The outputs from general circulation models (GCMs) lack spatial and temporal accuracy for local and regional studies due to their large-scale networks. Therefore, there is a need to make the scale of outputs from these models smaller to the station and point scales. This has led to the development of regional and statistical models that have wide applications in climate change studies from the beginning of their introduction and decision-making for facing and adapting to the consequences of climate change in recent years. The models based on statistical methods are more popular and applicable due to their ease of use and because they do not need high computational power. Among the statistical methods, LARS-WG and SDSM are the most commonly used and valid downscaling models. In this study, the results of our analysis related to the performance of these two models in simulating temperature and precipitation changes in western Iran are presented. The weather stations under study include 17 stations with a long-term statistical period (1989–2018) in three provinces of Kurdistan, Kermanshah, and Ilam. To evaluate the performance of the models, MSE, RMSE, MAE, and R2 are used. The results show that the two models have an acceptable level of ability in simulating temperature and precipitation changes in the area under study. However, different results are reported for different stations and within different weather parameters. A comparison between the performance of the two models in simulating temperature and precipitation changes reveals that both of them have higher accuracy in simulating temperature than precipitation. Furthermore, the SDSM model is more successful in a monthly simulation of temperature and precipitation with lower uncertainty. However, it has a time-consuming and complicated simulation process. The LARS-WG model is more efficient in simulating annual precipitation and is simpler with a higher performance speed. In a nutshell, none of these models is b
