一种适用于评估降水预报服务能力的评分方法

An evaluation method suitable for precipitation forecasts and services

参考TS (Threat Score)评分,针对降水预报及服务需求提出了一种改进的检验指标—双向TS评分(Double-sided Threat Score,DTS)。TS评分只关注事件发生的预报技巧,而DTS则同等关注事件的发生与不发生。理论分析和实例结果表明,DTS评分能合理地引导预报员或客观预报做出BIAS更接近1的预报。对2007—2018年逐年降水预报和2019年逐日降水预报检验表明,DTS评分随等级升高而降低,且各等级评分大都高于0.5,评分结果更容易让公众理解。另外,DTS评分描述的预报技巧增长趋势和波动规律与TS评分一致,便于同历史评分对比。

Based on the thread score(TS), an improved verification index, the double-sided thread score(DTS), is proposed for precipitation forecast and service. TS only focuses on the forecasting skills of occurrence of events, but DTS takes into account the forecasting skills for both occurrence and non-occurrence of the events. The theoretical analysis and real cases show that DTS can help to improve the subjective or objective forecasts with BIAS closer to 1. The verification of annual precipitation forecast from 2007 to 2018 and daily precipitation forecast in 2019 shows that the DTS score decreases with the increase of the grade, and the scores of each grade are mostly higher than 0.5, so the score results are easier for the public to understand. In addition, the growth trend and fluctuation of prediction skills reflected by DTS are consistent with those by TS, which is convenient for comparison with historical scores.