Monitoring Of Desertification On The Edge Of Badain Jaran Desert In Recent 20 Years Based On Remote Sensing Imagery

Badain Jaran Desert, which is located in the center of arid and semi-arid climate region in middle Asian, is the world's third largest, China's second largest desert, and is also one of the strongest dust storms source in northern China. Nearly half a century, under the influence of the westerly circulation, winter and summer monsoon wind, especially in the winter monsoon wind, climate change and human activities, the sand dunes in Badain Jaran Desert were expanding continuously, and a large number of lands on the edge of the desert were desertified actively. In this situation, Badain Jaran Desert gradually closed to the Tengger Desert in its southeastern and Ulan Buh Desert in its east and directly impacted on the people's livelihood and social development of Ejinaqi, Cosby in Inner Mongolia Autonomous Region and Minqin County, Gaotai County, Linze County in Gansu Province.In this study, we had the theory of Earth System Science as a guide, the spatial and temporal dynamic changes of desertification on the edge of Badain Jaran Desert as the research objects, and nearly 20a of the satellite remote sensing data, meteorological data and socio-economic data as the based analysis.Combined with field survey and experimental study, we analysed the spectral characteristics of desertified land, compared and discussed the accuracy and applicability of different methods monitored desertification on the base of remote sensing imagery, and with the methods of integrated interdisciplinary study, researched the features of spatial and temporal dynamic changes of desertification on the edge of Badain Jaran Desert in recent 20a and the interaction between these changes and climate change and human activities. We aimed to establish promptly monitoring system of desertification based on remote sensing for providing suggestions of desertification prevention and promoting the sustainable development among regional nature, ecological environment and human activities.Key results and conclusions were as follows:(1) After analyzing field spectral experimental data, we found that reflectance of sand increased at 350～760nm visible spectral region, relatively stabilized at 760～1000nm spectral region, and followed by a wavy change. The reflectance of sand shown a basic trend of gradually increasing as its color deepened and its granularity enlarged. As the fraction of vegetation coverage decreased, the surface reflectance gradually decreased at 350～700nm visible spectral region and increased at 700～1050nm near-infrared spectral region. And at 700～1050nm near-infrared spectral region, the more serious desertification, the greater its surface reflectance. Vegetation indicated to the degree of desertification in the the exterior level of landscape through its interior spectral characteristics at near-infrared region.(2) After analyzing the overall accuracy and Kappa coefficient of each classification method on test area 1 and 2, we found that the overall accuracy and kappa coefficient from feature space method were the best in both two test areas and separately to 76.99% and 0.6825 in the test area 2. The major factors in reducing the classification accuracy included the complexity, heterogeneity and fragmentation of each degree desertification landscape, as well as the small difference and low contrast among different degree desertified lands. The applicability of each method was different for specific monitoring regions and tasks. These classification methods based on vegetation indices had high accuracy in semi-arid and sub-humid regions with good vegetation condition and were good for the high accuracy, systematic and comprehensive desertification monitoring tasks.(3) The degree of desertification on the edge of Badain Jaran Desert gradually decreased from desert boundary regions to outside regions in 2010. The land of very severe desertification and severe desertification accounted for the main part, and of moderate and slight desertification were both less. From 1990 to 2000, the desertification on the edge of Badain Jaran Desert displayed a growth trend that the spread and the severity of desertification were coexistence. From 2000 to 2010, the desertification on the edge of Badain Jaran Desert expressed a decreasing trend that the reverse of desertification was greater than the expansion of it. The analysis of meteorological data and human activity indicated that the natural environment was the background of desertification process and human factors had an effect of aggravating or mitigating the process of desertification in the study area.