【作者】 韩文萍；

【导师】 王金亮；

【作者基本信息】 云南师范大学 ， 自然地理， 2007， 硕士

【摘要】 区域和地方尺度上的典型区土地利用/土地覆被变化(LUCC)研究,不仅可以深入地探讨土地利用变化的有关科学问题,同时也为土地利用变化的综合分析乃至全球环境变化研究提供丰富、准确的信息。区域土地利用/土地覆盖变化信息提取是一个多因素、多环节交织在一起的复杂过程,所要提取的变化要素不仅包括发生变化的空间位置和范围,还包括变化的类型和面积等信息,需要处理大量数据。因此,如何提高LUCC信息提取的效率和质量,是目前亟待解决的问题,对于土地资源的利用现状及其变化动态的快速调查和更新有极其重要的直接意义。在借鉴前人研究成果与经验的基础上,认为LUCC信息提取可分解为:变化信息的发现、变化范围的提取和变化类型转化关系的确定三个环节。本文以中分辨率遥感数据TM、ETM+为主要数据源,充分挖掘多源辅助数据信息,发展基于决策树的LUCC信息提取方法,初步实现了高原山地地区的LUCC变化信息自动发现-提取以及类型表示的自动化流程。主要研究内容和结论如下:1.完成研究区1989年、1999年四景影像的几何配准、辐射校正以及切割拼接,以保证变化信息提取的有效性。提取了研究区NDVI特征影像以及利用研究区DEM提取了高程和坡度信息,用于决策树规则建立;2.发展了基于中分辨率遥感数据的变化信息自动发现及变化范围提取方法;3.利用特征数据以及专家知识,构造试验区的决策树分类模型;4.构造了变化信息提取的决策树模型,最终确定研究区各地类转化关系,不同的变化类型提取后用不同的颜色表示出来。得到试验区1989年到1999年基于遥感影像的土地利用/土地覆盖变化动态监测图,并统计获得变化类型的面积信息,用转移矩阵表示。精度评价表明,分类精度和变化信息提取精度都较理想,能满足应用需求。结果表明,所采用的方法和技术提高了LUCC信息提取的自动化水平和精度。决策树能融合多来源、多类型数据,融光谱知识、专家知识、地学分析于决策规则中,来提高变化信息提取、变化转化关系确定的精度和自动化程度,是目前LUCC信息提取方法中的新方法,可以在更大范围内推广使用。更多还原

【Abstract】 Regional and local scales typical of land use/land cover change (LUCC), not only can in-depth discussion of land-use changes in the relevant scientific issues, as well as land-use changes and the comprehensive analysis of global environmental change research provide rich, accurate information. Regional land use/land cover changes in information extraction is a multi-factor, multi-link intertwined in a complex process, to be extracted elements of the changes include not only changes in the spatial location and scope also include changes in the type and size of such information is required to handle large amounts of data. Therefore, how to improve the LUCC information extraction efficiency and the quality is a serious problem, As for land resources utilization and the rapid changes of the investigation and is very important to update the direct significance.Draw on previous research results and experiences that the LUCC information extraction can be divided into: changes in the discovery of information, changes in the scope of the types of changes in extraction and conversion of the three sectors identified. In this paper, medium-resolution remote sensing data TM and ETM+ data as the main source of tap more sources of information supporting data, LUCC development of the decision tree method of extracting information, realize the plateaus of LUCC change information automatic discovery-type extraction and said the automation process. Main content and conclusions are as follows: 1. Completion of the study area in 1989, four in 1999 King of the geometric image registration, Radiation cutting and splicing correction to ensure that changes in the effectiveness of information retrieval. Extraction of the characteristics of the study area NDVI images and the use of the study area from the DEM elevation and slope information, decision tree for establishing rules;2. Based on the development of medium-resolution remote sensing data change information automatically changes the scope of discovery and extraction methods;3. Base of data and expertise, the establishment of experimental zones decision tree classification model;4. Changes in the structure of information extraction decision tree model, and the final study area into all types of relationships, changes in the different types with different colors said. District will be tested in 1989 to 1999 based on remote sensing images of land use/land cover change detection maps, Statistics obtained and the types of changes in the area of information, with the transfer matrix, accuracy evaluation, classification and the types of changes in the accuracy of more than 80%, can satisfy application needs. The results showed that the method used to improve the technology and information extraction LUCC level of automation and accuracy.Decision Tree integration of multi-source, multi-type data, spectral knowledge, expertise, to knowledge through changes in the rules for information extraction, transformation changes identified, LUCC is information extraction methods of the new methods to improve the information extraction LUCC level of automation and accuracy, in the greater context of use.更多还原