基于随机森林的AMSR2青藏高原有云地区水汽反演  

AMSR2 Retrieval of Water Vapor in Cloudy Areas of Qinghai-Tibet Plateau Based on Random Forest

作　　者：陈明 王永前[1] 吴锡[2] CHEN Ming;WANG Yongqian;WU Xi(College of Resources Environment,Chengdu University of Information Technology,Chengdu 610225,China;College of Computer Science,Chengdu University of Information Technology,Chengdu 610225,China)

机构地区：[1]成都信息工程大学资源环境学院,成都610225 [2]成都信息工程大学计算机学院,成都610225

出　　处：《遥感信息》2022年第2期84-90,共7页Remote Sensing Information

基　　金：第二次青藏高原综合科学考察研究项目(2019QZKK0206);国家自然科学基金项目(42075142);高原与盆地暴雨旱涝灾害四川省重点实验室科技发展基金项目(SCQXKJYJXMS202114)。

摘　　要：针对光学遥感数据在有云地区的水汽反演精度低,提出了一种利用星载微波辐射计数据基于随机森林回归模型反演的青藏高原有云地区大气水汽的方法。利用2017年1月和7月的GPS水汽数据和AMSR2亮温数据,通过随机森林算法构建非参数及非线性的回归预测模型,估算青藏高原有云地区的大气水汽,并利用GPS水汽数据和再分析ERA-Interim数据对反演结果进行对比分析。结果表明:基于随机森林回归模型估算的青藏高原有云地区大气水汽效果较好,模型反演的水汽与GPS水汽之间具有较高的相关性,平均相关系数为0.898,平均均方根误差为0.341 cm;以GPS水汽进行验证,模型反演的有云像元水汽精度高于MODIS水汽产品;反演得到的青藏高原大气水汽分布与ERA-Interim水汽分布特征基本一致。Aiming at the low accuracy of water vapor inversion of optical remote sensing data in cloudy regions,a method is proposed for the inversion of atmospheric water vapor in cloudy regions of the Qinghai-Tibet plateau based on a random forest regression model using satellite-based microwave radiometer data.Using GPS water vapor data and AMSR2 bright temperature data for January and July 2017,a nonparametric and nonlinear regression prediction model is constructed by the random forest algorithm to estimate atmospheric water vapor in cloudy areas of the Qinghai-Tibet plateau,and the inversion results are compared and analyzed using GPS water vapor data and reanalysis of ERA-Interim data.The results show that:1)The effect of atmospheric water vapor in the clouded areas of the Qinghai-Tibet plateau estimated based on the random forest regression model is better,and the water vapor retrieved by the model has a high correlation with GPS water vapor.The average correlation coefficient is 0.898,and the average root mean square error is 0.341 cm.2)Validated by GPS water vapor,the water vapor accuracy of cloud pixels retrieved by the model is higher than that of MODIS water vapor products.3)The distribution of atmospheric water vapor on the Qinghai-Tibet plateau obtained by inversion is basically the same as that of ERA-Interim data.

关 键 词：AMSR2 大气水汽 青藏高原 ERA-INTERIM 随机森林 GPS水汽 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]