在完善WRF集合均方根滤波同化系统的双向嵌套预报方案、雷达资料质量控制方案和多微物理过程积分方案的基础上,引入了一种扰动势函数、流函数的尺度较小、物理意义更明显的集合初始扰动新方案。通过台风“桑美”登陆过程中的多普勒雷达资料直接同化试验,对比新扰动方案与目前应用较多的背景误差协方差随机扰动和全场随机扰动方案,对集合滤波同化效果的影响。试验结果表明,3种初始扰动方案都有效改善了台风内部结构,但新方案由于尺度更小、结构更为合理,其同化效果也更为显著,提高了对台风路径、中心最低气压、最大风速及降水的预报质量。
On the basis of optimizing WRF ensemble square-root filter assimilation system with two-way nesting forecasting scheme and radar data quality control scheme, which also integrates multi-physics ensemble integral scheme, a new ensemble initial perturbation scheme with perturbed potential and stream functions is introduced. This new scheme has smaller perturbed scale and more obvious physical meaning. With direct assimilation of Doppler radar data on the landfalling of typhoon Saomai, this paper compares the impacts of the new perturbation scheme, the background error covariance random pertur- bation scheme and the whole space random perturbation scheme on the ensemble filter assimilation. Re- sult shows that all three initial perturbation schemes can improve typhoon structure, but the assimilation effect of the new scheme is more distinct due to smaller scale and better physical structure, which can improve the forecasting quality of typhoon track, center minimum pressure, maximum wind speed and precipitation.