Validation of a transient dynamical downscaling simulation approach with multi-model ensemble-based GCM bias corrections

Dynamical downscaling is the most widely used physics-based approach to obtaining fine-scale weather and climate information. However, traditional dynamical downscaling approaches are often degraded by biases in the large-scale forcing. To improve the confidence in future projection of regional climate, we used a novel bias-corrected global climate model (GCM) dataset to drive a regional climate model (RCM) over the period for 1980–2014. The dynamical downscaling simulations driven by the original GCM dataset (MPI-ESM1-2-HR model) (hereafter WRF_GCM), the bias-corrected GCM (hereafter WRF_GCMbc) are validated against that driven by the European Centre for Medium-Range Weather Forecasts Reanalysis 5 dataset (hereafter WRF_ERA5), respectively. The results suggest that, compared with the WRF_GCM, the WRF_GCMbc shows a 50–90% reduction in RMSEs of the climatological mean of downscaled variables (e.g. temperature, precipitation, wind, relative humidity). Similarly, the WRF_GCMbc also shows improved performance in simulating the interannual variability of downscaled variables. The RMSEs of interannual variances of downscaled variables are reduced by 30–60%. An EOF analysis suggests that the WRF_GCMbc can successfully reproduce the dominant tri-pole mode in the interannual summer precipitation variations observed over eastern China as opposed to the mono-pole precipitation pattern simulated by the WRF_GCM. Such improvements are primarily caused by the correct simulation of the location of the western North Pacific subtropical high by the WRF_GCMbc due to the GCM bias correction.