| Soil salinization is one of the main natural disasters in the world,especially in arid and semi-arid areas.Optical remote sensing has many characteristics,such as rich spectra information,reflecting earth surface characteristics plentifully,and imaging clearly.It has become an important data source for remote sensing monitoring on soil salinization.However,the characteristics and mechanism of optical remote sensing quantitative monitoring on soil salinization are not still being unsedstood enough.In view of the problems existing in the optical remote sensing monitoring on soil salinizayion,this paper deeply and systematically carried out optical remote sensing quantitative monitoring on soil salinization in different stages,degrees and types by choosing different study areas.The results and the characteristics of optical remote sensing monitoring on soil salinization were obtained.The research goals of this paper is to study the method of improving the precision of remote sensing monitoring on soil salinization,taking the measured soil hyperspectral data and remote optical sensing data as the main data sources,by combining the study of remote sensing monitoring on regional soil salinization with the study of regional soil salinity characteristics and change regularity.Optical remote sensing quantitative monitoring on soil salinization at regional scale was realized,based on regional soil salinization characteristics and the relation between soil salinity and soil spectra.By analyzing the relation between soil salinity,soil salt composition,soil salt composition content and soil spectra,the characteristics of optical remote sensing quantitative monitoring on soil salinization were explored to provide reference for optical remote sensing quantitative monitoring on soil salinization.Four aspects of the research were carried out around the research goals,including the relation between regional soil salinity,salt composition and soil spectra of remote sensing data;the relation between surface,deep soil salinity and soil spectra;optical remote sensing quantitative monitoring on regional soil salinization and its precision evaluation;experimental verification and analysis of optical remote sensing monitoring results of soil salinization.The study methods were correlation analysis,nonlinear regression analysis,mediating effect analysis,clustering analysis,partial least squares regression modeling,paired sample t test,hyperspectral data processing methods and modeling methods for nonlinear relation among variables,etc.The main work and achievements of this paper are as follows:Firstly,quantitative monitoring on soil salinization at the soil salinity decreasing stage in Kuqa oasis study area based on Hyperspectral data.At the soil salinity decreasing stage in Kuqa Oasis in October 2010,the soil salinity decreased gradually from top to bottom in 0-10 cm,10-30 cm and 30-50 cm layers,and the relationship between deep soil salinity and surface soil salinity was relatively strong power function,this law also existed for the main salt ions of Na+ and CI-content.NaCI is the main salt in the soil layers.For 0-10cm surface soil,the band with the greatest correlation between soil spectral reflectance and soil salinity,and main salt composition is same,HJ-1A hyperspectral data can be used to analyze the main salt composition in soil.Mediating effect analysis shows that deep soil salinity affects surface soil spectral reflectance by influencing surface soil salinity.A remote sensing modeling method integrating soil salinization mechanism and surface parameters is proposed to estimate soil salinity,salinity indices for soil salinity monitoring were constructed,the surface parameters for soil salinity monitoring are selected,and the accuracy of surface soil salinity estimation was improved.Secondly,quantitative monitoring on soil salinization at the soil salinity increasing stage in Kuqa oasis study area based on multi-pectral data.There was a strong power function relationship between deep layer soil salinity and surface layer soil salinity.Mediating effect analysis also shows that deep soil salinity affects surface soil spectral reflectance by influencing surface soil salinity.Based on the remote sensing modeling method integrating soil salinization mechanism and surface parameters,and the relationship between surface soil salinity and spectral reflectance of Ladndsat 5 TM data,different soil salinity indices were constructed,soil water content in deep layer,and NDWI(Normalized Difference Water Index)previously used for water body information extraction could improve surface soil salinity estimation accuracy.The selection of characteristic variables for the establishment of remote sensing monitoring model of soil salinity by multiple regression method is studied.For the multi-regression modeling using multi-source data,the relationship between characteristic variables will affect the accuracy of the estimation model,and the sensitive band is not necessarily the best modeling variable.The strong power function relationship between surface soil salinity and 0-10 cm soil salinity was used to estimate the salinity of deep soil.In this study area,the possibility of indirect monitoring on deep soil salinity by optical remote sensing existed at the soil salinity increasing stage.Thirdly,quantitative monitoring on soil salinization at the soil salinity decreasing stage in Ebinur Lake study area based on multi-spectral data.Na+CI-,SO42-and Ca2+content in surface soil are relatively large,and their differences are not very large.Soil salinity has a great correlation with these salt ions content,and the soil salinity of each layer is power function relationship.There are two main salts of NaCI and CaSO4 in the surface soil.The relationship between soil spectral reflectance and soil salinity,and Na' content,and CI' content show significant non-linear characteristics,but the relationship between soil spectral reflectance and SO42-content,and Ca2-content is linear or approximate linear.The sensitive band for soil salinity and main soil salt ions monitoring is consistent,in this region,multi-spectral data can be used to analyze the soil main salt composition to a certain extent.Several soil salinity indices based on the relationship between soil salinity and soil spectral reflectance were constructed,and are effective in improving the accuracy of estimating surface soil salinity.Finally,experimental verification and analysis of optical remote sensing monitoring results of soil salinization was carried out.Proved that the results of optical remote sensing monitoring of soil salinization are reliable,and some characteristics of quantitative monitoring on soil salinization by optical remote sensing are obtained.Based on the above results,this study has some innovations and contributions to optical remote sensing for quantitative monitoring on soil salinization and regional soil salinization.The innovations and contributions to optical remote sensing monitoring on soil salinization of this paper are as follows:?In mixed saline soils,if the salt composition of the soil was limited,especially in the case of only one main salt,the sensitive band for remote optical monitoring of soil salinity and the soil main salt content was consistent.In this case,the relationship between soil salinity and soil spectral reflectance was controlled by the main salts present in the soil,optical remote sensing can be used to analyze the main salt composition in soil to some extent.?The relationship between soil salinity and soil spectral reflectance is generally either linear or close to it.When mixed salts composed of two different salts are present,however,this relationship can be affected.If the mass ratio of the two salts is similar or identical,the relationship between soil salinity and soil spectral reflectance may have significant nonlinear characteristics.The influence of the nonlinear relationship on the accuracy of soil salinity estimation should be considered when using remote optical monitoring.? There are many factors that affect the selection of sensitive bands for remote optical monitoring of soil salinity,soil type,soil salt composition,and soil salt content.The proportion,quantity and mass of salts in mixed salts are some of these factors.? in some soils and for some soil salts,the relationship between the soil salt content and the soil spectral reflectance in some bands decreases with the increase of the salt content.? For the research field of optical remote sensing monitoring on soil salionization,the relation between soil salinity and soil spectra was extended to study the relation between soil salinity,soil salt composition and soil spectra,the research field is more microscopic and preciseMeanwhile,the innovations and contributions to the optical remote sensing monitoring on regional soil salinization of this paper are as follows:?There is a power function relation between soil salinity in the deep layer and that in the surface layer,it was possible to monitor deep layer soil salinity indirectly by using optical remote sensing data at the soil salinity increasing stage in the study area.? A remote sensing modeling method that integrating soil salinization mechanism and surface parameters for estimating soil salinity is proposed,the deep layer soil moisture,especially the soil moisture of 10-30cm layer and the NDWI that was used to extract water body information previously,can improve the soil salinity estimation precision in the surface layer of 0-10cm.In addition,the characteristics of this paper are mediating effect analysis on the relation between deep soil salinity and soil spectrum;the decision rules establishing on cluster analysis of soil salt ions content;the analysis of the relation between soil salt type and soil spectrum;and the selection of characteristic variables of multi-variable regression modeling for soil salinity estimation using remote sensing,and the views on sensitive waveband of soil salinization monitoring and modeling variables. |
PDF Full Text Request