Towards the improvement of surface forecasting in the NCEP GFS

Weather forecasts are so important, particularly those surface fields like temperature, humidity, and wind because they are used to protect life and property. The forecast of large-scale atmospheric general circulation has been improved significantly in the last several decades, however the same magnitude of the improvement in the surface fields has not been made because of the complexity of the surface-atmosphere interaction. The NCEP GFS has long been known the surface 2-m temperature bias issue. Even after the recent GFS implementation, large biases of surface 2-m temperature, humidity and 10-m wind are still found. In this study we will focus on the diagnosis of surface performance of the current operation GFS during the summer and winter time over CONUS. Some approaches are tested to improve such biases, which include the use of more recent land surface characteristic data, land surface parameter refinements, and the modified land surface process parameterization. The results of this study will provide some guidance for the next land surface upgrades in the next NCEP GFS implementation.