Dynamical downscaling of temperature extremes over China using the WRF model driven by different lateral boundary conditions

Abstract

Extreme temperature events have considerable impact on society and natural ecosystems, which receive less attention. This study examined the impact of lateral boundary conditions (LBCs) on the downscaling simulations of the Weather Research and Forecasting (WRF) model for temperature extremes over China for the period 1981–2010. The driving data comprised two reanalysis datasets from two different institutes, ERA-interim and NECP-R2. Twelve extreme temperature indices were calculated from the WRF simulations and compared with both reanalysis products and observations. The comparison includes their climate mean and interannual variation. Results indicated that WRF can skillfully reproduce the spatial distributions of extreme temperature indices, despite some biases in certain regions. The large biases in the warmest and coldest days, warm spell duration and cool days over eastern China from reanalysis data were reduced significantly. The simulation driven by ERA-interim outperforms NECP-R2. In terms of interannual variation, the WRF model was able to capture the observational trends except the number of days with warm spell and days with moderate warm. Furthermore, the accuracy of WRF in simulating interannual variation was consistent with that of the driving data. The WRF simulation driven by ERA-interim appeared best in terms of bias frequency distributions for most indices. Overall, the results revealed that WRF dynamical downscaling of temperature extremes is sensitive to selection of LBCs, and that performance could be improved by adopting driving data such as ERA-interim.