Evaluation of Seasonal Inflow Forecasting to Support Multipurpose Reservoir Management: A case study for the Upper Maule River Basin, Chile

Abstract: Seasonal hydrological forecasts of future streamflow volumes can provide water resources managers with valuable information to improve long-term water resources planning and water use efficiency. The latest generation of coupled ocean-atmosphere general circulation models provides an opportunity for the prediction of hydroclimatic variables (e.g. precipitation, streamflow; soil moisture) at long lead times, which is central to water resources management, agriculture and disaster planning. However given the inherent uncertainty and the large-scale resolution of climate model forecasts compared to the catchment scale, there is a need to evaluate their accuracy and value for water management. This study systematically evaluates several seasonal hydrological forecasting options for a mountainous catchment in central Chile, to assess seasonal reservoir inflow forecast skill in comparison with conventional seasonal hydrological forecasting. We find, in comparison with resampled historical precipitation forecasts, that bias-corrected seasonal precipitation and temperature forecasts from the latest global climate models can improve the accuracy and skill of inflow forecasts during the high-rainfall season (April-October) forecasts with periods of above-average rainfall typically associated with the El Niño-Southern Oscillation (ENSO). For reservoir managers, this improvement in forecast skill provides valuable information related to aid long-term planning of water resources management and hydropower production. Due to greater predictability of predominantly snowmelt-based inflow during the low-rainfall season (October-April), accurate simulation of initial hydrological conditions such as reservoir water levels and accumulated snow, can provide accurate seasonal hydrological forecasts for longer lead times with the use of forecasted precipitation and temperature data. However, the improvements in inflow forecast accuracy and skill, during the low-rainfall season, are marginal when compared with an Extended Streamflow Prediction (ESP) approach.