Performance comparison of four daily weather generators for historical period and downscaling future GCM scenarios

ABSTRACT Weather generators (WG) are one of the major tools for downscaling future scenarios of general circulation models (GCMs) for regional, especially hydrological, climate change impact assessment. A key capability of WGs in the downscaling process is their ability to transfer changes in the various statistical characteristics of weather variables, as projected by GCMs, to the downscaled series. The purpose of this paper is to evaluate the performance of four WGs for transferring the delta change of various statistical characteristics of weather variables predicted by GCMs to downscaled series. The performances of the WGs for downscaling shared socioeconomic pathway (SSP) 119, SSP 370, and SSP 585 scenarios of the Canadian Earth System Model version-5 (CanESM5) in six sites are evaluated. The WGs include LARS-WG, M-LARS-WG, IWG2, and D-IWG. Based on the results, overall, the performance of all the WGs in downscaling future GCM scenarios is reduced compared to the historical series simulation. However, IWG2 and M-LARS-WG by having monthly components performed better than D-IWG and LARS-WG, respectively. So, it is suggested that in addition to evaluating the performance of WGs in the simulation of historical variables, their performance in downscaling the future scenarios of the climate models should also be evaluated.