资源环境科技发展态势分析平台

An adaptive localization method using the hierarchical ensemble filter (HEF) technique was first applied to the ensemble Kalman filter (EnKF) radar data assimilation (DA) though observing system simulation experiments (OSSEs) for an idealized supercell storm. The HEF method calculates localization by minimizing the sampling error without defining the physical distance between observation and state variables. Four experiments using regular Gaspari and Cohn (GC) and HEF localization methods with different group sizes (group numbers of ensembles) were performed. The HEF localization experiments perform better than GC experiment, producing smaller analysis errors and larger ensemble spreads for most model state variables, and stronger wind, vertical vorticity, and cold pools. The best results are obtained when the largest group size is used in the HEF experiment. With this improved analysis, the forecast error is reduced for most variables and levels. The dynamical and thermal distributions of HEF experiments are found to better promote the development of the supercell storm which helps to improve the forecast reflectivity in terms of location, intensity, and area coverage. This suggests the potential of HEF localization for an EnKF radar DA. A larger group size for this method is expected to lead to improved analysis and forecast results compared with the regular GC localization method.