摘要
FY-3D可降水量(precipitable water vapor,PWV)是我国新一代极轨气象卫星提供的新型遥感水汽产品。针对降水、云、地表反射等因素影响导致的FY-3D PWV产品精度不高的问题,提出了一种融合GNSS和随机森林算法的FY-3D PWV高精度校正模型。该模型使用随机森林算法作为核心结构,充分考虑FY-3D PWV产品的多类型影响因素,以FY-3D像元的PWV值、时间、位置、高程、地表类型以及观测姿态等特征参数为输入变量,以高精度GNSS PWV为目标变量,利用美国西部地区2020年的实测数据进行了模型训练和测试。实验结果表明,相比于初始产品和线性校正模型,该模型分别使FY-3D PWV产品的精度提高了40.68%和36.07%,说明了其在FY-3D水汽产品校正中的优越性和稳定性。
FY-3D precipitable water vapor(PWV)is a new remote sensing water vapor product offered by China’s latest generation of polar-orbiting meteorological satellites.However,the accuracy of FY-3D PWV products is often affected by precipitation,clouds,and surface reflection.To address this issue,a high-precision calibration model for FY-3D PWV combining global navigation satellite system(GNSS)and random forest algorithm is proposed.The proposed model utilizes the random forest algorithm as its core structure and takes into account various factors that influence the accuracy of FY-3D PWV products.The PWV value,time,position,elevation,surface type,and viewing posture of FY-3D image pixels are set as the input variables of the model.Additionally,it uses high-precision GNSS PWV as the target variable and conducts model training and testing using measured data from the western region of the United States in 2020.The experimental results show that compared with the initial product and linear correction model,the proposed model improves the accuracy of FY-3D PWV product by 40.68%and 36.07%,respectively,which illustrates the superiority and stability of this model in the correction of FY-3D water vapor product.
作者
贺致芬
石娜
宋禄楷
焦英华
张文渊
HE Zhifen;SHI Na;SONG Lukai;JIAO Yinghua;ZHANG Wenyuan(Satellite Application Center,Shandong Provincial Institute of Land Surveying and Mapping,Jinan 250102,China;Basic Surveying and Mapping Center,Shandong Provincial Institute of Land Surveying and Mapping,Jinan 250102,China;School of Environment and Spatial Informatics,China University of Mining and Technology,Xuzhou,Jiangsu 221116,China)
出处
《遥感信息》
CSCD
北大核心
2023年第6期117-123,共7页
Remote Sensing Information
