Impact of different reanalysis data on WRF dynamical downscaling over China

Abstract

High-resolution simulations were conducted over China during 2000–2014 with the Weather Research and Forecasting (WRF) model forced by the National Centers for Environmental Prediction (NCEP) Global Final Analysis (FNL) and ERA-interim reanalysis data. ERA-interim showed better agreement with observed temperature and precipitation than FNL. In general, WRF downscaling resulted in a more reliable spatial distribution of temperature than FNL, but not better than ERA-interim. For precipitation, ERA-interim performed the best, while WRF produced a wet bias in southeastern and southwestern China. A better spatial distribution and annual cycle were achieved by WRF in most of the subregions compared with FNL. In the simulation of extreme precipitation, WRF captured the main features of the extreme indices, and the spatial pattern correlations were higher than those for FNL. The simulation of precipitation and extreme precipitation was sensitive to the different reanalysis data boundaries, where significant differences occurred. The precipitation and extreme precipitation of WRF forced by ERA-interim were relatively better than those forced by FNL. This suggests a positive influence of the lateral boundary conditions of ERI in improving the WRF dynamical downscaling capabilities. However, compared with the reanalysis data itself, the improvements in the WRF simulation forced by FNL were more obvious than those forced by ERA-interim. This indicates that greater improvements can be made to the downscaling skill of the WRF forced by the FNL reanalysis data than to that forced by the ERA-interim data.