【作者】 张华；

【导师】 曾光明； 黄国和；

【作者基本信息】 湖南大学 ， 环境工程， 2004， 硕士

【摘要】 由于环境问题本质上都不同程度地包含有某种地理意义，引入遥感和GIS这两种信息时代的地学前沿方法，符合环境问题的本性，同时使研究者能够从空间和光谱的角度更加完善地考虑问题。在系统介绍遥感和GIS的基本概念、原理及主要应用的基础上，首先探讨了运用遥感技术对内陆河流的水质进行的定量监测。以湘江长沙段为研究区域，通过相关性分析成功建立了多时相遥感信息模型。然后阐述了遥感和GIS在土地利用／覆被研究中的应用。选取长沙市为研究区域，精确解译多时相的Landsat影像，采用重心迁移、转化矩阵、动态指数等方法，对长沙市土地利用／覆被格局的变化进行了定位、定性、定量的研究。利用KT变换对ETM影像进行光谱增强，综合监督方式和非监督方式获取了分类所需的混合模板。各地类的混合模板在特征空间中体现出了良好的可分离性，获得了较好的分类结果，并从光谱特征和空间分布的角度对分类错误进行了分析。在以上分类研究的基础上，引入分层二阶随机抽样方法改善专题图的精度评价效果。通过建立包含多层次单元的抽样体系和进行二阶随机抽取，获得了空间分布均匀的样本集，并有效降低了成本。所建立的概率型误差矩阵有效反映了抽样过程中所形成的中选概率的差异，所生成的用户精度、制图精度、全局精度和Kappa系数比传统方法所得出的结果具有更高的可信度。最后通过总结在一系列环境规划项目的实践，进一步说明遥感和GIS在环境监测与分析中具有显著的应用价值。更多还原

【Abstract】 Due the geographic essence of most environmental problems, introducing the two most informational geographic techniques, remote sensing and geographic information system (GIS), into the environmental research is compatible with its essence and will improve it both in spatial analysis and spectral analysis. After introducing the principles of remote sensing and GIS as well as their basic applications, remote sensing for quantitative monitoring of inland water quality is presented. By a series of pertinence analysis, a multi-temporal, universal remote sensing information model is successfully established for the Changsha Section of the Xiang River. Then, remote sensing and GIS are employed in the research of land use/cover change. Corresponding Landsat imageries are interpreted accurately. By using transformation matrix and change indexes, the characteristic of this change is analyzed, as well as the change of Bary center. Furthermore, spectral enhancement via the Kauth-Thomas transformation is applied to improve the classification of Landsat ETM image. Classification signers are obtained by a combination of supervised method and unsupervised method, and show good separability in the feature space. Sound result of classification is obtained, and the spectroscopic feature and spatial distribution of classification error is analyzed. Based on the previous research of classification, stratified two-stage sampling is introduced into the accuracy assessment of thematic map. The sampling design employs a nested, hierarchical partition constructed of spatial units of different size, which provides a sound spatial distribution of specimen and reduces the sampling cost. Different inclusion probability is assigned to different specimen, and a probabilistic confusion matrix is established based on this. Producer’s accuracy, user’s accuracy, and overall accuracy are analyzed as well as Kappa coefficient and their improvement is analyzed. At last, some experience in related environmental planning is summarized, which shows the great potential of remote sensing and GIS in the environmental monitoring and analysis, additionally.更多还原