多源降水量数据融合分析及评价  

Analysis and evaluation of multi-source merging precipitation data

作　　者：王福增[1] 温凯 何山 谷晓平 WANG Fu-zeng;WEN Kai;HE Shan;GU Xiao-ping(Key Laboratory of Atmospheric Sounding of China Meteorological Administration,College of Electronic Engineering,Chengdu University of Information Technology,Chengdu 610225,China;Ecological Meteorology and Satellite Remote Sensing Center of Guizhou Province,Guiyang 550005,China)

机构地区：[1]成都信息工程大学电子工程学院,中国气象局大气探测重点实验室,成都610225 [2]贵州省生态气象和卫星遥感中心,贵阳550005

出　　处：《兰州大学学报（自然科学版）》2023年第5期647-654,共8页Journal of Lanzhou University(Natural Sciences)

基　　金：国家自然科学基金面上项目(4207050768);国家重点研发计划项目(2018YFC150720,2018YFC1506605);四川省自然科学基金项目(2022NSFSC0208)。

摘　　要：利用2018年5月贵州省黔西南州自动气象观测站的降水量资料、兴义站雷达资料和CMORPH卫星降水反演产品资料,采用最优插值法对3种降水量数据开展融合试验.试验设计4种降水量数据融合方案,分别对融合的降水量数据进行区域精度检验.结果表明,经概率密度匹配法订正后的卫星-地面融合降水量与订正之前相比,相关系数由66.19%提升至79.68%,偏差由-0.544mm/h降至-0.247 mm/h,均方根误差由2.944 mm/h降至1.353 mm/h.当融合降水模型仅输入2种降水量数据源时,具有较高精度的一方对融合降水量的精度影响更大.卫星-雷达-地面数据融合降水量的相关系数为81.32%,偏差和均方根误差分别为-0.219、1.037 mm/h,精度高于其他降水量数据融合方案,说明融入更多有效降水量数据源能明显提高面降水量的精度.With the precipitation data of an automatic meteorological observation station in Qianxinan Prefecture of Guizhou Province in May 2018,radar data of Xingyi Station and the CMORPH satellite precipitation products,an optimal interpolation method was used to carry out merging experiments on three types of precipitation data.In this experiment,four precipitation merging schemes were designed,and the regional accuracy of the merging precipitation data tested respectively.The results indicated that satelliteground merging precipitation,compared to that before the correlation coefficient,decreased from 66.19%to 79.68%,the deviation decreased from-0.544 to-0.247 mm/h and the root mean square error decreased from 2.944 to 1.353 mm/h after the probability density matching method was adopted.The accuracy of the merging precipitation was higher after the probability density matching method was adopted.When only two precipitation data sources were input to the merging precipitation model,the one with a higher accuracy had a greater impact on the accuracy of the merging precipitation.The results of satellite radar ground merging precipitation evaluation showed that the correlation coefficient reached 81.32%,the deviation and root mean square error were-0.219 and 1.037 mm/h respectively,and the accuracy higher than other merging precipitation schemes,meaning that the accuracy of area precipitation could be significantly improved by integrating more effective precipitation data sources.

关 键 词：CMORPH卫星 最优插值法 高分辨率降水数据融合 

分 类 号：P412[天文地球—大气科学及气象学]