中文摘要：

利用相临过去时段预报结果中同一时刻不同时效的模式预报场差异,计算预报误差协方差,并基于集合-变分混合同化系统将其与静态背景场误差协方差结合,从而在同化系统中构建了具有各向异性和一定流依赖特征的背景场误差协方差。单点观测理想试验显示本方案改善了静态模型化背景场误差协方差的各向同性和流依赖性问题。“凡亚比”台风的一系列同化及模拟试验表明,从台风路径、强度等方面本文方案的效果都要优于三维变分法。本文方案在不需要集合预报,计算量与三维变分法相当的情况下,给同化系统引入了各向异性、一定流依赖特征的背景误差协方差,因此本方案适于在计算资源较为紧缺情况下,对时效要求较高的预报业务中应用。

英文摘要：

Based on ensemble-variational hybrid data assimilation system, the anisotropic and some flow-dependent background error covariance was introduced into data assimilation systems by combining historical forecast error co- variance with the static background error covariance. The historical forecast error covariance was calculated from the forecasts of difference between the different forecasts respectively valid at the same time. Single observation ex- periments demonstrate that the background error covariance modeled by the new method has the anisotropic and some flow-dependent information. A series of assimilation and simulation experiments for typhoon Fanapi show that the track, minimum sea level pressure and wind speed using the method were better than that of 3DVar. The historical foreeast error eovariance not need ensemble forecasts and the anisotropic and some flow-dependent information are taken into account in the data assimilation system, then the cost of the calculation is similar to that of 3DVar, so the method would be beneficial to some operational centers and research communities with limited eomputational resources.