| Soil is a significant component of ecosystem, and it is also the organic carrier forthe connection of animal and plant. With the development of society, soil environmentproblem is becoming more and more serious. Presently, land degradation anddesertification are the key limitation to potential of soil. Soil salinity has a greatinfluence on Songliao Plain. Salinization often occurred in areas where salinegroundwaters are elevated to where they approach the ground surface and evaporationexceeds precipitation. Human activities may also cause soil salinity. Soliao Plain isthe area with the highest concentration of soil salinity, where salinized soils are nearly60%of the total land area and soil salinity damages soil properties, while at the sametime restricting the development of agriculture industry and breaking the ecologicalbalance. Soil salinity has become a giant limitation to human production activity.Through the study of soil salinity, its information can be gathered from multi-angles.Besides, the degree of salinization, the distribution range, the geographic location andthe development process can also be effectively distinguished. The thematic map ofsoil salt content will provide assistant information for land administration departmentand agriculture production department. It will also offer basis for policy making andthe prevention of soil salinization.Previous techniques for identifying and monitoring salinized land are mainlybased on sampling, result testing and field survey, which not only makes thetechnique difficult and time-consuming, but also brings great difficulty to broad-scalequantitative evaluation. Therefore, efforts are being made to develop morecost-effective monitoring system. The development of remote sensing provides abrand-new method for the monitoring and mapping of soil salt content. This techniqueis non-contact and remote, which is able to monitor fast and effectively from amicroscopic perspective. It not only is economic, but also improves accuracy. At thebeginning, when remote sensing was applied to land monitoring, salt-affected soil isdistinguished mainly by multispectral remote sensing. The development ofhyperspectral remote sensing is based on analysis of component and structure.This paper is to study the salt composition of Songliao plain saline soil, and takespectrum curve characteristics as the foundation, by the method of the spectralanalysis, drawing the sensitive range of this area, and using the least square to supportvector machine regression analysis method to establish the relative model of the soilsalt content and its correlation judgment index and saline soil spectral reflectance.Oncondition of the hyperspectral satellites feasibility testing environment, apply themodel to the content of soil salt space distribution for information extraction, combined with the field test data sample to estimate the result evaluation of the salinesoil salt content space distribution in Songliao plain to reveal the monitoring ability inenvironmental disaster reduction satellite in soil salinization and the salt quantitativeestimate ability.The primary results of this dissertation are listed below:1. Soil content feature of Songliao Plain. Chemical analysis of samples showsthat in the top layer of soils, the predominant cation is Na+, which is86.30%of thetotal cation and that the predominant anion are HCO3-and CO32-, which is85.28%ofthe total anion. Through statistics analysis of the data, it can be shown that there isobvious correlation between soil salinity and HCO3-, CO32-and Na+. This may be dueto the soil forming materials causing ion combining ways mainly by sodium ions andcarbonated root ion and bicarbonate.2. Analysis of the feasibility of environmental mitigation satellite hyperspectraldata application. The application of environmental mitigation satellite hyperspectraldata is not widespread. In order to test whether the hyperspectral obtained from thesatellite has practical value for the study of soil salt content. This paper adopts theapproach of comparative data, and uses the presently wide Hyperion data ascompared object, removing the bad line and stripe, and applying the FLAASH modelto the image data by the atmosphere correction, using the MNF transform atmosphereto correct the post-processing, get the optimization of the spectrum image data At last,through correlation analysis and differential analysis, after the comparison of the twogroups of data, it shows that environmental mitigation satellite hyperspectral data haspractical application value.3. Based on the spectral features of salt-affected soil, sensitive wave bands ofsalt-affected soil are confirmed. Combined with previous findings, it is assured thatthe main clay mineral contents of Songliao Plain are kaolinite, chlorite, illite andmontmorillonite. Use USGS spectrum library data, and draw the clay mineralextraction of spectral features and do the analysis. And with the foundation of thesample point spectral data, and take the method of spectral analysis, with the originalspectrum curve, first order differential curve and second order differential curve. Anddetermine primarily the sensitive bands of this area’s saline soil are in the range of460-508nm,540-571nm, and765-827nm.4. Environmental mitigation satellite hyperspectral remote sensing soil saltcontent estimation. In order to make soil salt content distribution map, NDVI wasused in this study to make the separation between soil and vegetation, water andtowns. The models that are previously established based on LS-SVM were applied toEnvironmental mitigation satellite hyperspectral data for mapping of soil salt content.R2=0.784, RMSE=1.037are obtained for environmental mitigation satellite method.Through the analysis of soil salt content distribution map, it is obvious that the resultof this thesis is in accordance with previous findings.5. Environmental mitigation satellite hyperspectral remote sensing soil saltcontent estimation. In order to make soil salt content distribution map, NDVI wasused in this study to make the separation between soil and vegetation, water andtowns. The models that are previously established based on LS-SVM were applied to Environmental mitigation satellite hyperspectral data for mapping of soil salt content.R2=0.784, RMSE=1.037are obtained for environmental mitigation satellite method.Through the analysis of soil salt content distribution map, it is obvious that the resultof this thesis is in accordance with previous findings. |
PDF Full Text Request