Assessing future changes in daily precipitation extremes across the contiguous United States with the extended Generalized Pareto distribution

There have been multiple record-breaking precipitation events in the past and extremes are projected to intensify in the future. The reported change in extremes is contingent upon the methodology and the datasets used for the analysis. Widely used extreme value statistics disregard important information in the data by selectively sampling extremes, an issue addressed by the extended Generalized Pareto distribution (ExtGPD), an extension of the Generalized Pareto distribution that models the entire data range. We apply the ExtGPD to assess the projected changes in daily precipitation extremes across the contiguous United States (CONUS) using 22 global climate models (GCMs), part of the 6th Climate Model Intercomparison Project Phase (CMIP6) for four different emission pathways. We first assess the performance of the GCMs in reproducing observational records and select a subset of well-performing models (used a median of ten GCMs at most grids); we then determine the projected changes in daily precipitation magnitude in the middle (2031-2064) and end (2067-2100) of the 21st century relative to the 1981-2014 period. We observe daily precipitation extremes to intensify across CONUS. A projected increase of over 50% in the magnitude of 100-year daily precipitation extremes relative to the historical climate is expected in the southern, northeastern, and western United States by the end of the century under high emission scenarios. In general, the results are consistent across the models, with low inter-model variability and over ten GCMs projecting a significant intensification of the extremes. The findings corroborate that emission reductions significantly mitigate the intensification of daily precipitation by the end-century, highlighting the need for rapid climate action.