Understanding the model uncertainty of future changes in extreme precipitation events

Abstract. Despite high confidence in the global intensification of extreme precipitation under warming, substantial uncertainty remains in regional projections across climate models. Developing a process-based understanding of the physical drivers underlying this uncertainty is critical for improving future projections and informing adaptation strategies. Here, we apply a physics-based diagnostic framework to decompose projected changes in precipitation extremes and their uncertainty into thermodynamic and dynamic contributions. The thermodynamic contribution is relatively consistent across models and explains the globally mostly uniform intensification of extremes, whereas the dynamic contribution varies substantially among models and emerges as the dominant source of uncertainty, particularly in the tropics and midlatitudes. We find that the model uncertainty in the thermodynamic contribution cannot be simply explained by the local seasonal warming difference. Instead, there is a pronounced shift in the seasonal timing of precipitation extremes and thus the change in temperature on the day of extremes may substantially deviate from the seasonal mean warming, particularly across northern midlatitudes. We demonstrate that in many places to what extent the day of precipitation shifts into a cooler climate is the dominant uncertainty source of thermodynamic changes. Meanwhile, uncertainty in the dynamic contribution is primarily associated with inter-model differences in changes of updrafts. Notably, the change in updrafts at the 700 hPa level alone accounts for much of the model spread in precipitation extremes across the globe. These results highlight the key physical processes driving uncertainty in extreme precipitation projections and provide a foundation for targeted model evaluation and the development of observational constraints.