Validating the Mark-HadGEM2-ES and Mark-MIROC5 Climate Models to Simulate Rainfall in the Last Agricultural Frontier of the Brazilian North and North-East Savannah

The MarkSim climate data generator is designed to have global validity in scales of up to five degrees. Thus, the objective of this study was to evaluate the performance of the MarkSim-HadGEM2-ES and MarkSim-MIROC5 models to estimate average rainfall in the last agricultural frontier of the savannah in the north and north-east regions of Brazil. For this purpose, the simulated data were compared with those observed and recorded by the National Institute of Meteorology, being evaluated by statistical measures of correlation, bias and performance. The results revealed high bias and relative error, with unsatisfactory performance in the micro regional and regional scales. Calibration by means of regression improved performance and showed that in order to reproduce the current climate and make reliable projections in these spatial scales possible, there is a need to correct the systemic errors of these models.