Resumen
Many scientists assume that RCM output is directly used as input for climate change impact models, while it consists of systematic errors. Consequently, RCM still requires bias correction to be used as an input model. The purpose of this study was to analyze the RCM performance before and after bias correction, its best performance from several models, as well as to clarify the importance of bias correction before it is used to analyze climate change. As a result of this, the method used for bias correction was Distribution Mapping method (for rainfall) and Average Ratio-method (for air temperature). While the Generalized Extrem Valuedistribution (GEV) was used to analysis extreme rainfall. To determine the performance of the model before and after bias correction, statistical analysis was used namelyR2, NSE, and RMSE. Furthermore, ranking for every single model and Taylor Diagram was used to determine the best model. The results showed that the RCMs performance improved with bias correction. However, CSIRO-Mk3-6-0, CCSM4, GFDL-ESM2M, and MPI-ESM-MR models can be ignored as ensemble models, because they demonstrated poor performance in simulating rainfall. From this study, it was suggested that the best model in simulating daily and monthly rainfall was ACCESS1-0, while MIROC-ESM-CHEM (daily air temperature) and ACCESS1-0 (monthly air temperature) were best models used in simulating air temperature. Key words: RCM, bias correction, performance, rainfall, air temperature