【作者】 刘璞；

【导师】 吴嘉平；

【作者基本信息】 浙江大学 ， 遥感与信息技术， 2008， 硕士

【摘要】 浙江钱塘江流域地处长江三角洲经济发达地区。近十几年来,人口、经济快速增长,土地利用显著变化,国民生产总值逐年递增,成为我国经济发展最快的地区之一。但由于在经济发展过程中对土水资源的合理开发利用和生态环境的保护重视不够,造成了土地利用方式的不合理以及土水资源的大规模无序开发,从而加剧了钱塘江流域水质下降和环境恶化。研究表明,流域内各种土地利用/覆盖类型比例的变化是造成土水质量下降和生态环境恶化的重要原因。伴随着钱塘江流域经济的飞速发展,科学大范围地进行土地资源科学有效管理迫在眉捷。因此,利用遥感技术进行土地利用/覆盖自动分类方法研究及遥感监测时空变化分析,正确认识浙江钱塘江流域土地利用/覆盖变化情况及时空演变趋势,对优化土地利用结构,实现土地资源的优化配置和集约利用将提供新的方法途径,对钱塘江流域乃至整个浙江省的可持续发展具有极其重要的意义。本文将光谱角制图(SAM)方法与多源信息相结合,探索了适应于地形地貌复杂、土地利用细碎地区,简便、高效的能够满足应用需求的高精度土地利用/覆盖计算机自动分类的新方法。并运用此方法,对浙江钱塘江流域1991年、1997年和2004年三年跨度近15年土地利用/覆盖进行自动分类,实现动态监测。随后,对1991年、1997年和2004年三期的土地利用/履盖变化进行时空特性分析,揭示了浙江钱塘江流域土地利用/覆盖变化的规律,寻找出该流域土地利用/覆盖变化的热点地类及热点区域,以及探明了各地类重心转移的程度和方向。为优化浙江钱塘江流域土地利用结构,实现土地资源的优化配置和科学管理,提供强有力的信息支持和研究分析的新路径方法。本研究的主要结果如下:1.基于SAM和多源信息的土地利用覆盖自动分类方法研究土地利用/履盖变化是全球环境变化的重要组成部分和主要原因之一。如何在地形地貌复杂、土地利用细碎地区,获得能够满足应用需求的高精度土地利用/覆盖自动分类专题图是遥感工作者所面临的挑战。为此,本研究将多源信息与光谱角制图(SAM)自动分类相结合,探求多源信息在TM图像自动分类中的作用。对引入的七个信息:海拔高程、坡度、坡向、归一化植被指数(NDVI)、归一化水体指数(NDWI)、归一化建筑指数(NDBD和归一化裸土指数(NDBaI)的作用,进行了分析。以具有代表性的浙江富阳市和义乌市为研究区,进行试验验证。结果表明,坡度信息的辅助作用最为突出。复合坡度信息后的TM图像SAM分类比单独TM图像SAM分类总精度提高10%左右。以坡度作为辅助信息与TM图像进行融合,再利用SAM方法进行分类,简便高效,对研究地区的土地利用解译工作具有实际的应用价值。2.浙江钱塘江流域1991年、1997年、2004年土地利用/覆盖动态监测结合分层分区分景的方法,根据研究区的地物特征和地形地貌,将研究区划为3层16区87个地貌分景区,运用SAM自动分类方法对浙江钱塘江流域1991年、1997年、2004年三期共87个地貌分景区进行土地利用/覆盖的自动分类,并对分类结果进行了随机点精度评价和实测点精度评价。两次评价结果,三期影像的分类总精度均在80%左右,总Kappa系数均在0.70以上,分类质量较好。浙江钱塘江流域1991年、1997年、2004年土地利用/覆盖动态监测结果:林地基本保持不变;水田呈逐年下降趋势;建筑用地呈逐年上升趋势;旱园地在1991年和1997年面积保持不变,在2004年有明显下降;水域在1991年和1997年面积变化不大,在2004年明显上升。3.浙江钱塘江流域1991年到2004年土地利用/覆盖时空变化分析浙江钱塘江流域1991年到2004年间,建筑用地变化速度最快,是浙江钱塘江流域土地利用/覆盖变化的“热点”地类;早园地和水田变化频繁,是浙江钱塘江流域土地利用/覆盖变化的“敏感”地类。1991年到2004年年间土地利用/覆盖程度不断加深,其中,1991到1997年间是浙江钱塘江流域快速发展的时期。1991年到2004年间浙江钱塘江流域土地利用/覆盖结构均衡度在不断增加,优势度在不断下降,目前流域土地利用/覆盖结构正向着逐步均衡的方向发展。浙江钱塘江流域1991年到2004年间诸暨市、义乌市是土地利用/覆盖变化的热点地区。其中,1991年到1997年间诸暨市、遂昌市、萧山区、富阳市、桐庐县、义乌市为土地利用/覆盖变化的热点地区,1997年到2004年间萧山区、金东区、婺城区为土地利用/覆盖变化的热点地区。浙江钱塘江流域1991年至2004年期间,水域和建筑用地的重心漂移量较大。建筑用地的重心漂移量最大,漂移方向为东北方。其次是水域,漂移方向为东南方;水田、旱园地、林地的重心漂移量均较小,漂移方向分别水田向北方,旱园向西北方,林地向西南方。更多还原

【Abstract】 Qiantang River Watershed(QRW)is situated at the Yangtze River Delta where is one of the economically most developed areas.For the past decades,the economics and population of QRW have a swift growth,the land use/cover changes remarkably and the Gross National Product(GNP)increases substantially year by year.The area has becomes one of economic development fastest areas in China.However,soil and water resources have not been used and conserved properly in the course of economic development,which leads to the water quality declination and the environment quality deterioration.Previous research indicated that the proportion and changes of land use/cover in the drainage basin was the primary reason why the quality of water and soil declined and the ecological environment deteriorated.Along with the swift development of economy in QRW,there is an urgent need for us to utilize and manage the land resource scientifically and effectively.Therefore,using the remote sensing technique to classify land use/cover,to monitor land use/cover changes over space and time domains will have extremely vital significance to the sustainable development of QRW and even the entire Zhejiang Province.The paper found that the SAM method by integrating TM imagery and slope data was simple,efficient,and straightforward,and it had greatpotentials to be applied in the area with complex physiography and diversed land uses.Then using this method,the thesis studied the dynamics of land use/cover of QRW and performed spatio-temporal analysis for the land use/cover of the area in 1991,1997 and 2004,Which could convey substantial information for supporting the land use structure optimization and provide deep insight for the management of land resources in a scientific manner.The major results from this thesis were as follows:1.Land use/cover classification using multi-source data with SAMLand use/cover change is one of main causes and components of global change.It has been a great challenge for imagery interpreter that how to classify remote sensing imagery with high accuracy and apply in the area with complex physiography and diversed land uses.By using multi-source data with spectral angle mapper(SAM) algorithm,this study tested the effectiveness of these data.Fuyang City,Zhejiang Province was selected as the study site,while Yiwu City of the Province was as a verification site.Results showed that slope information provided the best assistance in land use/cover classification,comparing with other source data such as Elevation, Aspect,normalized difference vegetation index(NDVI),normalized difference water index(NDWI),normalized difference build-up index(NDBI)and normalized difference bareness index(NDBaI).In comparison with the result from TM data only,the classification accuracy with the slope information added increased by about 10%. The SAM method by integrating TM imagery and slope data is simple,efficient,and straightforward,and it has great potentials to be applied in other study areas.2.Land use/cover dynamic monitoring of Qiantang River Watershed in 1991,1997 and 2004With the classification scheme of by layers,regions and sub-region,the study areas were divided into 3 layers,16 regions and 87 sub-regions according to the local characters of geomorphology and landform.Land use/cover in 87 sub-regions of QRW in 1991,1997 and 2004 were classified with SAM from TM images and slope data.The classification results were assessed by the method of examining the random points and by the method of contrasting field-measured points.Both results of accuracy assessment showed that the classification accuracy of all three years were about 80%,with Kappa coefficients above 0.70.The main findings of land use/cover dynamic Of QRW in 1991,1997 and 2004 were:forest relatively invariant;paddy field decreasing, built-up increasing yearly;dry land/garden stable from 1991 to1997,while decreased significantly in 2004;and there was little change of water area in 1991 and 1997,whereas it increased in 2004.3 Spatio-temporal analysis of land use/cover dynamics of Qiantang River Watershed from 1991 to 2004Between 1991 and 2004,built-up was the hotspot type since its changing ratio was the largest;dry land/garden and water were the sensitive types of land since they changed frequently.The diversity of land use/cover kept increasing between 1991 and 2004,with the highest change from 1991 to 1997 in QRW.Land use/cover equilibrium kept increasing,while it was diminishing for some dominant types in QRW from 1991 to 2004.Nowadays,land use/cover of QRW has been gradually approaching the balancing point.Zhuji City and Yiwu City were the hotspot of land use/cover change from 1991 to 2004 in QRW.Zhuji City, Suichang County,Xiaoshan District,Fuyang City,Tonglu County and Yiwu City were the hotspots of land use/cover change from 1991 to 1997,while Xiaoshan District,Jindong District and Wucheng District were the hotspots of land use/cover change from 1997 to 2004.The gravity center of water and built-up shift dramatically.The gravity center of built-up drifted most towards northeast,while water towards southeast;The gravity center of paddy field,dry land/garden,forestland drifted slightly,and their directions were north,northwest,and southwest in sequence.更多还原