Evaluation of General Circulation Models in Predicting Monthly Rainfall for the San Francisco River Basin

The effects of climate change on water resources can be assessed by simulations from Global Climate Models (GCMs). The Intergovernmental Panel on Climate Change (IPCC) has periodically published reports pointing out to several results of models for the prediction of current and future climate scenarios based on historical information. The evaluation of the performance of GCMs’s predictions and their intra-annual variation is an important step for the application of downscaling techniques that allow producing future projections with a lower degree of uncertainty. In this sense, the present study aims at evaluating the performance of 44 GCMs to predict the monthly precipitation distributed over the São Francisco River basin (SFRB). To do so, a new index indicator (Im) was proposed. Im is composed by a combination of Pearson correlation coefficient (r), the Root Mean Square Error (RMSE), the Precision percentage of the cell in the grid (PCell) and the Seasonality Bias (SB). The proposed index was effective in evaluating the performance of global climate change models in the prediction of rainfall in the São Francisco River Basin. The performance of the GCMS investigated decreased in predicting the rainfalls for areas by the mouth of the basin. The best performance, based on the index Im applied along the three large regions of the São Francisco River Basin was obtained with the EC-EARTH GCM.