面向强降水预报的邻域空间检验技术研究

Traditional skill scores used in heavy rainfall verification are affected by the so-called “double penalty” problem caused by slight displacements in either space or time with respect to the observations. The fractions skill score (FSS), a neighborhood spatial verification method, can effectively compensate the deficiency of traditional skill score. However, the original idea of the FSS method is to obtain a "useful and skillful scale" which can hardly provide a certain score for precipitation assessment to analyze temporal variabilities of daily forecast, thus limiting its operational application ability. We intent to propose two sets of analysis plans for FSS method: First, analyzing the scale-dependent characteristics of the FSS method and examine the feasibility of fair comparison at a certain scale. The second is to develop a new comprehensive score that satisfies a Gaussian weight function to average the FSS over all scales. By the sensitivity analysis, a more advantageous analytical approach is selected as a preferred analytical approach to the FSS method. Based on 4-years heavy rainfall cases in summer, the applicability analysis is investigated to compare the advantage relative to the traditional skill score. By synoptic type method, we investigate the differences of FSS method in different synoptic type process so as to provide effective technical support for the better diagnosis of heavy rainfall forecasting. Meanwhile，the research results will be applied to the whole year forecasting products to analyze the effect of assessment in different seasons in order to provide support for improving the ability of forecasting heavy rainfall.

传统降水检验受到空间及时间微小差异所带来的"双重惩罚"影响，无法客观反映强降水预报能力。邻域空间检验FSS方法可有效弥补传统检验的不足，但其旨在获取降水“可用预报尺度”，难以反映研究人员更为关注的强降水逐日预报差异，极大限制了该方法的实际应用价值。为弥补这一不足，申请人拟提出两套FSS备选分析方案：通过分析FSS方法尺度依赖特征，考察特定尺度下公平比较的可行性；构建满足高斯分布可融合不同尺度于一体的FSS综合指标，通过敏感性分析优选更具优势的分析思路作为FSS优选分析方案。针对多年强降水个例开展综合分析，以获取FSS方法在强降水检验中的评估性能及相对传统检验方法的改进优势；采用天气学分型技术，考察FSS方法在不同类型强降水过程中的检验差异性规律，为更好诊断强降水预报性能提供有效的科学判据；将其应用于整年预报中分析该方法在分季节强降水预报中的综合评估效果，为改进强降水预报能力提供技术支持。

申请人针对FSS方法最初的分析思路旨在获取降水“可用预报尺度”，难以为降水逐日预报差异提供单一的检验指标，因此限制了FSS方法的实际应用能力的问题，改进FSS方法分析思路，开发了一种满足高斯分布，可融合所有尺度信息的FSS方法综合指标，通过试验表明该方法可有效延续前期分析中邻域空间检验技术的所有优势，又可不受邻域尺度人为定义的影响，具有实际的应用价值。. 随着集合预报分辨率的不断提高，集合降水预报同样存在与确定性预报相似的问题。项目组成员将FSS方法拓展至集合预报领域，构建适用于集合预报的降水空间检验指标EFSS(Ensemble Fraction Skill Score)，利用ECMWF(European Centre for Medium-Range Weather Forecasts)集合预报模式2018年夏季降水预报产品及国家气象信息中心提供的格点化降水融合产品进行分析发现，EFSS评分不受集合成员数影响，可获取一致性的评估结论。通过与适用于集合预报的常规技巧评分EETS (Ensemble Equitable Threat Score)对比分析发现，常规技巧评分受限于评分过低而无法有效反映强降水过程间差异性特征，EFSS方法则可有效提升强降水预报检验辨识度。. 针对东北地区强降水预报问题，采用EFSS方法对2018年夏季不同天气学分型条件下个例分析，通过过程分析及常规技巧评分与所研发的集合空间降水检验方法的对比分析发现，常规技巧评分在降水范围较广，强度较强时具有一定的评估能力，但当强降水分布零散，点降水占主导时，其自身的评估劣势将明显显示出来，即其很难在点对点的检验中完美匹配预报与观测，从而由于严重的“双重惩罚”问题影响评估结论的可靠性及辨识度。而集合降水空间检验方法可有效弥补这一不足，显著提高评估辨识度，可为了解集合降水预报能力提供更为可靠的检验方案。