Use of dynamical downscaling to improve the simulation of Central U.S. warm season precipitation in CMIP5 models



[1] Despite supporting exceptionally productive agricultural lands, the Central U.S. is susceptible to severe droughts and floods. Such precipitation extremes are expected to worsen with climate change. However, future projections are highly uncertain as global climate models (GCMs) generally fail to resolve precipitation extremes. In this study, we assess how well models from the Coupled Model Intercomparison Project Phase 5 (CMIP5) simulate summer means, variability, extremes, and the diurnal cycle of Central U.S. summer rainfall. Output from a subset of historical CMIP5 simulations are used to drive the Weather Research and Forecasting model to determine whether dynamical downscaling improves the representation of Central U.S. rainfall. We investigate which boundary conditions influence dynamically downscaled precipitation estimates and identify GCMs that can reasonably simulate precipitation when downscaled. The CMIP5 models simulate the seasonal mean and variability of summer rainfall reasonably well but fail to resolve extremes, the diurnal cycle, and the dynamic forcing of precipitation. Downscaling to 30 km improves these characteristics of precipitation, with the greatest improvement in the representation of extremes. Additionally, sizeable diurnal cycle improvements occur with higher (10 km) resolution and convective parameterization disabled, as the daily rainfall peak shifts 4 h closer to observations than 30 km resolution simulations. This lends greater confidence that the mechanisms responsible for producing rainfall are better simulated. Because dynamical downscaling can more accurately simulate these aspects of Central U.S. summer rainfall, policymakers can have added confidence in dynamically downscaled rainfall projections, allowing for more targeted adaptation and mitigation.