Performance Evaluation of Four Different Land Surface Models in WRF

This study presents a performance evaluation of four different land surface models (LSM) available in Weather Forecast Research (WRF). The research site was located in Haean Basin in South Korea. The basin is very unique by its geomorphology and topography. For a better representation of the complex terrain in the mesoscale model were used a high resolution topography data with a spatial resolution of 30 meters. Additionally, land-use layer was corrected by ground mapping data-sets. The observation equipments used in the study were an ultrasonic anemometer with a gas analyzer, an automatic weather station and a tethered balloon sonde. The model simulation covers a four-day period during autumn. The result shows significant impact of LSM on meteorological simulation. The best agreement between observation and simulation was found in the case of WRF with Noah LSM (WRF-Noah). The WRF with Rapid Update Cycle LSM (WRF-RUC) has a very good agreement with temperature profiles due to successfully predicted fog which appeared during measurements and affected the radiation budget at the basin floor. The WRF with Pleim and Xiu LSM (WRF-PX) and WRF with Thermal Diffusion LSM (WRF-TD) performed insufficiently for simulation of heat fluxes. Both overestimated the sensible and underestimated the latent heat fluxes during the daytime.