Medium-range reservoir inflow predictions based on quantitative precipitation forecasts

SummaryForecasts of inflow into major reservoirs of the Brazilian hydroelectric power system are needed for the operational planning over periods ranging from a few hours to several months ahead. Medium-range forecasts of the order of a few days to two weeks have usually been obtained by simple ARMA-type models, which do not utilize information on observed or forecast precipitation, nor streamflow observations from upstream gauging stations. Recently, several different hydrological models have been tested to assess the potential improvements in forecasts that could be obtained by using observed and forecast precipitation as additional inputs. We present results from the use of a large-scale hydrological model applied to part of the Paranaíba river basin between Itumbiara and São Simão power plants (75,000 km2) using precipitation forecasts from the regional Eta model run by the Brazilian Center for Weather Prediction (CPTEC). Results were compared with those from the currently-used ARMA model and it is shown that forecast errors can be reduced considerably, during both wet and dry seasons. Further reductions in prediction errors may be anticipated from improved rainfall forecasts and of data quality used by the hydrological model.