Projected changes of precipitation extremes in North America using CMIP6 multi-climate model ensembles

In this study, we estimated a comprehensive set of extreme precipitation indices defined by the Expert Team on Climate Change Detection and Indices (ETCCDI) over North America (NA) using 18 selected higher-resolution GCMs of the sixth Coupled Model Intercomparison Project (CMIP6). The GCMs are evaluated in their historical simulations (1981-2010) against those estimated from in-situ (Daymet) and reanalysis (NARR) dataasets, followed by the projected changes in precipitation extremes of NA in response to climate warming (SSP1-2.6, SSP2-4.5 and SSP5-8.5) scenarios of CMIP6 in terms of temporal variations, spatial distributions, seasonal patterns and model agreement over the 21st century relative to 1981-2010. Results show that the CMIP6 ensemble median (CMIP6-EnM) can generally capture the observed extreme precipitation patterns of NA, and it outperforms all 18 individual GCMs with negative RMSE ' XY, higher R statistics (R > 0.8) and the signal-to-noise ratio (SNR) > 1 in most sub-regions of NA. In general, extreme precipitation events are projected to occur more frequently and severely in the future, particularly in latitudes above 55 degrees N of NA and in coastal areas. In contrast, dry conditions in southern NA (e.g., NCA and SCA) are projected to intensify over the 21st century. Seasonal changes are projected to be more significant in winter than in summer for northern NA (e.g., GIC, NWN and NEN), while in Central America (e.g., WNA, CAN, ENA), precipitation extremes will likely be less severe in both summer and winter. However, unlike strong model agreements in high (100%) and mid-latitudes (80%), there is often a wide disparity among the signs of projected changes between GCMs in southern NA. Unless greenhouse gas emissions are drastically reduced immediately, NA will likely be vulnerable to the impact of worsening extreme precipitation events especially in northern and coastal regions.