| The vegetation coverage is recognized as the most effective index for land desertification evaluation. Desertification not only causes huge economic loss, but also decreases land resources. Desertification will cause natural disasters such as dust storms which results to ecological environment and a serious threat to the safety of human existence. Choosing typical desert area Ebinur Lake Basin as the study area, which is the main air wind inlet and important source of sand. Therefore, the research of vegetation coverage in Ebinur Lake basin not only can evaluate the degree of desertification in this region, but also can provide effective technical support for the formulation of desertification control strategy.The methods to obtain vegetation coverage are mainly field measurement and remote sensing interpretation method. Remote sensing method is fast and efficient, but being easily effected. There still exists problems which the spectral contribution under drought conditions is easily hided and spectral features different from green vegetation under moist condition. Therefore, how to improve the ability of remote sensing in arid areas has been one of the most serious challenge that is faced by the remote sensing vegetation. Based on this, choosing high end element involved in TM data extraction of vegetation coverage can significantly improve the accuracy with the characteristics of high resolution of GF-1 and long time dimension of TM data.This thesis first normalized GF-1 date and TM data. Use band regression method, NDVI regression method, pixel decomposition method respectively to extracte vegetation coverage, which were used to regression analysis field survey data. Through the comparison of the accuracy, we found that acquiring the image endmembers from GF-1 participating TM data extraction has higher accuracy. Through the four vegetation coverage of spatio temporal changes and driving force analysis,We Came to conclusion as following:(1) Extracting desert vegetation coverage, band regression model, NDVI regression model, the image decomposition method are all feasible; Extracting vegetation coverage from TM data based on the element decomposition method which select endmember from GF-1 can significantly improve the extraction accuracy about 3.74%.(2) The vegetation coverage of Bozhou in 1990, 2001, 2007 and 2014 are 0.58,0.61,0.59,0.64 respectively.The vegetation coverage of the Ebinur Lake Wetland is 0.49, 0.54,0.55,0.59 respectively. From the spatial location, vegetation coverage in Ebinur Lake Protection Zone is lower than Bozhou.From the view of spatial location, high vegetation in Bozhou region are around the northwest, southwest, south mountain and plain farming area; low vegetation area is mainly distributed in the Ebinur Lake region. The specific distribution characteristics is: in Bozhou, 1990-2001, middle mountain district(1000 m-2500 m) >high mountain(2500m and above) > hilly area(less than 1000 m) > the Ebinur Lake wetland protection area. 2001-2014, middle mountain district(1000 m-2500 m)> hilly area(less than 1000 m) > high mountain(2500m and above) > the Ebinur Lake wetland protection area.In Ebinur Lake Wetland, 1990-2001, region of experiment > region of buffer > region of core area; 2001-2014, region of buffer > region of experiment > region of core area.The coverage of vegetation are mainly concentrated in the 0-30%, which is decreasing with the passage of time. The vegetation coverage between 30% and 60% is relatively stable, between 60% and 80% and above 80% are a growing.(3) Through correlation analysis, we found: The positive correlation between coverage and nature(average annual precipitation, annual mean temperature) is less than coverage and human factors(population, GDP) in Bozhou. In addition, the establishment of protected areas in Ebinur Lake wetland take a big role in vegetation coverage increasing. |
PDF Full Text Request