Long‐term flow forecasts based on climate and hydrologic modeling: Uruguay River basin

This paper describes a procedure for predicting seasonal flow in the Rio Uruguay drainage basin (area 75,000 km 2 , lying in Brazilian territory), using sequences of future daily rainfall given by the global climate model (GCM) of the Brazilian agency for climate prediction (Centro de Previsão de Tempo e Clima, or CPTEC). Sequences of future daily rainfall given by this model were used as input to a rainfall‐runoff model appropriate for large drainage basins. Forecasts of flow in the Rio Uruguay were made for the period 1995–2001 of the full record, which began in 1940. Analysis showed that GCM forecasts underestimated rainfall over almost all the basin, particularly in winter, although interannual variability in regional rainfall was reproduced relatively well. A statistical procedure was used to correct for the underestimation of rainfall. When the corrected rainfall sequences were transformed to flow by the hydrologic model, forecasts of flow in the Rio Uruguay basin were better than forecasts based on historic mean or median flows by 37% for monthly flows and by 54% for 3‐monthly flows.