多次取样的网格化暴雨预报“配料法”研究

利用2008—2019年甘肃省陇东342个测站的逐日降水资料和ECMWF_ERA5再分析资料,总结了不同暴雨天气过程的环流特征,建立了精细至模式预报每个格点、综合考虑暴雨日每个预报时次的多次取样网格化暴雨预报“配料法”方案,并通过2020年5次暴雨过程进行验证。结果表明,陇东暴雨可分为副高扰动型、冷槽切变型和偏北气流对流型;由于不同类型暴雨的环流特征差异明显,降水范围及持续时间不同,具有代表意义的关键物理量及阈值也不相同;网格化配料法的过程预报准确率相比ECMWF_THIN提高了20%,站点预报准确率提高了7.8%,性能优于ECMWF_THIN模式的暴雨预报;配料法对小区域预报效果优秀,预报范围越大预报准确率越低。

Research on "Batching Method" for multi-sampling grid rainstorm

Abstract: Based on the daily precipitation data of 342 stations in Gansu Province from 2008 to 2019 and the ECMWF_ERA5 reanalysis data, the circulation characteristics of different rainstorm weather processes are summarized, and a multi-sampling grid rainstorm prediction "batching method" scheme is established, which is precise to model prediction at each grid point and comprehensively considers each forecast time of rainstorm day. It was verified by 5 rainstorm processes in 2020. The results show that the heavy rain in Eastern Gansu can be divided into subtropical high disturbance type, cold trough shear type and northward convection type. Because the circulation characteristics of different types of heavy rain are obviously different, the precipitation range and duration are different, and the key physical quantities and thresholds with representative significance are also different. Compared with ECMWF_THIN, the process prediction accuracy of grid batching method is improved by 20%, and the site prediction accuracy is improved by 7.8%, and the performance is better than that of ECMWF_THIN model. The batching method is good for small area prediction, and the greater the forecast range, the lower the forecast accuracy.