CMPAS 1 km融合降水产品在四川省的精细化评估

Refined Evaluation of CMPAS 1 km Fusion Precipitation Products in Sichuan Province

  • 摘要: 利用2020年9月—2021年8月四川省地面自动站逐小时降水观测数据,对比检验了1 km逐小时降水实时(RT)与近实时(NRT)融合产品在四川省的适用性。结果表明:(1)RT与NRT的整体评估结果相近,NRT略优于RT。(2)两种融合降水产品的晴雨准确率均为96.83%,能较准确地反映出有无降水。(3)国家站在盆地与川西高原以高估为主,在攀西地区则以低估为主;区域考核站与非考核站在盆地均以高估为主,在川西高原与攀西地区则以低估为主,RT较NRT的分布更离散。(4)两种产品均具有与观测较一致的日变化趋势,误差峰值出现在强对流天气频发的夏季,谷值出现在降水较少的冬季。(5)NRT的分量级评估结果略优于RT;当降水量级为小雨时,融合产品均高估于实况;当量级为中雨至大暴雨时,融合产品低估于实况,且随着降水量级的增大,误差随之增大,TS评分减小。

     

    Abstract: Using the hourly precipitation observation data of Sichuan Automatic Ground Station from September 2020 to August 2021, the applicability of the 1 km hourly precipitation real-time (RT) and near real time (NRT) fusion product developed by the National Meteorological Information Center in Sichuan Province was evaluated carefully. The results show that: (1) The overall evaluation results of RT and NRT are similar, and NRT is slightly better than RT. (2) The accuracy of the two fusion precipitation products is 96.83%, which can accurately reflect whether there is precipitation. (3) From the perspective of spatial distribution, the national stations in the Basin and the West Sichuan Plateau are mainly overestimated, while the national stations in Panxi are mainly underestimated. The regional assessment stations and non-assessment stations in the Basin are mainly overestimated, while the regional assessment stations in the West Sichuan Plateau and Panxi are mainly underestimated. The distribution of RT is more discrete than that of NRT. (4) From the time distribution point of view, both products have a daily variation trend which are consistent with the observation. The error peak value appears in the summer with frequent strong convective weather, and the valley value appears in the winter with less precipitation. (5) From the test of precipitation component level, fusion products overestimate the reality when the precipitation level is light rain, fusion products underestimate the reality when the magnitude is moderate rain to heavy rain, and with the increase of precipitation level, the error increases and the TS score decreases, and the NRT component level evaluation is slightly better than RT.

     

/

返回文章
返回