Improved and Expanded Probabilistic Downscaling of CMIP6

The goal of this project is to develop a gridded dataset of future hourly and daily precipitation for the entire U.S. and U.S. territories out to the year 2100. A state-of-the-art statistical downscaling methodology which translates global climate model output to realistic high-resolution gridded precipitation will be used. Each climate model from the Coupled Model Intercomparison Project Phase 6 (CMIP6) will be downscaled individually so that the degree of uncertainty in future projections can be properly accessed. Sophisticated methods for accounting for topography and recent advancements in machine learning will be used to increase the skill and accuracy of the downscaling. The downscaling methodology is well suited for reproducing precipitation extremes and will provide valuable information on future point precipitation frequency estimates. These estimates are important for the design stormwater management systems, drainage systems, flood control measures, and other infrastructure to handle extreme rainfall events.