Evaluation of the Brazilian Global Atmospheric Model in the Simulation of the Water Budget Components in the Amazon Basin

Abstract This study evaluated the performance of the Brazilian Global Atmospheric Model (BAM) in the representation of water balance components in the Amazon basin for the present climate (1979-2015). The BAM model reproduces the space-temporal variation of the water balance components in the Amazon basin, despite the deficiency in correctly positioning the maximum precipitation and moisture convergence over the region. BAM underestimated precipitation (8,8%) and runoff (36,8%), and overestimated evapotranspiration (5,3%). Moisture sink behavior was well represented by BAM, since precipitation is always greater than evapotranspiration (P > E) in the Amazon basin. The non-closure of the water budget (52,6%) in the basin shows the limitation of BAM physical parameters in the representation of convection (precipitation) and moisture convergence, proving the need for better adjustments. Overall, despite underestimating rainfall, BAM adequately simulated annual and seasonal rainfall. The evapotranspiration was overestimated, especially in the rainy season, but was underestimated in the dry season. The results showed that BAM needs adjustments and calibration in the representation of surface processes, since it presented difficulty in simulating the seasonal variation of evapotranspiration. Systematic errors found in water balance components are of great importance in determining the degree of confidence for present climate simulations and future projections.