Abstract:

Abstract:Based on ARPS and WRF model, a mesoscale ensemble forecasting system with multi-initial conditions and multi-physical processes was established on the basis of different physical parameterization schemes. Moreover, a southwest low-vortex rainstorm process in the Sichuan basin which occurred from August 16 to 18, 2015 was analyzed with the above system. The results indicate that this ensemble system could well predict the heavy rainfall. Some ensemble products, such as ensemble mean and ensemble probability had better forecast skills for the rainstorm compared with the deterministic forecasts. The ensemble forecast of initial value perturbations was optimal to rainstorm and below, while that of physical process scheme was optimal to heavy rainstorm and above. The ensemble forecast considering two kinds of perturbation was better than that of single perturbation, which significantly improved the forecast of precipitation. The forecast differences of weather systems (upper trough, low-vortex, shear line) and meteorological elements (wind field, water vapor) between the “good” and “bad” clusters of ensemble members were the key factors on the forecast skills. In addition, the physical processes and initial values were more sensitive to the disturbance of the low-level (especially the forecast of wind field) than that of high-level. The ensemble forecast had a significant positive contribution to the forecast skills of heavy rain and above. However, the forecast of extraordinary storm was very uncertain, and there was still a large space for improvement.

Key words: low-vortex rainstorm, ensemble forecast, initial value perturbations, physical processes, clustering analysis

摘要：

利用ARPS和WRF模式，基于不同的物理过程参数化方案建立了一个多初值、多物理过程的中尺度集合预报系统，并对2015年8月16—18日发生在四川盆地的西南低涡暴雨过程进行试验分析。结果表明：建立的中尺度集合预报系统对此次暴雨过程有较好的预报能力，与确定性预报相比，降水集合平均以及概率预报等产品对暴雨的预报有一定的指示意义。初值扰动的集合预报对暴雨及以下量级的预报较优；物理过程扰动的集合预报对大暴雨及以上量级的预报较好；同时考虑2种扰动的集合预报总体上好于单因子扰动的集合预报，使降水预报效果得到明显改善。对天气系统（高空槽、低涡、切变线）以及气象要素（风场、水汽）的预报差异是造成“好”、“坏”成员预报效果差异的主要原因，且低层形势场（尤其是风场的预报）对初值及物理过程的扰动比高层要敏感。集合预报对大雨及以上量级预报技巧的提高有明显正贡献，但是对特大暴雨的预报不确定性很大，还有较大的改进空间。

关键词: 关键词：低涡暴雨, 集合预报, 初值扰动, 物理过程, 聚类分析