Comparison of LARS-WG and SDSM Downscaling Models for Prediction Temperature and Precipitation Changes under RCP Scenarios

Various methods developed to convert large-scale data to regional climatic data. In few studies , the results of these methods have been statistically compared. The main purpose of this study was to compare SDSM and LARS-WG models for Downscaling output data of CANE-SM2 and HADGEM2-ES general circulation models under RCP2.6, RCP4.5 and RCP8.5 scenarios. For this study, precipitation, minimum and maximum temperature of Arak synoptic station were used as the base period (1980-2005). The results of the predictions of the SDSM and LARS-WG models were compared for three periods (2021-2040, 2041-2060 and 2061-2080). The RMSE, R2, MAE and NSE criteria were used to evaluate the performance of SDSM and LARS-WG models. Based on the results of the evaluation criteria, both models have an acceptable resuts in simulation of climate variables. Downscaling results of both models showed that minimum and maximum temperatures should increased compare to baseline in all scenarios and periods from January to June, but LARS-WG model had a over estimation compared to the SDSM model. Changes in predicted precipitation by SDSM and LARS-WG models did not have a clear trend. According to the results, the SDSM model shows more changes in precipitation, wherase, the LARS-WG model shows more changes in minimum and maximum temperature compared to the base period. Based on the results of this study, it is not possible to determine the exact superiority of each of the models, But SDSM model had a better prediction for precipitation and maximum temperature, while, LARS model had better results for maximum and minimum temperature.in general it can be said that the prediction results of both models are statistically significant (P<0.01) in most cases.