【作者】 杨德生；

【导师】 何政伟；

【作者基本信息】 成都理工大学 ， 地球探测与信息技术， 2011， 博士

【摘要】 重庆市渝北区作为重庆主城和新型工业化发展的主要拓展区,建区以来在加快建设重庆都市圈,大力发展城郊型经济发展中发挥了重要作用,但同时也承受了高强度的人为干扰过程,使地表景观及生态环境状况均发生较大变化,对生态环境保护工作提出了更高的要求。另外,作为中国内陆唯一的国家级开发开放新区“两江新区”的主体组成部分,未来一段时期,其必定面临着高强度开发的问题,人类发展与生态环境的矛盾将表现得尤为明显,生态环境保护及经济与环境的协调发展急需先进、科学的技术手段与方法支持。在地表覆被信息提取过程中,首先对遥感数据源的光谱和纹理特征进行分析。在分析不同类型的地表覆盖在TM/ETM遥感影像中的光谱特征的基础上,以样本均值光谱曲线及盒须图为工具,对各地类光谱的差异性及可分性进行归类;通过绘制纹理特征光谱曲线,确定了适用于特定覆盖类型分类的TM影像纹理特征。结合地质、地貌、土壤、DEM数据以及对目标地物的影像光谱、纹理认识,采用CART决策树构建基于决策规则的、适合多源数据的、复杂地区地表景观类型精确识别与提取的技术体系。在完成研究区地表覆盖状况准确提取的基础上,通过斑块类型水平的TA、PLAND、NP、PD、LPI、LSI,景观水平NP、PD、LSI、IJI、AI、CONTAG共12个景观指数,进行各地表覆盖类型动态变化的分析,挖掘区域景观发展规律;针对区域的城市建成区、道路、矿山等典型人工景观进行了扩展分析,揭露其发展特点与规律,为区域规划提供科学依据。结合区内气候、环境、经济社会数据开展研究区景观格局变化的生态环境响应研究,重点分析了城市化进程的生态环境影响。建立了符合研究区实际的区域生态环境质量评价指标体系,基于综合指数法开展区域生态环境质量分区评价,确定了生态环境优质区、良好区、正常区、退化区的空间分布状况,为生态环境保护工作提供了依据。取得以下主要结论:(1)兼容影像特征、DEM、地貌、地质、土壤多源数据的遥感分类方法适用于地表景观破碎的丘陵区土地覆被信息提取,经实地验证分类精度较高,能满足研究需求,并完成了研究区1988年、2000年、2007年、2010年四期土地利用(建设用地、水体、草地、林地、耕地)信息的提取,精度分别达到71.89%,85.87%,78.8%和86.96%。(2)应用景观生态学的理论与方法对区域建设用地、耕地、水体、草地、林地五类景观在1988-2010年的变化进行分析,结果表明:水体景观面积不断减少并且破碎化严重;建设用地景观面积一直不断增加,且呈现出高度集中的趋势;林地景观在2000年前受到较大破坏,但之后面积大幅增加而斑块数量减少;草地景观变化与林地景观的规律正好相反,因此可以判断林地景观的变化应该或至少是造成草地景观变化的重要因素之一;耕地是区域景观中的主体类型,变化不明显。在1988-2007年渝北建成区发展主要是以片状蔓延式和轴向延展式为主要扩展模式,但2007年至2010年期间,则主要是以跳跃式扩展和轴向延展式为主;渝北区道路在2000年至2010年期间,得到了大力的扩展建设,道路密度明显增加,总体与城市建成区扩展模式相同;矿山开发对区内的生态系统有重要影响,从2000年到2010年间区内矿山数量明显增多,开发强度加大。(3)结合城市化进程分析区域景观格局变化的生态环境响应,可以看出区域气候、空气质量、水质、区域生态系统服务价值均发生不同程度变化,这和区域的城市人工景观增多、城市建设工程规模扩大、人口大幅度增加、温室气体排放加剧有密切关系。通过生态环境质量的综合评价,划分出生态环境优质区、良好区、正常区、退化区四个分区,面积分别为:326.19km2,682.25km2,340.37km2,108.14km~2。最后根据渝北地理特征、自然条件、资源禀赋、产业基础及发展现状,提出了区域生态建设与保护的合理对策。更多还原

【Abstract】 As the major area of Chongqing city and new industrial development, YubeiDistrict has contributed significantly to Chongqing’s urbanization and suburbeconomic development; meanwhile, its landscape structure has gone through greatchanges under the high-intensity human disturbance, therefore the environmentprotection presents a great challenge and requires higher standards. Furthermore,being in the heart of "Liangjiang New Area", the only national development zone inmiddle-west of China ,Yubei will be face the challenging problems coming along withhigh intensity development during the next period. The conflict between humandevelopment and the environment will be inevitable, how to protect ecologicalenvironment and balance economic and environmental development need the supportof advanced technical means and scientific methods.In the process of extracting land cover data, firstly, the spectral signature andtexture feature of different land use type based on TM/ETM data are analyzed . thedifference of ground objects spectrum are classified using mean spectral curve andbox plot through sample analysing. then, proper TM spectral signatures for specificland cover type are identified. texture feature spectral curve also be confirmed forinformation extracting.By combining the image spectral signatures of geological,terrain, soil, DEM and land cover data, CART decision tree will be introduced toremote sensing image classification to formulate a decision-based, multisource-basedlandscape identification and extraction system.Based on accurate extraction of land cover of study area, different land coverdynamic change are analyzed through patch Class-level index(TA、PLAND、NP、PD、LPI、LSI)，Landscape-level index(NP、PD、LSI、IJI、AI、CONTAG )to mine outthe trend of regional landscape development. Extensive analysis is also carried out onother man-made landscape, such as, city built-up area, road, mine, to providescientific evidence for regional planning. Using climate, environmental and social-economic data, the study of ecological response to landscape change in studyarea are carried out, putting an emphasis in the effect of urbanization oneco-enviornment. This study also proposed a feasible regional ecological environmentquality evaluation index system ,with compositive index method to classify theeco-enviornment quality of the study area into four categories: high quality area, goodarea, normal area and degraded area, and thus provided an alternative mean toeco-enviornment protection.Major conclusions as followed:Classification accuracy of land cover information extraction at the hilly area withmulti-source data has been proven to be effective in the research. Based on TMimages texture features and spectral analysis, combined with DEM, topography,geology and other data, CART decision tree are introduced to remote sensing imageclassification, finished the 1988, 2000, 2007, 2010, four periods of land-use(construction land, water, grassland, woodland, arable land) information extraction.Using the error matrix to assess accuracy, the classification accuracy obtained is asfollows: 71.89%, 85.87%, 78.8% and 86.96%, indicating that the classification iseffective.By applying landscape ecology theory and methodology to analyze the change ofbuilt-up area, arable land, water, grassland, woodland landscape between 1988 and2010 year, the results showed that: the decrease and fragmentation of water landscapearea is serious from 1988 to 2010; area of construction land has been increasing,showing a trend of concentration; forest landscape is subject to greater damage beforethe year 2000, but a substantial increase and plaque area decreased; grassland andwoodland change follows the contrary law, it can be to determine that the changes inforest landscapes are caused by turf or at least an important factor in landscape change;farmland is the main type of landscape, it has not changed significantly. Built-up areasof Yubei extended as sheet style and axial spread mode during 1988-2007, but theleapfrog extension and axial spread mode is the main type of growth from 2007 to2010; roads of YuBei has been greatly extended between 2000 and 2010, road densitysignificantly increased, which has the same pattern with the overall expansion ofurban built-up areas; mine development have a major impact on the local ecologicalsystem, the number of mine and development strength increased significantly from2000 to 2010.Combined urbanization process to analyze the ecological effects of landscapechange, we can see that the regional climate, air quality, water quality, the value of ecosystem services are changing to different extent, which are closely related withartificial landscapes increasing, the expanded scale of urban construction projects, asubstantial increase in population and increased greenhouse gas emissions. Throughthe comprehensive evaluation of local ecological environment quality, high qualityarea, good area, normal area and degraded area were classified, and a reasonableenvironmental recovery and protection measure was proposed. And finally, this paperalso provided reasonable countermeasures of ecological construction and protectionaccording to natural conditions, natural resources, industrial base and development ofthe present situation.更多还原