珠海一号高光谱反演表层土壤有机质含量方法研究  

Inversion of Farmland Soil Organic Matter Content by Hyperspectral Remote Sensing Images of Zhuhai-1

作　　者：赵俊 杜文涛 李新萍 王智枭 孟成真 张启超 张辉 孙子文 张惠怡 ZHAO Jun;DU Wentao;LI Xinping;WANG Zhixiao;MENG Chengzhen;ZHANG Qichao;ZHANG Hui;SUN Ziwen;ZHANG Huiyi(Henan First Geology and Mineral Survey Institute Co.,Ltd.,Luoyang 471000,China;Key Laboratory of Au-Ag-Polymetallic Deposit Series and Deep-seated Metallogenic Prognosis of Henan Province,Luoyang 471000,China;School of Geography and Tourism Luoyang Normal University,Luoyang 471000,China;Mineral Resources Exploration Center of Henan Geological Bureau,Zhengzhou 450006,China)

机构地区：[1]河南省第一地质矿产调查院有限公司,河南洛阳471000 [2]河南省金银多金属成矿系列与深部预测重点实验室,河南洛阳471000 [3]洛阳师范学院地理与旅游学院,河南洛阳471000 [4]河南省地质局矿产资源勘查中心,河南郑州450006

出　　处：《地理空间信息》2025年第4期67-70,共4页Geospatial Information

基　　金：河南省自然资源厅2021年科技项目(豫自然资函[2021]157号)。

摘　　要：高光谱反演有机质研究大部分聚焦在小尺度区域研究,为了更好地体现其大尺度模拟的有效性和实用性,以伏牛山脉北麓的伊川县为研究区,珠海一号高光谱卫星(orbita hyper spectral,OHS)和农田实测表层(0~20 cm)土壤有机质(soil organic matter,SOM)含量为对象,采用多元逐步回归、偏最小二乘和随机森林方法探索OHS数据反演县尺度有机质含量。结果表明,随机森林反演模型的拟合度较高(R^(2)=0.68),均方根误差为1.89。利用OHS数据反演有机质能够快速掌握监测区域有机质空间分布状况,为高光谱大尺度反演有机质研究提供了依据和参考。Most hyperspectral inversion research of organic matter focus on small scale area.In order to demonstrate the validity and practicability of large scale simulation,taking Yichuan County at the northern foot of Funiu mountain as the study area,taking Zhuhai-1 hyperspectral satellites(OHS) and measured surface(0~20 cm) SOM content as the objects,we used multiple stepwise regression,partial least squares and random forest methods to invert county SOM content from OHS data.The results show that the random forest inversion model has a high degree of fit(R~2=0.68),RMSE is 1.89.The inversion of organic matter with OHS data can quickly grasp the spatial distribution of organic matter in the monitoring area,and provide a basis and reference for hyperspectral large-scale inversion of organic matter research.

关 键 词：高光谱 有机质 土壤反演 珠海一号 

分 类 号：P237[天文地球—摄影测量与遥感]