Avaliação do modelo climático regional RegCM4 sobre o Nordeste brasileiro para o período do outono

Abstract

The regional models have systematic errors in different regions, mainly in the tropics, due to adjustments in physical parameterizations, cumulus convective and precipitation grid scale. On this, the objective of this work was to evaluate the precipitation and temperature of parameterizations Grell and Emanuel Regcm4 model on the northeastern region of Brazil during the period from 1998 to 2008. To evaluate this model was necessary to do a bilinear interpolation through the Climate Data Operators (CDO) program, where all data was interpolated to grid of 0.25 of the Tropical Measuring Mission satellite (TRMM). Then was made an average of autumn precipitation in mm/day and temperature in °C. Posteriorly were realized cluster analysis, correlation, variance (ANOVA), differences between test averages (t-student), and calculations of errors as the Absolute Mean Error (MAE), Mean Squared Error (EQM) and Root Mean Squared Error (REQM). Concluded that Emanuel parameterization overestimate the precipitation in relation to estimate by TRMM, especially in regions that are more accumulated rainfall. The Grell parameterization simulating precipitation was minor errors in the simulations on all clusters. In relation the temperature simulated by the Regcm4, the best results were obtained in simulation settings of this variable in all clusters.