[p2] Parameter optimization


A terrain-following Covariance Localization for the WRF-LETKF Radar Data Assimilation System

Pin-Ying Wu (National Central University, Taiwan), Shu-Chih Yang (National Central University, Taiwan), Chih-Chien Tsai (Taiwan Typhoon and Flood Research Institute)


The WRF-LETKF radar data assimilation system has brought great benefit for quantitative precipitation nowcasting (QPN) in Taiwan. This work seeks to improve the localization strategy in order to adapt to the complex topography in Taiwan. In the ensemble-based data assimilation, the covariance localization is applied to avoid the sampling error in the estimate of error covariance at longer distance. When dealing with high-resolution assimilation over complex topography, a terrain-following localization was proposed in this study insteading of a horizontal localization which was only the function of a distance.
The terrain-following localization was applied to a case of typhoon Morakot (2009), which dumped over 350 mm in 6 hours over the mountainous area of Taiwan. Positive influence, can be derived by using the new localization strategy, e.g. improving the accuracy of the wind analysis over the central mountain range of Taiwan. As a result, QPN over mountain area is improved.