【作者】 李述；

【导师】 刘勇；

【作者基本信息】 兰州大学 ， 地图学与地理信息系统， 2006， 硕士

【摘要】 在干旱、半干旱区,植被覆盖度低、抗干扰性弱、区域生态脆弱,环境的小幅波动都会造成生态系统的深刻变化,从而表现为地表土地覆盖的明显变化,所以干旱区土地覆盖的形成和发展深受全球环境变化的影响,是全球变化特别是LUCC研究的重要场所。遥感技术有大面积同步观测、时效性、经济性等优点,被广泛应用于土地利用/土地覆盖变化和荒漠化研究中,因此开展干旱、半干旱区的土地利用/土地覆盖变化和荒漠化遥感监测研究有着重要的意义。论文的主要内容有: 本文以1:5万地形图、野外调查数据、1987年的Landsat TM影像和2002年的Landsat ETM+影像、中宁县和景泰县1987～2002年的气象观测数据、1987年～2001年的NOAA/AVHRR 8km×8km的NDVI数据为基础,对数据完成了以下处理和分析: 完成几何纠正后,对数据进行了辐射和大气校正,地表反射率反演,并对校正的结果进行了评估。为解决研究区的同物异谱和异物同谱现象,基于空间位置信息和发现的知识对监督分类结果进行了分类后处理,最后基于波谱阈值划分了盐碱地、沙漠和沙漠化土地、其它荒漠类型,完成了影像的分类,并对分类结果进行了精度评估。利用状态转移矩阵表统计了1987年和2002年各种土地覆盖类型的面积变化,并对面积变化较大的地物类型进行了空间分布变化分析,然后对1987年和2002年研究区的植被覆盖度进行了分析,最后结合中宁县和景泰县1987～2002年的气象观测数据、研究区1987～2001年的NOAA/AVHRR 8km×8km的NDVI植被指数对腾格里沙漠东南边缘地区近15年来的降水、自然植被和生态环境变化做了初步的探讨。 得到下述结论: ① 基于遥感影像进行多时相的对比分析或利用遥感进行定量分析的研究中,辐射和大气校正十分重要。本文通过对遥感影像校正前后数据的差异对比发现,校正前两景陆地卫星影像在第1波段和第5波段的差异分别为85%和70%以上,而校正后的差异都减小到5%以内。更多还原

【Abstract】 It is characterized by exiguous vegetation, weak resistance and fragility ecosystem in arid and semiarid region, minor environmental change will induce great change of ecosystem and land-cover of the earth’s surface, so the forming and developing of land-cover in arid and semiarid region is greatly influenced by global change and it become a key site of global change and LUCC study. Characterized by synchronous broadly observation, speedy and economical, Remote sensing is widely used in the study of LUCC and desertification. In a word, the study of land-use/cover change and desertification by remote sensing has a great significance in arid and semiarid area.In this paper, based on 1:50,000 topographic map, field investigation data, Landsat TM image in 1987 and Landsat ETM+ image in 2002, meteorological data (from 1987 to 2002) of Zhongning county and Jingtai county, NOAA/AVHRR 8km×8km NDVI vegetation index, we processed and analyzed the data:First, we performed geometric correction, radiometric correction and atmospheric correction to the images, then validated the result of correction. To resolve the common phenomenon of same object having different spectral character and different object having same spectral character in study area, we extracted residential area and the grass in shelter belt based on knowledge discovery after supervised classification, distinguished salinized land, sandy desertification land and other harsh desert based on spectrum threshold, finally we got the accuracy assessment reports. We count the area change information of all land-cover type by transfer matrix and analyzed the spatial change of typical land-cover type. After the analysis of fractional vegetation cover, we integrated meteorological data (from 1987 to 2002) of Zhongning county and Jingtai county, NOAA/AVHRR 8kmx8km NDVI vegetation index (from 1987 to 2001) for the primary analysis of precipitation, natural vegetation and ecological environment change at the Southeastern Marginal Area of the Tengle Desert.The conclusion of this paper is as folio wings:?Based on the contrast analysis of multi-temporal remotely sensed images or quantitative study of remotely sensed images, radiometric and atmospheric correction of the images are very important. In this paper, we contrast the remote sensing images before correction with that after radiometric atmospheric correction, the result show that radiometric correction can greatly reduce the difference of different data.?The traditional auto-classification based on the spectral character always has poor accuracy because of the common phenomenon of same object having different spectral character and different object having same spectral character especially in arid and semiarid regions. Experiment results show that the classification based on multi-features has great accuracy compared with traditional auto-classification based on the spectral character.?The land-use/cover at the Southeastern Marginal Area of the Tengle Desert has taken place great change in the 15 years. The area of farmland increased 67 percent;residential area increased 149.5 percent;a great deal of cultivated land and woodland has been destroyed with the rapid development of industrialization and urbanization. There was much precipitation in 2002 compared with the precipitation in 1987, it made desert shrub got much better, much of salinized land and sandy desertification land turn into other harsh desert which has a better fractional vegetation cover. The grass and shrub in shelter belt along Baotou-Lanzhou desert-railway got little better at different extent, the protection work was efficient.? Through the analysis of NOAA/AVHRR NDVI data (from 1987 to 2001) and the precipitation data in the same period we found that grass and desert shrub was obviously restricted by precipitation. There is a lag between the growth of desert vegetation and the precipitation, furthermore, the correlation between NDVI vegetation index and precipitation two months before is much better than that between NDVI vegetation index and annual precipitation.?In the fifteen years focused on, the desert vegetation and precipitation in Jun and July had a little fluctuation increase from 1987 to middle of 1990s, but desert vegetation degenerated acutely in droughty 1991. From middle of 1990s, desertvegetation started to degenerate because of precipitation reducing, after then the ecological environment started to deteriorate, NDVI vegetation index and precipitation got the minimum in 2001. In 2002 the desert vegetation had an obvious recovery resulted from the precipitation increasing and grazing forbidden policy.?Result from much precipitation in 2002 compared with that in 1987 and grazing forbidden policy, the desert shrub got much better, much of salinized land and sandy desertification land converted into other harsh desert which has a better fractional vegetation cover, the ecological environment ameliorates. However, the grass and shrub in shelter belt in northwest margin of the oasis, both sides along the Yingshuiqiao-Yangulatu desert road, the farmland margin in the Nanshantaizi, sand desertification is still smart.更多还原