【作者】 白卫国；

【导师】 陆守一； 李增元；

【作者基本信息】 北京林业大学 ， 森林经理学， 2004， 博士

【摘要】 目前，国内对草地生态系统区变化的监测研究多是以牧草生物量为对象进行的，而没有对各种土地利用方式综合考虑。变化研究没有对草地生态系统区各种土地利用的空间变化方式分析及驱动因素分析；没有多尺度综合研究。建立的众多草地监测模型只能在特定的区域、特定的时间段内有效，而缺乏大范围内草地生态系统区土地利用变化的监测模型。本研究以高寒草地生态系统区的土地利用变化情况为对象，借助遥感和GIS技术，探索高寒草地生态系统区土地利用空间变化的分析方法，并在选择的试验区进行了方法试验，结果如下： (1) 在对研究区域土地利用特点分析的基础上，结合本研究的目的，对所获得的两期遥感影像数据，采用ISODATA算法进行图像分割。采用判读解译、NDVI分析和地物波谱特征分析相结合的方法对遥感影像进行解译，将区域土地利用类型划分为高盖度草地、中盖度草地、低盖度草地、农田、弃耕地或休耕地、水体和裸地。居民点、道路和河流采用已有的数字化地图。对分类后的结果进行了检验，精度达到90.91％。 (2) 根据目前土地利用变化研究中的实际需求，提出并实践一种新的土地利用变化监测模型，并以通用商业遥感软件ERDAS为支撑平台，对该模型进行了实现。应用该模型对试验区1987年～2002年的土地利用变化进行试验，验证结果表明，该模型对土地利用变化监测的精度只受土地利用分类精度的影响，模型对分类专题图的处理结果准确度可达100％。该模型的优点在于简单、易于理解和应用，适宜推广应用。 (3) 对土地利用变化研究中应用的两种重心模型进行了实例检验和对比分析，得出重心模型二优于重心模型一的结论。针对地学研究中的三维空间特点，结合高寒草地生态系统区土地利用的特征及本研究的目的，对重心模型二进行了改进，将其由2维模型扩展为3维模型。 (4) 对研究区域土地利用的变化分析得出：1987年～2002年，高盖度草地类型面积减少了90634hm~2，中盖度草地类型面积减少了124600hm~2，低盖度草地类型面积增加了157539hm~2，裸地类型面积增加了21675hm~2，现耕地类型、弃耕地或休耕地类型面积分别增加18840hm~2和18409hm~2。 (5) 应用改进的重心模型对土地利用类型的分布进行了分析，结果表明，高盖度草地类型的分布以山地区比较集中，中盖度草地类型主要分布于山区与平坦地形的交错地带，低盖度草地类型、现耕地类型、弃耕地或休耕地类型、裸地类型和水体类型主要分布在平坦地形区；对土地利用类型在空间上的变化分析结果表明，2002年与1987年相比较，低盖度草地的分布向东南移动了28732m；裸地类型的分布向东南移动15671m，其它类型的分布均发生了明显的移动；对土地利用类型的分布高程变化研究结果表明，中盖度草地类型的分布重心海拔上升241 .27m，高盖度草地类型的分布重心海拔上升182.48m，裸地类型的分布重心海拔下降173.48m，低盖度草地类型的分布重心海拔下降127.74m，弃耕地或休耕地类型的分布重心海拔上升42.78m，现耕地类型的分布重心海拔上升36.45m。 (6)对山地区和平坦地形区1987年一2002年的土地利用类型分布及变化的空间特征进行了较为详细的对比分析。 (7)采用选择窗口法，应用景观指数对1987年和2002年土地利用的景观特征分别在类型水平和景观水平进行了分析。 (8)采用与地图制图比例尺相联系的景观尺度确定方法，分别在比例尺l:5万、l:10万、l:20万、l:25万、1:50万和1:100万的尺度上，对1987年和2002年各个选择窗口内的景观表现特征、变化进行了分析和比较。研究认为，对高寒草地生态系统区土地利用的景观特征及变化分析，采用以1:25万比例尺的空间数据为基础进行分析比较合理。更多还原

【Abstract】 The resent survey research about grassland ecosystem placed more attentions on grass biomass productivity on the grassland, and take no integrative consideration about various types of land use. Studies about grassland ecosystem change, lacking of the analysis of spatial trend and drivering force,were also not be operated in a multi-scale integrative analysis. Many models that had been built for grassland survey can only work correctly in particular condtion and in limited period. There is still no survey model that can be used in large area extent to inspect land-use changes in gralssland ecosystem.In this dissertation, the changes of land use in the Alpine Grassland Ecosystem (AGES) were analyzed by using the technologies of Remote Sensing (RS) and Geographical Information System (GIS), and a new set of methods suitting to analysis spatial change in the AGES were worked out and proved effectively in experiment area. With large quantity of analysis, the following results were mainly reached.(1) With the analysis of land use characteristics of experiment region in Alpine grassland ecosystem, and taking consideration of the aims research, the two gainned images, Landsat TM image gainned in 1987 and Landsat ETM+ image gainned in 2002, were segmented by used the arithmetic of Iterative-Orgnizing Data Analysize Technique (ISODATA). Then these fragments were identified by using land-use indicators, which were built during the image interpreting work prcocess in field, such as, the color, texture, shape, position, etc.,and also by using the methods of NDVI and characteristic analysis of spectrum of land-use types in a man-compute interactive manner. The final land-use thematic maps were achieved in the images, and also the land-use types of the High Coverage Grassland (HCG), the Middle Coverage Grassland (MCG), the Low Coverage Grassland (LCG), Field (F), abandoned/rested Field (F1), Bareness land (B), and Water body (W) were achieved. The distribution condition of resident sites, roads and rivers were got from digited maps. Precision was assessed on the final land-use classified results, and a precision of 90.91% was gainned.(2) For the purpose of meeting the need in land-use change research, a new land-use change survey model were built and carried out to use on the basis of commom business Remote Sensing software ERDAS in this study. Finally, this model was used to inspect the land use changes in the experiment area from 1987 to 2002. Outcome showed that the precision of land use change inspection was only influenced by the precision of land-use classification, and for the land-use thematic maps, the precision of land-use type change detection can reach 100%. It is simple to beunderstood and easy to use, suitable to generalization and use in LUCC research field.(3) Two types of gravity center model were detected and compared by using the land-use thematic maps of the experiment area in this study. It was concluded that model 2 is super to model 1. Considerring the 3-dimension characteristics of objects studied in geo-science field, the gravity center model 2 was advanced from 2-dimension to 3-dimension.(4) By analyzing the area of different land-use types in the experiment region from 1987 to 2002, it was found that the area of High Coverage Grassland reduced 90634hm2, the area of Middle Coverage Grassland reduced 124600hm2, the area of Low Coverage Grassland increased 157539hm2, the area of Bareness Land increased 21675hm2, the area of Field increased 18840hm2, and the area of abandoned/rested Field increased 18409hm2.(5) Analyzing the spatial distribution and changes of different types of land use in experiment with the advanced gravity center model, the results showed the distribution of different types of land use , that the High Coverage Grassland mainly distributed in mountainous area, the Middle Coverage Grassland mainly distributed between mountainous sub-region and flat sub-region in the experiment area, the Low Coverage Grassland, the Field, the abandoned/rested Field, the Bareness land,更多还原