Global climate models have difficulties in simulating extreme weather events, especially extreme precipitation (EP), due to a lack of sufficient spatial resolution. Large-scale precipitation is simulated fairly well in low-resolution climate models (LRM, 1.0°), but the smaller scale and more intense precipitation events need a higher resolution model to be represented properly. With rapid computer power growth and model improvements, the hope is to capture EP better than before. Recently high-resolution climate models (HRM, 0.25°) have been developed but the major challenge of simulating EP using the HRM is that the computational cost would be 50-100 times greater than the LRM. In contrast, the RRM approach uses only 10% of the computational resources of an HRM, while 10% of the global area is refined to a higher resolution. Also, the RRM approach, unlike a nested regional modeling approach, can retain the nonlocal interactions while simulating the circulation detail in a particular region.
In this work, we evaluate a regionally-refined version (1° to 0.25°) of the atmospheric model version 1 (EAMv1) of the Energy Exascale Earth System Model (E3SM). To understand how well the model predicts EP, we first analyze the precipitation statistics in the RRM, comparing it to both HRM and LRM, as well as observational data from ground-based and satellite measurements. Further, we carry short-term (2-4 weeks) forecasts initialized by observation to identify the impact of model bias on the simulation of EP.