ECMWF极端天气指数在华东台风 降水预报中的应用和检验

选取2015—2018年影响华东地区的13个台风个例,分析降水极端天气指数EFI (Extreme Forecast Index)和SOT (Shift of Tails)与台风降水之间的统计关系。结果表明:EFI和SOT与降水气候百分位之间存在明显的正相关关系。EFI和SOT越大,强降水发生概率越高。随着预报时效的增加,EFI和SOT指数对暴雨和大暴雨的预报效果逐渐变差。对于短期(72 h以内的时效),EFI预报技巧优于SOT,而随着预报时效的延长,SOT的预报技巧逐渐接近并超过EFI。以TS评分最大为标准兼顾合理的预报偏差,得到两种极端天气指数不同预报时效、不同等级暴雨的预报阈值。总体而言,事件越极端,EFI和SOT的预报阈值越大,对于暴雨和大暴雨,EFI指数的预报阈值随着预报时效的延长有减小趋势,而SOT的预报阈值基本保持不变。在台风极端降水预报中,EFI和SOT可以作为EC定量降水预报的补充,有助于减少强降水的漏报,并提早发出预警信息。

Application and verification of ECMWF extreme index in typhoon precipitation forecast in East China

Thirteen cases of typhoon that affected East China during 2015-2018 are selected to study the statistical relationship between the Extreme Forecast Index (EFI) as well as "Shift Of Tail" index (SOT) from ECMWF ensemble forecast system and the observed rainfall cause by typhoon. Results show that there is a significant positive correlation between the EFI and SOT indices and the percentile of precipitation, and we found that the higher the EFI and SOT, the greater probability of rainstorm occurrence. For the forecast of rainstorm and heavy rainstorm, the forecast effectiveness of EFI and SOT gradually get worse with the increase of forecast time. For short-term forecast time within 72 h, the forecast skill of EFI is better than that of SOT. However, with the extension of forecast time, the forecast skill of SOT gradually approaches and exceeds that of EFI. By taking the maximum TS score as a criterion and taking into account reasonable forecast deviations, the forecast thresholds of the two types of EFI with different forecast times and different levels of heavy rainstorm were obtained. In general, the more extreme the event of precipitation, the larger the forecast threshold for EFI and SOT indices. For rainstorm and heavy rainstorm, the forecast threshold of the EFI tends to decrease with the extension of forecast time, while the threshold of the SOT remains almost unchanged. In the forecast of typhoon extreme precipitation, the EFI and SOT index can be used as supplement for bettering EC quantitative precipitation forecast. The combination of EC forecast and EFI and SOT indices can help to reduce the probability of missed detection of heavy precipitation in order to provide early warnings.