中文摘要：

混合变分/集合同化是基于变分＋集合思路新兴发展起来的一种资料同化方法。利用WRF和WRFDA最新版本3.5构建混合集合同化的流程,通过单点试验直观体现出混合集合同化方案＂流依赖＂背景误差协方差的影响。对台风＂海鸥＂路径和强度的模拟和分析表明,混合集合同化方案在台风路径和强度的预报上要强于三维变分同化方案。对比两种方案同化后的分析场表明,混合集合同化方案分析出的台风中心强度优于三维变分同化方案分析结果,其原因一方面是混合集合同化方案初始场是采用集合平均的结果,另一方面是混合集合同化方案采用＂流依赖＂背景误差协方差的影响,这两个因素对台风的预报准确性有一定作用。

英文摘要：

A Hybrid ensemble Kalman transform filter-variation data assimilation method （Hybrid DA） is a new assimilation method based on variational analysis and ensemble methods. In this study, we developed a Hybrid DA cycle system based on the WRF and WRFDA＇s newest version 3.5 to simulate Typhoon Hai ou （2008）. A single-point observational test showed the effects of the Hybrid DA system＇s ＂flow-dependent＂ ensemble background covariance whereas 3DVAR did not. The errors of the track and intensity forecasts by the Hybrid DA method are smaller than those by the 3DVAR. Experiments showed that such improvements are due to the use of the ＂flow-dependent＂ ensemble background covariance and ensemble mean background. Detailed diagnostics further revealed that the increments produced by the Hybrid DA and 3DVAR were different for both the analyses of the typhoon itself and its environment.