【作者】 杨凯；

【导师】 曾永年；

【作者基本信息】 中南大学 ， 地图学与地理信息系统， 2008， 硕士

【摘要】 城市区域土地利用/覆盖的快速变化,改变了地域土地覆盖的空间格局,从而对城市及其周边地区的自然、生态和社会经济过程产生了巨大影响。深入开展城市土地利用/覆盖变化及其空间格局分析,对城市可持续发展及保持良好的城市生态环境具有重要的意义。本文以长沙市地区为例,就城市土地利用/覆盖分类方法以及城市空间格局进行分析研究。主要研究内容及结论如下:(1)基于地表生物物理参数的土地利用/覆盖遥感分类方法研究针对基于地物波谱反射特征的土地利用/覆盖分类方法,难以很好地解决“同物异谱、异物同谱”问题,分类精度不高的状况。本文充分利用TM数据的多光谱特征,提取了植被指数NDVI、地表温度Ts、温度植被角度TVA和温度植被距离TVD四种分类特征进行最大似然分类和决策树分类,通过对不同的组合方案的分类结果比较,发现NDVI和Ts作为分类特征参与到多光谱影像分类中,无论采用最大似然分类还是决策树分类,分类结果比较理想;将TVA、TVA和TVD分类特征与多光谱影像相结合进行分类不但没有提高反而降低了分类精度;加入NDVI、Ts以及TVD特征信息参与最大似然分类得到的分类精度却有所提高;决策树分类中,在Ts、TVA、TVD波段,各类地物的光谱值范围都有一定程度的混淆。因此,在多光谱影像中单独加入Ts、TVA或TVD特征波段,不能有效区分地物。为此,分别将NDVI、TVA和TVD作为特征波段与多光谱影像相结合参与决策树分类,其精度比利用多光谱波段进行最大似然分类有了很大的提高。(2)城市空间格局梯度分析研究基于多时期的Landsat影像,提取了1973年、1993年、1998年和2001年四个时期的土地利用/覆盖信息。在此基础上,采用“梯度分析”和“景观格局指数”相结合的方法分析长沙市城市化过程中空间格局的变化。结果表明:①对于长沙市来说,半径大小为“1km”的移动窗口是研究长沙市景观空间格局梯度分析较适宜的尺度,既能避免景观指数的较大波动,又能全面地体现城市景观梯度变化的规律;②随着城市化程度的推进,在类型水平上,从城市近郊到边缘,各种土地覆盖类型随着距离城市中心的远近,各种景观指数则表现出明显的梯度变化。在景观水平上,城市中心区域主要以城镇用地为主,故景观的破碎化程度低,斑块较为完整,形状较为简单,连通性好;城市近郊是城镇用地、裸地、耕地和林地的交界地带,破碎化程度较高,形状复杂,斑块连通性差;城市远郊是长沙市景观格局变化幅度最大的区域,原来的自然景观状态被打乱,林地和耕地被其他土地利用类型所蚕食,特别是城镇用地,导致景观斑块的形状变得复杂,破碎化程度也比以前严重,景观的优势度降低。(3)城市化水平与土地利用/覆盖景观格局的关系研究基于多时期的Landsat影像,提取了多时相长沙市五区的城镇用地。同时在景观水平上定义了城镇用地比例来表达城市化水平,并分别从密度指标、斑块面积指标、聚散性指标、多样性指标和形状指标中选取了六种景观指数,以便讨论城市化过程与土地利用/覆盖景观格局变化的关系。城镇用地比例与这六种景观指数的回归分析表明城市化水平与这些指数之间有明显的相关性。更多还原

【Abstract】 Urban land use/cover changes have impacts on urban surface charaicteristics, as well as the socio-economic developmemt. Therefore, it is important to analyze the urban land use/cover and spatial pattern changes. Taking Changsha city as a case study area, this paper studied the classification method of unban land use/cover and the analyzed landscape pattern gradient changes. The main conclusions are as follows:(1) Land Use/Cover Classification Using Remotely Sensed Surface Biophysical ParametersThe classification based on the spectrum reflectivity can not solve the phenomena of "the same kinds of targets with different spectral or the different kinds of targets with the same spectral". Therefore, the classification accuracy is not satisfied to analyze urban land use/cover change. This paper extracts four classification features including Vegetation Index (NDVI), land surface temperature (Ts), Temperature-Vegetation Angel (TVA) and Temperature-Vegetation Distance (TVD) for Maximun Likehood classification and decision tree classificaion. Compared several image processing routines, the results indicate that it has the highest classification accuracy combining NDVI and Ts and the multi- spectrum image for Maximun Likehood classification and decision tree classificaion. TVA、TVA and TVD that are used in the multi- spectrum image almost had no effect for improve the classification accuracy. Whereas, based on the multi-spectrum image, it has higher classification accuracy using NDVI, Ts, and TVD. In the process of decision tree classification, the spectrum of the types is mixed up in the bands of Ts、TVA、and TVD, and it can not distinguish the land use/cover types. Therefore, NDVI、TVA and TVD are used to joined in the classification. Its accurarcies are highter than the classification accuracy that only using Maximun Likehood classification.(2) The gradient analysis of urban landscape patternFirstly, land use/cover information for the four years is extracted based on the Landsat imagery. Then, landscape metrics combined with gradient anlysisis employed to analyze the spatial pattern changes of Changsha city. The results show that①the mobile window that the radius is 0.5km may be more appropriate for unban landscapes pattern analysis. It avoids the great fluctuation of landscape metrics, and it can reflect the change rule of urban landscape gradient;②with the development of the urbanization process, In the class level, from the the center of the city to the edge, various landscape metrics reveral significant gradient variation according to the distance between the center of the city and the types of land use.In the landscape leve, low fragmentation and good connectedness is appeared in the center of the city. Beacause city, bare land, argriculture and forest exist in the outskirts, the landscape has high fragmentation, poor connectedness; the exurb is the area that its landscape pattern changes most. The nature landscape is disturbed, and its shape becomes more complex. Moreover, the high fragmentation also is appeared and the dominance of the landscape failed.(3) The relationship between the urbanization and the landcape patternLandscape patterns in four different years (1973, 1993, 1998 and 2001 year) are extracted using Landsat images. At the same time, the urbanization level is expressed as the city area proportion and six landscape metrics are chosed from density metrrcs, class area metrics, contagion metrics, diversity metrics and shape metrics in order to discuss the relationship between the urbanization level and the change of land cover landcape pattern. The regression analysis shows that there are significant correlations between the urbanization and six metrics.更多还原