and applicability of the discrete Kalman filter algorithm for predicting short-term floods. The algorithm is applied to the basin of the Angel Albino Corzo (Penitas) dam, which is part of the Grijalva Hydroelectric System, as well as to the Sayula Hydrometric Station. It is used to determine the response function for the basin and thus forecast flows into the reservoir. To that end, both flow data and precipitation recorded at weather stations located in the study area are used, as well as calculated inflows to the basin. This analysis evaluates multiple time increments and different response functions, as well as their associated parameters, using the Nash-Sutcliffe coefficient. Highly acceptable values were obtained, such that the filter is found to be useful for short-term flow forecasting, highlighting its usefulness as a tool to support policy development and the operational control of reservoirs.