Impact of Land Surface Information on WRF’s Performance in Complex Terrain Area

By using the 1-km resolution digital elevation model dataset of China and the moderate-resolution imaging spectroradiometer (MODIS) land use and vegetation fraction data recorded in 2006, this study assesses the impact of land surface information on the performance of the Weather Research and Forecasting (WRF) model for the river valley city of Lanzhou in winter. Results indicated that the near surface temperature was very sensitive to the land surface data resolution, and the near surface wind field was not sensitive to the land surface data resolution. The WRF model reproduced the near surface temperature better than the wind field, and its performance on near surface temperature was clearly affected by land surface information. The hit rate of the WRF simulated near surface temperature to observations increased by 15.8% when model default land surface data were replaced by MODIS derivations and the 1-km resolution digital elevation model dataset of China; the improvement measured in the night was more obvious than that measured in the day. The impact of surface conditions on temperature extended throughout the boundary layer, which indicates that accurate land surface information is vital for improving near surface and boundary layer simulation of the WRF model. Moreover, the WRF model accurately simulated the evolution of the wind field. The error of WRF modeled wind speed slightly decreased and the error of WRF modeled wind direction slightly increased with updated land surface information. The impact of the initial values of soil moisture and initial integration time on the model's performance is more obvious noticed in winter in arid and semi-arid regions.