Are high-resolution NASA Unified WRF simulations credible tools for predicting extreme precipitation over the Great Plains?

Abstract:

Accurate simulation of extreme weather events remains a challenge in climate models. Previous studies indicate that regional climate models better reproduce extreme precipitation with their higher spatial resolution than coarser resolution global climate models. This study utilized radar-based hourly precipitation data with a resolution of 4 km to evaluate rainfall characteristics simulated with NASA Unified Weather Research and Forecasting (NU-WRF) model at horizontal resolutions of 24, 12 and 4 km. We also examined the impact of spectral nudging on the performance of NU-WRF. The rainfall characteristics in the observations and simulations were defined as a joint probability distribution function (JPDF) between duration and peak intensity of precipitation events over the Great Plains in summer. The Regional Climate Model Evaluation System (RCMES) is an open source software suite developed jointly by NASA’s Jet Propulsion Laboratory and the University of California, Los Angeles. RCMES facilitates evaluation of NU-WRF evaluations by providing tools to process a vast amount of observational and model datasets with high resolutions. Using RCMES, we calculated JPDF for each dataset and the overlapping area between observed and simulated JPDFs to measure the similarity between the two JPDFs. The performance of NU-WRF simulations based on the precipitation JPDF is strongly dependent on their resolutions. The simulation with the highest resolution of 4 km shows the best agreement with the observations with the same resolution in simulating short-duration downpour events over the Great Plains. Our analysis indicates that even the regridded high-resolution simulation on low-resolution grids shows better performance than low-resolution simulations. The simulations with lower resolutions of 12 and 24 km show reasonable agreement only with the observational data whose resolutions are similar to the simulations.