基于光谱指数建模的河套灌区盐渍土盐分反演  

Salinity Inversion of Saline soil in Hetao Irrigation District Based on Spectral Index Modeling

作　　者：玉玉 吴月茹 左合君 王海兵 闫敏 YU Yu;WU Yue-ru;ZUO He-jun;WANG Hai-bing;YAN Min(College of Desert Management,Inner Mongolia Agricultural University,Hohhot 010018,Inner Mongolia,China;Key Laboratory of Aeolian Physics and Desertification Control Engineering from Inner Mongolia Autonomous Region,Hohhot 010018,Inner Mongolia,China;Inner Mongolia Autonomous Region Colleges and universities innovation team development plan,desert sand ecological protection and management technology innovation team,Hohhot 010018,Inner Mongolia,China)

机构地区：[1]内蒙古农业大学沙漠治理学院,内蒙古呼和浩特010018 [2]内蒙古自治区风沙物理与防沙治沙工程重点实验室,内蒙古呼和浩特010018 [3]内蒙古自治区高等学校创新团队发展计划,沙漠沙地生态保护与治理技术创新团队,内蒙古呼和浩特010018

出　　处：《中国农村水利水电》2024年第11期185-195,共11页China Rural Water and Hydropower

基　　金：国家自然科学基金地区项目(41961051);人才引进优秀博士科研项目(NDYB2017-11)。

摘　　要：探究遥感影像高效准确的提取河套灌区土壤盐渍化信息的机理,通过构建特征模型对河套灌区盐分进行反演,为河套一带土壤盐渍化治理提供数据参考。基于Landsat 7 ETM+、Sentinel-2遥感影像数据,根据不同波段提取光谱参数:盐分指数SI(Salinity Index,SI)、归一化植被指数NDVI(Normalized Difference Vegetation Index,NDVI)、地表反照率指数(Albedo)以及修改型土壤调节植被指数MSAVI(Modified Soil-Adjusted Vegetation Index,MSAVI)。结合以上光谱参数构建遥感盐分监测指数模型(Salinization Detection Index,SDI),对河套灌区盐渍化土壤信息进行提取分析,并结合实测数据构建混淆矩阵对模型精度进行了验证,选取精度验证结果较高的模型分析河套灌区土壤盐渍化程度。通过混淆矩阵对模型进行精度评价后得到SI-Albedo特征空间模型SDI2模型的总体精度最高,达到86.79%,Kappa系数为0.82,SDI1、SDI3和SDI4与混淆矩阵的总体分类精度分别为79.25%、45.28%和69.81%。结果表明,SI-Albedo特征空间模型SDI2对河套灌区盐分信息提取及反演较为适宜,在四类遥感盐分监测指数模型中,SI-Albedo特征空间模型SDI2对研究区的盐分反演具有较强的参考意义。This paper intends to explore the mechanism of efficient and accurate extraction of soil salinization information in Hetao Irrigation District by remote sensing images,and invert the salinity of Hetao Irrigation District by constructing a feature model to provide data reference for soil salinization control in Hetao area.Based on Landsat 7 ETM+and Sentinel-2 remote sensing image data,spectral parameters were extracted according to different bands:salinity index(SI),normalized difference vegetation index(NDVI),surface albedo index(Albedo)and modified soil-adjusted vegetation index(MSAVI).Combined with the above spectral parameters,a remote sensing salinity detection index(SDI)model was constructed to extract and analyze the salinized soil information in Hetao Irrigation District,and the accuracy of the model was verified by combining the measured data and constructing the confusion matrix.The model with higher accuracy verification results was selected to analyze the degree of soil salinization in Hetao Irrigation District.After evaluating the accuracy of the model through the confusion matrix,the SI-Albedo feature space model SDI2 model has the highest overall accuracy,reaching 86.79%,and the Kappa coefficient is 0.82.The overall classification accuracy of SDI1,SDI3 and SDI4 with the confusion matrix is 79.25%、45.28%and 69.81%,respectively.The results show that the SI-Albedo feature space model SDI2 is more suitable for the extraction and inversion of salt information in Hetao Irrigation District.Among the four types of remote sensing salt monitoring index models,the SI-Albedo feature space model SDI2 has a strong reference value for salt inversion in the study area.

关 键 词：光谱指数 河套灌区 遥感盐分监测模型 混淆矩阵 

分 类 号：TV93[水利工程—水利水电工程] S156.41[农业科学—土壤学] TP751[农业科学—农业基础科学]