S2S次季节到季节预报对全球降水预报的性能评价

Performance of Sub-Seasonal to Seasonal (S2S) Products for Global Precipitation Forecasts

摘要: 次季节到季节(Sub-seasonal to Seasonal, S2S)预报研究项目是由世界天气研究计划和世界气候研究计划联合发起的，旨在弥补中长期天气预报到季节预测之间的空隙，近年来被逐步应用于季风、寒潮预报等领域，但目前尚无在全球尺度上针对S2S降水产品的预报能力进行评估的研究。本研究采用了相关系数、绝对值偏差、探测率、空报率、Heidke技巧评分指数5项评价指标，对全球10个预报中心生成的S2S降水产品进行了综合评价,以期为该数据产品在水文、农业等领域的应用提供参考。结果表明：S2S降水预报的预报能力随预见期的增长而逐渐降低，在第10天后几乎丧失了预报能力。在10天预见期内，ECMWF、UKMO、KMA、CNR-ISAC的降水产品与其它预报中心相比在全球大部分区域都具有更高的可预报性；而HMCR的降水产品在全球大部分区域的可预报性较低，其它模型表现居中。空间分布而言，KMA、UKMO的降水产品在澳大利亚、北欧、东亚能够很好地捕捉实际降水特征，而其余的降水预报表现略差。

Abstract: The Sub-seasonal to Seasonal (S2S) Prediction Project is established by the World Weather Research Program (WWRP) and World Climate Research Program (WCRP) to fill the gap between medium-range weather and long-range or seasonal forecasts. In recent years, the S2S forecasts have become to apply in monsoon and cold wave prediction. However, the performance of S2S products in forecasting daily precipitation is not evaluated at the global scale. Accordingly, this study evaluated the S2S precipitation products from 10 forecasting centers using five evaluation metrics including correlation coefficient, absolute bias, probability of detection, false alarm ratio and Heidke’s skill score, aimed to provide reference for the application of S2S precipitation products in hydrology, agriculture and other fields. The results show that the skill of S2S precipitation forecasts decrease with the increase of the lead time, and they almost have no forecasting skill after the 10th lead time. Within the lead time of 10 days, precipitation products from ECMWF, UKMO, KMA, CNR-ISAC models have higher predictability than those from other models, while the HMCR model consistently performs worse than other models in terms of forecasting daily precipitation for most regions of the world, other models perform moderately. In terms of the spatial distribution, precipitation products from KMA and UKMO models are better at capturing the characteristics of observed precipitation than other models in Australia, Northern Europe and East Asia, while other precipitation forecasts perform slightly poor.