青海MODIS可降水量数据校正研究  

Study on the data correction of MODIS precipitable water vapor of Qinghai province

作　　者：王佳 李琼[2] 魏加华[1,2,3] WANG Jia;LI Qiong;WEI Jiahua(School of Water Conservancy and Electric Power,Qinghai University,Xining 810016,China;State Key Laboratory of Plateau Ecology and Agriculture,Qinghai University,Xining 810016,China;State Key Laboratory of Hydro Science&Engineering,Tsinghua University,Beijing 100084,China)

机构地区：[1]青海大学水利电力学院,青海西宁810016 [2]三江源生态与高原农牧业国家重点实验室,青海大学,青海西宁810016 [3]清华大学水沙科学与水利水电工程国家重点实验室,北京100084

出　　处：《青海大学学报》2021年第2期77-84,共8页Journal of Qinghai University

基　　金：国家重点研发计划项目(2017YFC0403600);青海省科学技术厅重点研发与转化计划项目(2019-SF-146);青海省科学技术厅自然科学基金项目(2019-ZJ-968Q)。

摘　　要：为了获取高精度且空间连续的水汽数据,利用无线电探空数据对地基GNSS探测反演的PWV数据和MODIS PWV数据进行精度验证和相关性分析;应用地基GNSS PWV数据对MODIS PWV数据进行校正,分别建立基于线性回归和BP神经网络的MODIS PWV数据校正模型,并对校正后的MODIS PWV数据进行效果评价。结果表明:GNSS PWV数据与探空数据的平均偏差为1~2 mm,而MODIS PWV数据与探空数据相比平均偏差为1~6 mm,需进行校正。两种方法构建的模型对MODIS PWV数据校正后精度都有所提高,平均相对误差减少了0.89 mm,精度提高了27%,BP神经网络方法的校正效果更好。In order to obtain the data of water vapor with high-precision and spatial continuity,radiosonde data are used in this paper to make precision verification and correlation analysis for PWV data retrieved from atmospheric sounding of ground-based GNSS and MODIS PWV data.MODIS PWV data are corrected by using PWV data of ground-based GNSS.The correction models of MODIS PWV data based on linear regression and BP neural network are established respectively,and the effects of the corrected MODIS PWV data are evaluated.The results show that the average deviation between GNSS PWV data and radiosonde data is 1-2 mm,while that between MODIS PWV data and radiosonde data is 1-6 mm,which needs to be corrected.The accuracy of the corrected MODIS PWV data is improved by the two methods.The average relative deviation is reduced by 0.89 mm,and the accuracy is improved by 27%.BP neural network method has better correction effect.

关 键 词：MODIS GNSS 可降水量 线性回归方法 BP神经网络 

分 类 号：P413[天文地球—大气科学及气象学] TP79[自动化与计算机技术—检测技术与自动化装置]