| Oasis was the particular geographical landscape and the supporting body of socioeconomic development in arid area. Large-area land was abandoned and the NPP of vegetation descended because of soil salinization and secondary salinization, which heavily limit the development of agriculture. All of these had severely threatened the ecological safe, the sustainable development of society and economy.The downstream of Kaidu River was taken as the study area. Combined with remote sensing data, the soil salinization was graded based on traditional field investigation and the technique of GIS. The spatial and temporal changes were quantitatively analyzed. A prediction was done for the changes of land salinization in future based on the gray predictable model-GM(1,1). The driving forces of soil salinization were analyzed in the study area. The suggestion and contermeasures for preventing and controlling soil salinization were put forward.This paper got the following conclusions:1. The salt-affected soil of different types were analyzed by ASD in typical region. The conclusions were drew that reflective curve of light spectrum were controlled by five fold lines (0.35~0.71μm, 0.71~1.47μm, 1.47 ~ 1.97μm ,1.97~2.21μm and 2.21 ~ 2.50μm) and four absorbing belts (0.69~0.74μm, 1.38 ~ 1.53μm ,1.81~2.09μm and 2.09~2.25μm). Quantitative remote sensing retrieval of the salt-affected soil by an analysis of hyper spectral data was done.2.The downstream of Kaidu River was taken as the study area, the first band, the second band, the third band and the fourth band of CBERS images and NDSI index were synthesized, then with the synthesized images, the principal component transform, soil index transform and soil salinity index transform were processed. The false color composite image by these three indices to be synthesized was supervised classification and extracted information on soil salinization.3. The composite images of downstream of Kaidu River were supervised classified, and extracted information on soil salinization by the Maximum Likelihood,the Minimum distance method and the support vector machine method. The classification of support vector machine method was the best method by comparision and analysis. The kappa coefficient of classification was 0.894 and the result was coincident with current situation, which has offered the reliable basis for extensive dynamic monitoring of the salinized soil in the area.4. The images of MSS of 1973,TM of 1990, ETM+ of 2000,CBERS of 2007 were supervised classification by support vector machine method. The area and distribution of four periods were analyzed . According to the data of low-grade salinization soil, middle grade salinization soil, heavy-grade salinization soil in four periods. The prediction was done for the changes of land salinization in future based on the gray predictable model-GM(1,1).5. The origin of soil salinization in study area was analyzed, and studied the influence factors and dynamic changes of soil salinization. Population, utilization of water resources and model of agriculture development were the major factors. Combined with remote sensing technique and geostatistic method, the influence factors of the forming and development of soil salinization were studied, and the suggestion and contermeasures for preventing and controlling soil salinization were put forward.
|
PDF Full Text Request