马尾松林地土壤有机碳遥感估测  

Remote Sensing Estimation of Soil Organic Carbon in Masson Pine Forest Land

作　　者：吴南锟 刘健[1] 郑文英[1] 项佳[1] 张佳奇[1] 余坤勇[1] Wu Nankun;Liu Jian;Zheng Wenying;Xiang Jia;Zhang Jiaqi;Yu Kunyong(University Key Lab for Geomatics Technology and Optimize Resources Utilization in Fujian Province,Fujian Agriculture and Forestry University,Fuzhou 350002,P.R.China)

机构地区：[1]3S技术与资源优化利用福建省高校重点实验室(福建农林大学)

出　　处：《东北林业大学学报》2020年第1期68-73,87,共7页Journal of Northeast Forestry University

基　　金：国家自然科学基金面上项目(31770760)

摘　　要：采用卡萨生物圈(CASA)模型的遥感间接估算法对长汀县河田镇马尾松林地土壤有机碳进行模型构建。结果表明:结合多种植被指数构建的综合植被指数(ICV),缓解了归一化植被指数(INDV)在植被净第一性生产力(NPP)反演的饱和现象及高估现象,拟合精度比归一化植被指数提高了17.4%,说明综合植被指数在研究区土壤有机碳(SOC)的估算上有更高的契合度;运用综合植被指数构建的随机森林回归模型,对土壤有机碳预测综合精度(R^2=0.597 2,RMSE=1.76,RM=94.85%)比其他回归模型高,适用于研究区SOC的估算。By the CSAS model-based remote sensing indirect estimation method, the model inversion of soil organic carbon in Pinus massoniana forest land in Hetian Town, Changting County was constructed. The CVI(Comprehensive Vegetation Index) effectively alleviated the saturation phenomenon and overestimation of NDVI(Normalized Vegetation Index) in the NPP(Net Primary Productivity) inversion process which constructed by combining multiple vegetation indices, and the fitting accuracy was improved by 17.4% compared with NDVI, indicating that CVI is a higher degree of fit in the estimation of the SOC(Soil Organic Carbon). The SOC prediction based on CVI constructed random forest regression model(R^2=0.597 2, RMSE=1.76, RM=94.85%) is the most comprehensive model in multiple regression models, and is suitable for estimating SOC in the study area.

关 键 词：土壤有机碳 植被净第一性生产力 归一化植被指数 卡萨生物圈模型 遥感 

分 类 号：S771.8[农业科学—森林工程]