ANALYSIS OF SEASONAL STREAMFLOW FORECASTS BASED ON THE ECMWF SEAS5 SYSTEM FOR THE 1983 SOUTH AMERICAN HISTORICAL FLOOD AT ITAIPU DAM

Código

XXIV-SBRH0329

Autores

Ingrid Petry, Fernando Mainardi Fan, VINÍCIUS ALENCAR SIQUEIRA, CAMILA FREITAS, Cássia Silmara Aver Paranhos

Tema

SE02.F - Previsão hidrológica

Resumo

Seasonal forecasting is the process of making predictions of streamflow months in advance and is a powerful tool for planning and management of water resources. A Hydrologic Ensemble Prediction System (H-EPS) uses precipitation ensemble forecast data as input to hydrological models. In South America, seasonal forecast finds demand in the hydroelectric sector, responsible for 65% of the energy produced in countries such as Brazil. In this work we analyzed seasonal streamflow forecasts issued by a H-EPS for Itaipu Dam specifically for the years 1982 and 1983, when a strong El Niño caused extreme floods in several South American regions, especially in the La Plata Basin. For the experiment, we used a continental-scale hydrological model forced with seasonal precipitation hindcast data from the ECMWF model (SEAS5) after bias correction. The hindcast data consist of 25 members with lead times up to 215 days and are available for every first day of the month. The resulting forecasts were compared with those produced by the Ensemble Streamflow Prediction method (ESP), commonly used as benchmark. Results show that the H-EPS method was able to predict the occurrence of the February peak approximately one month before its occurrence, although the predictive skill was reduced with the increase in lead time. The H-EPS performed better than the ESP, resulting in predicted discharges closer to the observed values over the forecast lead times.