基于实测高光谱和电磁感应的绿洲土壤含水量估测  

Estimation of soil moisture spatial distribution based on measured spectral and electromagnetic induction instruments in the arid oasis

作　　者：周倩倩[1,2] 丁建丽[1,2] 杨爱霞[1,2] 宁娟[1,2] 谭娇[1,2] 杨斌[3] 

机构地区：[1]新疆大学资源与环境科学学院,乌鲁木齐830046 [2]绿洲生态教育部重点实验室,乌鲁木齐830046 [3]武警黄金第八支队,乌鲁木齐830057

出　　处：《干旱区资源与环境》2018年第3期152-157,共6页Journal of Arid Land Resources and Environment

基　　金：国家自然科学基金(U1303381;41261090);自治区重点实验室专项基金(2016D03001);自治区科技支疆项目(201591101);教育部促进与美大地区科研合作与高层次人才培养项目资助

摘　　要：以新疆渭-库绿洲为研究区,利用电磁感应和高光谱技术并以地形因子作为辅助参量,构建土壤含水量的支持向量机回归(Support Vector Regression,SVR)预测模型,采用泛克里金法(Universal Kriging)将预测的土壤含水量进行插值在地图上实现空间可视化。结果表明:1)表层土壤含水量与表观电导率具有良好的相关性,两种模式相结合建立的土壤含水量解译模型的拟合优度达到0.853。2)原始反射率光谱经微分变换后更能凸显出细微差异,利用原始一阶微分建立的SVR土壤含水量模型,预测集决定系数(R2)为0.913,相对分析误差(RPD)为2.06,该模型具有较高的预测精度和稳定性。3)表层含水量空间分布不均,由绿洲内部到荒漠-绿洲交错带再到荒漠呈现逐渐减少的趋势。综上所述,利用该支持向量机模型对绿洲土壤含水量的预测具有实际与理论意义。In this contribution,the delta oasis between the Weigan River and the Kuqa River was selected as our study area. The Support Vector Regression（ SVR） was established by using electromagnetic induction and hyperspectral techniques and topographic factors. And then the soil moistures were predicted by using Universal Kriging method to realize the data visualization. The results show that,firstly,the correlation coefficient of EM38 and soil moisture multivariate regression model reach up to 0. 853; the surface water content in the oasis varies from 2. 17 to 32. 30（ %）; the value of the coefficient of variation is 56. 80. Secondly,first-order differential transformation and second-order differential transformation is better than the original soil spectral reflectance.Thirdly,the support vector machine model has higher accuracy,its determination coefficient（ R2） is reached up to 0. 913; the ratio of performance to deviation（ RPD） is 2. 06. Accordingly,the support vector machine regression model can be used as a desirable monitoring model for oasis arid soli water content. It provides a new method for rapid,reliable and non-destructive estimation of arid oasis soil water content.

关 键 词：土壤含水量 电磁感应 高光谱 地形因子 

分 类 号：S152.7[农业科学—土壤学]