To compromise the computational cost and the model resolution and accuracy of wind field over regions with complex terrain, the diagnostic downscaling of prognostic meteorological fields is developed and employed in recent years. This approach aims at providing a simple and efficient algorithm to generate high resolution wind data. An EPA-approved diagnostic model (CALMET) is being used in this study for downscaling the initial guess field generated from the output of NCAR prognostic model (WRF). Although this method has been widely used, especially in countries or communities where computational resources is limited, the validity of the terrain adjustment schemes in the diagnostic model and its sensitivity to the model’s parameterization is not well documented. This kind of pa...[ Read more ]

To compromise the computational cost and the model resolution and accuracy of wind field over regions with complex terrain, the diagnostic downscaling of prognostic meteorological fields is developed and employed in recent years. This approach aims at providing a simple and efficient algorithm to generate high resolution wind data. An EPA-approved diagnostic model (CALMET) is being used in this study for downscaling the initial guess field generated from the output of NCAR prognostic model (WRF). Although this method has been widely used, especially in countries or communities where computational resources is limited, the validity of the terrain adjustment schemes in the diagnostic model and its sensitivity to the model’s parameterization is not well documented. This kind of parameterizations is of crucial importance on the accuracy of the final wind field output from the model, especially in regions with complex terrain like Hong Kong and PRD or regions where observational data is not available for model evaluation. This study aims at exploring the different terrain adjustment schemes in the CALMET model. Their limitations is discussed and documented after numerical experiments on the sensitivity of the model to different parametric relations. Improvements on the parametric relation are also suggested which is expected to also improve the accuracy of the model. Finally, the structure similarity of the diagnostic and prognostic downscaled wind field is evaluated to select a currently best model option. The downscaled wind field from this option with 100m resolution is also compared to the observation data to access the performance and seek for further improvements.