Evaluating North American winter extreme precipitation variability and its drivers in CMIP6 models

Accurate simulation of extreme precipitation is crucial for climate models to support emergency preparedness and risk management. This study evaluates 90 simulations from 32 global climate models (GCMs) for their ability to capture monthly Rx5day (maximum 5-day precipitation) variability and large-scale drivers over North America (NA) during extended winter (November–March, 1961–2010), relative to reanalyses (ERA5 and JRA-55). An empirical orthogonal function (EOF) analysis identifies three distinct modes of Rx5day variability with regions of maximum variance over central-western, southwestern, and southeastern NA. These modes are associated with large-scale circulation anomalies from the North Pacific across NA to the North Atlantic. Moisture transport, shaped by these circulation patterns, is primarily directed into western NA from the North Pacific and into the southeastern region from the Gulf of Mexico and western Atlantic. These modes are partially related to ENSO-like SST and tropical heating anomalies, which can generate Rossby wave trains and modulate moisture transport pathways. Specific humidity and moisture flux anomalies are stronger at 850-hPa than at mid-troposphere, revealing the dominant role of low-level moisture transport and convergence, primarily driven by background mean moisture and transient winds. The GCMs capture these linkages, with the multi-model mean providing a more robust representation than individual models. The initial-condition ensemble spread, comparable to structural uncertainty, underscores the importance of internal climate variability in the extreme precipitation drivers. Overall, these findings demonstrate both the capabilities and limitations of evaluated GCMs in simulating extreme precipitation, highlighting the need for continued model improvements to strengthen risk management.