基于电磁感应的干旱区土壤盐渍化定量遥感研究

以南疆典型干旱区Landsat 7 ETM+遥感图像为数据源,利用决策树分类法提取农业用地,并对农业用地进行移动式电磁感应调查(简称磁感调查)和光谱特征提取,同时分析磁感数据和图像光谱特征与土壤盐分含量的相关性,从而建立土壤盐分的定量反演模型。研究结果表明:土地利用类型决策树的分类精度达到93.75%,Kappa系数达0.915 4;经多元逐步回归分析,磁感调查获得的土壤盐分含量与差值植被指数(DVI)、ETM+图像第二波段像元值(B2)以及比值植被指数(RVI)间具有显著相关性,由此建立的遥感反演模型可用于土壤盐分含量的定量反演。经89个样点检验,基于磁感调查的土壤盐分遥感反演精度虽低于基于磁感调查的地统计空间分析的精度,但遥感定量反演值与磁感调查实测值仍具有良好的相关性,而且精度较高,因此利用本文方法进行土壤盐渍化大面积监测是快速有效的途径。

Research on Quantitative Remote Sensing of Soil Salinization in the Arid Area Based on Electromagnetic Induction

For the sake of quantitative remote sensing study of soil salinization in the typical arid area,Landsat 7 ETM+ image of the typical arid area in South Xinjiang was obtained.The land use type of farmland was extracted by decision tree classification.The correlation between soil salinization,etectromagnetic induction data and spectrum characteristics was analyzed by mobile electromagnetic survey and extraction of spectrum characteristics in farmland.On such a basis,a quantitative inversion model of soil salinization was obtained.Some results have been obtained: the land use classification has a favorable accuracy with a total precision of 93.75% and a Kappa coefficient of 0.9154;multiple regression indicates that there exists significant correlation between soil salinization detected by the mobile electromagnetic survey and DVI(Difference Vegetable Indice),B2(the value of band 2 of ETM+ images) and RVI(Ratio Vegetable Index),and that the inversion model of soil salinization can be used to identify salinized soils quantitatively.Results from 89 verification points show that,although the quantitative inversion accuracy of remote sensing is a little lower than that of geo-statistics analysis based on electromagnetic induction,the correlation between the inversion values and the measured values is favorable,and the accuracy is acceptable.Thus the means put forward in this paper is an rapid and effective technology for large-scale soil salinization monitoring.