【作者】 于新洋；

【导师】 张安定；

【作者基本信息】 鲁东大学 ， 自然地理学， 2012， 硕士

【副题名】以烟台市区为例

【摘要】 城市作为人口集聚的增长极和经济发展的动力源，历来是学者研究的重点区域。虽然目前城市面积仅占全球陆地面积的很小部分，却集中了人类半数以上的人口。城市区域的快速扩展会对社会-经济-环境生态系统的可持续发展产生难以估量的影响，这些影响理应引起人们的警示和思考，并采取必要措施减少城市发展引起的消极影响，以保证人类社会的可持续发展。非渗透表面，又被称为非透水表面，是指水分无法透过其深入土壤的人造表层，如道路、停车场、建筑物屋顶等，它是城市和村镇地表的重要组成部分。非渗透表面不仅是城市化程度的重要表征指标，而且是城镇环境质量的表征指标之一。准确提取和监测非渗透表面密度信息以及建立非渗透表面特征数据库对全球变化和人地交互关系的研究具有重要意义。自获取首幅陆地遥感影像以来，遥感影像以其多波段、信息连续和获取方便等优点，逐渐成为非渗透表面密度信息提取的主要数据源，各种提取方法和理论模型也应运而生。目前，基于遥感影像的城市非渗透表面信息提取方法可归纳为像元尺度的非渗透表面信息提取法和亚像元尺度非渗透表面信息提取法。前者主要应用传统分类方法研究和提取城市非渗透表面信息，后者则基于像元分解技术，按密度大小定义并研究城市非渗透表面信息。本研究以烟台市区为研究区域，以1989、1995、1999、2000、2006和2009年获取的Landsat TM/ETM+遥感影像数据为主要数据源，以研究区中巴地球资源卫星影像、谷歌地球影像、土地利用图以及行政区域矢量图等为辅助资料，经过几何校正、辐射校正等影像预处理手段，利用植被-非渗透表面-土壤模型和归一化线性光谱混合模型对研究区进行非渗透表面密度信息提取，研究其空间变化模式，并从两个方面展开对非渗透表面密度的应用研究：以多时相非渗透表面密度信息为决策判断条件，利用决策树分类模型对研究区进行土地覆被分类，与传统分类方法进行精度对比，并对人类活动影响的城市用地的动态变化模式加以研究，以求更为准确的监测研究区的城市用地变化情况；与此同时，对研究区城市非渗透表面密度/归一化差值植被指数与地表温度的相关关系进行初步的探索，为非渗透表面密度和归一化差值植被指数作为地表温度的表征指数的可靠性提供例证。研究结果表明：（1）将植被-非渗透表面-土壤模型与归一化光谱线性分解模型结合起来提取烟台市区城市非渗透表面密度的准确度较高，得到的非渗透表面密度信息可以作为后续研究的基础数据源；（2）对2009年烟台市非渗透表面密度进行密度分级后发现烟台市的城市低密度区和中密度区仍占据较大区域范围，高密度城市区域分布范围较小，烟台市区总体上仍处于低密度分散发展阶段；（3）土地覆被分类结果显示融入非渗透表面密度信息的决策树模型的土地覆被分类精度高于作为对比的最大似然法和人工神经网络分类法的分类精度。对研究区进行城市用地变化监测发现自1989年至2009年，由于人类活动影响的不断深入，烟台市区城市用地面积扩展了约35.83km2，主要转自农用地、裸地和植被覆被区域；（4）非渗透表面密度与地表温度呈显著的线性正相关关系，非渗透表面密度可以作为地表温度的表征指数；归一化差值植被指数受季节影响较为显著，与地表温度的线性关系不明显，无法作为地表温度的表征指数。对于本研究区而言，利用非渗透表面密度作为地表温度的表征指数较为适宜。更多还原

【Abstract】 Urban area has been subjected to the attention of researchers since it is the mostpowerful source of economic development. Although now urban area covers a smallfraction of the earth land area, it has had profound impact on the development of riverbasin and even the global environment. High-speed urban expansion has producedincalculable damage to the sustainable development of the environmental andsocial-economic-ecological systems, and these repercussions should be directed moreattention by decision makers to make informed decisions which could balance thepositive and negative aspects of urban sprawl to ensure sustainable development.Impervious surface refers to the artificial surface through which water cannotpenetrate into the soil, such as roads, parking lots, buildings, roofs, etc. In recent years,impervious surface became an indicator of the urbanization degree; at the same time,impervious surface is also an important indicator of urban environmental quality. It isof significance to extract and monitor impervious surface area for environmentalmonitoring and urban planning.The extraction methods of urban impervious surface based on remotely sensedimages can be divided into per-pixel classification and sub-pixel classification.Per-pixel classification utilizes traditional methods such as maximum likelihoodmethod to monitor and extract the urban impervious surface area information inremote sensing images; sub-pixel classification algorithms decompose pixels,defining the impervious surface information depending on the density of impervioussurface.Jiaodong Peninsula is the largest peninsula and is one of the most developedeconomical areas in China. Yantai is one of the most representative mid-sized coastalcities in China with150years history since it opened the harbor for internationalbusiness in May1861. Now it is the largest fishing seaport and the second largestindustrial city in Shandong province. With the help of multitemporal remote sensingimages, it is of great practical significance to take Yantai city as the case study area toextract and monitor Yantai city’s impervious surface information. However, until nowthere is no research studying the impervious surface information extraction in Yantaicity. In this study, Yantai City, Jiaodong peninsula, was chosen as the case study area,the Landsat TM/ETM+remotely sensed images of1989,1995,1999,2000,2006and2009were selected as the main data sources, and China-Brazil Earth ResourcesSatellite images, digital elevation model data, land use maps and vector administrativedata were chosen as the auxiliary data. Image resolution and unity, imageenhancement, registration, geometric correction, minimum noise fraction, pixel purityindex transform were performed to the images. After that, the vegetation-impervioussurface area-soil (V-I-S) model and normalized spectral mixture analysis (NSMAS)model were used to extract the impervious surface area density component diagram tostudy the urban developing dynamics of Yantai city. Then the percent impervioussurface area was introduced into the decision tree classification method as thedecision rules to classify the land covers of the case study area in order to improve theaccuracy of monitoring the land cover change in Yantai city; and it was also used toexplore the relationships of percent impervious surface area and surface temperaturepreliminarily.The results showed that:(1) It is of high accuracy to combine vegetation-impermeable surface area-soil modeland the normalized spectral mixture analysis model for the extraction of theimpervious surface density of Yantai city, and the density information can be used asthe foundation data for the further researches;(2) The low-density and medium density area of Yantai city still cover a large urbanareas, generally Yantai city is still in the low-density development stage.(3) The accuracies of decision tree classification methods taking the three phases ofthe Yantai urban impervious surface density as decision rules were highest among thethree classification methods (Impervious surface area-decision tree method, maximumlikelihood method and artificial neural network classification method as comparison).Statistics of the classification results found that from1989to2009, the extendingimpervious area of Yantai City accounted to approximately35.83km2, and wastransferred from agricultural land, bare land and vegetation land cover areas;(4) The study of the relationships between percent impervious surface and landsurface temperature, normalized difference vegetation index and land surface temperature extracting from the four different seasons of1995,2000,2006and2009.The results demonstrated that there was a strong linear relationship between landsurface temperature and percent impervious surface area for the four seasons, whilethe linear relationship between land surface temperature and normalized differencevegetation index is much less strong and is seasonal effected. The results suggest thatpercent impervious surface area provides an alternative parameter to the typicalnormalized difference vegetation index for analyzing land surface temperaturequantitatively in urbanized environments.更多还原