基于MODE方法的SCMOC降水产品在河西走廊干旱区的检验

Li Lingping1,Li Tianjiang1 *,Wu Wen1,Li Yanying1

摘要:

模式预报效果的客观检验和评估,对改进模式预报性能和提高预报员本地化订正提供重要参考依据。利用2019-2022年5-9月国家气象中心精细化网格降水预报指导产品SCMOC和三源融合网格实况降水分析资料(CMPAS),基于经典检验和MODE空间检验法,检验分析智能网格预报 SCMOC在河西走廊的降水预报表现。结果表明:(1)SCMOC对各量级降水预报TS随降水量级大小而不同,量级越小,TS越大,量级越大,偏差越大,命中率越小。(2)SCMOC对暴雨以下量级降水预报目标相似度达到0.71～0.75,暴雨目标相似度为0～0.49。(3)目标轴属性检验,SCMOC对各量级降水目标轴长度和宽度都有扩大现象,中雨及以下量级雨带偏向西北-东南的较东北-西南明显偏多,大雨和暴雨雨带和实况接近。(4)目标面属性检验,SCMOC对各量级降水预报面积比实况面积偏大(除暴雨),量级越大,偏大程度越大；对小雨强度预报偏强,中雨及以上量级预报强度偏弱；小雨预报较实况偏西偏南、中雨及以上量级偏西偏北。(5)典型个例检验,SCMOC对大雨以下量级降水空间分布形态能够较好地把握,雨带和雨强预报较好,但对大雨及以上量级雨带及雨强预报能力较弱,位置偏北偏西,强度偏弱。

Abstract:

The objective test and evaluation of the model forecast effect provide an important reference for improving the model forecast performance and improving the forecaster"s localization correction. Using the refined grid precipitation forecast guidance product SCMOC and the three-source fusion grid live precipitation analysis data (CMPAS) of the National Meteorological Center from May to September 2019-2022, based on the classical test and the MODE spatial test method, the precipitation forecast performance of the intelligent grid forecast SCMOC in the Hexi Corridor was tested and analyzed. Results showed: (1) The TS of precipitation forecast by SCMOC varies with the magnitude of precipitation. The smaller the magnitude is, the larger the TS is, the larger the magnitude is, the larger the deviation is, and the smaller the hit rate is. (2) The target similarity of SCMOC for precipitation forecast below rainstorm reaches 0.71 ~ 0.75, and the target similarity of rainstorm is 0 ~ 0.49. (3) The target axis attribute test shows that SCMOC has expanded the length and width of the target axis of each magnitude of precipitation. The rain bands of moderate rain and below are more obvious in the northwest-southeast than in the northeast-southwest, and the heavy rain and rainstorm rain bands are close to the actual situation. (4) The target surface attribute test shows that the forecast area of SCMOC for each magnitude of precipitation is larger than the actual area (except rainstorm), and the larger the magnitude, the greater the degree of bias ; the forecast intensity of light rain is stronger, and the forecast intensity of moderate rain and above is weaker. The light rain forecast is more westward and southward than the actual situation, and the moderate rain and above are more westward and northward. (5) The typical case test shows that SCMOC can better grasp the spatial distribution pattern of precipitation below heavy rain, and the forecast of rain belt and rain intensity is better, but the forecast ability of rain belt and rain intensity of heavy rain and above is weak, and the position is north and west, and the intensity is weak.