Does Dynamical Downscaling at 12km Resolution Add Value to the Representation of Precipitation?
This work analyzes how well historical precipitation across the contiguous United States (CONUS) is simulated in a 12 km Weather Research and Forecasting model version 4.2.1 (WRF v 4.2.1)-based dynamical downscaling of the fifth-generation ECMWF atmospheric reanalysis (ERA5).
We comprehensively evaluate 3- and 24-hr precipitation characteristics (diurnal and annual cycles, annual and seasonal means, precipitation frequency, and their probability distribution) across regions and seasons over the contiguous United States. Such evaluation is necessitated by the fact that the dominant precipitation processes depend upon timescale, season, and regions.
WRF well simulates the timing and magnitude of the summer precipitation diurnal cycle and the precipitation annual cycle. The biases in the 12-km WRF are comparable to the convection-permitting WRF simulations. WRF exhibits seasonally dependent precipitation biases across the CONUS. WRF slightly overestimates 3 and 24h precipitation maximum over the CONUS, in contrast to ERA5, which generally underestimates these quantities mainly over the eastern half of the CONUS. Although dynamical downscaling improves the fine-resolution representation of precipitation, it still needs reasonable bias correction before being used as input data for domain-specific models.