基于目标的降水检验方法及应用

Object-Based Precipitation Verification Method and Its Application

针对目前天气预报业务和模式研究中常用的如TS评分等基于点对点的降水检验方法的不足,本文对近十年发展起来的基于目标的降水量检验方法进行了讨论。基于目标的检验是将降水场看作多个天气系统相伴随的降水目标组成,它将观测场和预报场中的降水场进行识别和配对后再对各个降水目标的位置、尺度和形状等属性进行检验,因此能够提供更加详细且定量的具有天气学含义的检验信息。此类检验方法包含三个基本的步骤:目标识别、目标配对和目标检验。本文详细介绍了此方法的算法流程,并对目标识别算法中度量目标之间临近度和目标配对算法中度量目标之间匹配度的算子进行了改进。将本文改进的算法应用到对T639、ECMWF和日本模式降水预报的检验中,开发出包含强降水雨带位置、形状和范围等属性信息的检验产品,为业务预报提供定量的参考信息。

According to deficiency of the commonly used point-based precipitation verification methods in weather forecast operation and model research such as TS scores, the object-based verification method de- veloped in near decade is discussed in this paper. In this method, precipitation field is regarded as a combi- nation of many precipitation objects accompanied by several weather systems, and the verification to the position, scale and shape of each object is carried out after these precipitation objects in the observation field and prediction field are identified and matched. So, the method can provide more detailed and quanti- tative verification information that contains more synoptic meanings. There are three steps in this method： object identification, object pairing and object verification. In addition, this paper introduces the algorithm and processes of the object-based verification method in detail and makes some improvements to the opera- tions of the close degree during object identification and the match degree during object pairing. The im- proved method is applied to the verifications of precipitation of T639, ECMWF and Japan models, and ver- ification products of the position, shape and range of heavy rainfall are created, which can provide quantifi- cational information for the operation of weather forecasting.