【作者】 韩贵锋；

【导师】 徐建华；

【作者基本信息】 华东师范大学 ， 地图学与地理信息系统， 2007， 博士

【摘要】 植被是陆地生态系统的重要组成部分，是生态系统中物质循环与能量流动的中枢，也是对人类社会经济活动有重要贡献的资源。选择我国东部地区作为研究区（113°-123°E，21.5°-35.5°N），以SPOT／VGT-NDVI时间序列影像为主要数据源（1998-2005），分析植被覆盖的时空变化及其人为因素的影响。从1998到2005年，我国东部地区植被的总面积有缓慢下降趋势，但是植被活动在增强。植被减少趋势与经济发展呈正相关的关系。在植被分布较多的地区，植被活动呈现出明显的下降趋势；植被较少的地区，呈现出明显的增加趋势。长江以北地区的植被增加趋势明显，尤其在安徽北部的阜阳和毫州地区以及鄱阳湖周边地区；长江以南地区植被减少趋势十分显著，尤其在江苏和上海接壤地区、上海市、浙江东南沿海地区、广州周边地区、南昌西南地区、福建厦-漳-泉地区以及湖南长-株-潭地区。东部各省（市）的植被活动变化差异十分显著：上海表现出明显的下降趋势；福建、浙江和江西有轻微的下降；安徽和江苏两省的植被表现出了相似的上升趋势。东部地区的植被重心1998到2005年均位于江西北部的上饶境内，向西北方向移动的趋势明显。植被的年内变化表明，东部地区植被的生长期为240-320天，生长期最长的位于北纬28-30度之间的低海拔地区，NDVI峰值出现的时间最早为第18旬即6月下旬，最晚为第28旬即10月上旬。8年平均的年积分NDVI的大小顺序为，福建＞浙江＞江西＞安徽＞江苏＞上海；低值区的植被增加趋势明显，而高值区的增加不明显。为了提取不同植被类型的NDVI年内变化曲线，使用主成份分析和ISODATA非监督分类方法对东部地区的植被进行了分类，分类kappa指数高达0.82，得到了18类主要的植被类型，其中一年两熟农作物和亚热带常绿阔叶林所占面积较大，每类植被均有其特殊的NDVI年内变化曲线。东部地区人口密度大，经济比较发达，城市扩张明显。植被分布与人口、GDP和建设用地GDP之间有明显的负相关关系，相关性在空间上有显著的异质性，在经济发达地区负相关性较高，而在河流、湖泊以及近海岸周围，存在正相关性。结合城市化的统计数据，应用面板数据分析方法，分析得出植被和城市化率之间的负相关性逐年增强，植被和城市化率之间有对数关系。采用多环缓冲带取样，研究发现城市化对植被物候的影响也是明显的；在长三角的上海市、杭州市、南京市、苏州市、无锡市和常州市等6个城市中，除杭州市外，城市化使植被的减少在10km缓冲带内表现得十分显著；城市化提前了始率期（SOS），延后了终绿期（EOS），从而延长了生长期（GSL），但是使NDVI年内极差(NDVI_amp)降低幅度更大；从6个城市的平均水平看，城区植被和8-10km缓冲带植被相比，生长季节延长了5天，NDVI_amp减少了0.21。利用TM影像，使用变化轨迹方法对典型快速城市化地区——上海市植被的研究表明：14年来，上海市的植被面积呈现持续下降趋势，浦东新区植被减少最多；三期（1989、1997、2003）均为植被的面积占总面积的一半以上，其次是三期均为非植被的面积占总面积的1／5。早期植被转化为非植被的地区主要在城区周围，而近期的转化发生在距城区较远的地区；非植被向植被的转化面积很小。Logistic回归分析发现，14个影响因子对植被变化轨迹的影响强度大小依次是，离道路的距离＞离行政中心的距离＞离植被—非植被边界的距离＞离商业中心的距离=1990年人口密度＞离河流的距离＞离高速公路的距离＞土地利用多样性＞人口密度差（2003-1990）=离上海市中心的距离。其中，离高速公路的距离对植被变化的影响反常，即离高速公路越近，植被分布越多或者转化为非植被的概率越小。经过检验，回归模型的精度是满意的，其中二分类Logistic模型精度高于多分类Logistic模型。更多还原

【Abstract】 Vegetation is an important component of the terrestrial ecosystem, a centrum of materials cycle and energy flow and the vital resource for human development. Eastern China is selected as this study region with the longitude ranging from 113°E to 123°E and the latitude ranging from 21.5°N to 35.5°N. Based on SPOT/ VGT-NDVI images in different years, the aim of this research is to analyze spatio-temporal change of vegetation cover and influence of human factors on it.The total area of the vegetation in east China decreases gradually from 1998 to 2005, but the vigor of the vegetation in this area has an increasing trend. There is a positive correlation between vegetation decrease and economic development. The vegetation in northern region with low dense vegetation cover of Yangtze River has an obvious increasing trend especially in Fuyang city and Bozhou city of Anhui Province and the area around Poyang Lake. In the southern region of Yangtze River with high dense vegetation cover, vegetation has an obvious decreasing trend especially around Shanghai city, Suzhou city and Guangzhou city, the southeastern Zhejiang province, the southwestern Nanchang city, Xiamen-Zhangzhou-Quanzhou region in Fujian province and Changsha-Zhuzhou-Xiangtan region in Hunan province. The change of vegetation vigor differs among provinces in the study area. There is a distinct decreasing trend in Shanghai city, a slight decreasing trend in Fujian province, Zhejiang province and Jiangxi province and an increasing trend in Anhui province and Jiangsu province. The gravity centre of the vegetation in the study area always lies in Shangrao city of Anhui province and has been moving toward northwest since 1998.Then the change of the vegetation in any year from 1998 to 2005 is analyzed. Result shows that the Growth Season Length （GSL） of the vegetation in east China ranges from 240 to 320 days and the vegetation with longer GSL often distributes in low latitude region locating with the range of latitude from 28°N to 30°N. The earliest peak NDVI appears on the 18^th dekad （the third dekad of June）, and the latest peak NDVI appears on the 28^th dekad （the first dekad of October）. A descending sort of the Integrated NDVI （INDVI） in a year is Fujian province, Zhejiang province, Jiangxi province, Anhui province, Jiangsu province and Shanghai city. The vegetation in the region of low integrated NDVI increased obviously, whereas unobvious in high INDVI region. In order to extract the NDVI curves of a year of the vegetation, the vegetation in east China is classified into 18 types using the methods of Principal Components Analysis （PCA） and Iterative Self-Organizing Data Analysis Technique Algorithm （ISODATA）. Classification accuracy is satisfying and kappa index is 0.82. Double-cropped crop and sub-tropical evergreen broadleaf forest have larger area than other types of vegetation. Each type of vegetation has its own NDVI curves of a year.There are most cities with high population density, quickly economic growth and obvious urban expansion in the study area. The vegetation distribution correlates negatively with Gross Domestic Product （GDP） per square kilometer, population density and GDP per square kilometer on construction land respectively. However, the situation of the correlation has evident heterogeneity in space. Strong negativecorrelation occurs in relatively developed areas, while positive correlation occurs around river, lake and coast. Analysis of the statistical data from 1998 to 2005 with the panel method shows that negative correlation between vegetation and urbanization has been increasing year by year. And logarithmic relationship is found between the vegetation cover and the urbanization. This research also shows that influence of urbanization on vegetation growth in a year is obvious according to multi-ring buffer sampling. In Yangtze River Delta, such 6 cities as Shanghai, Nanjing, Hangzhou, Suzhou, Wuxi and Changzhou, vegetation cover reduces especially within 10 kilometer far away from city center except Hangzhou city. Urbanization makes Start of Growing Season （SOS） earlier, makes End of Growing Season （EOS） later. So GSL is lengthened, but NDVI amplitude is declined more obviously. It’s clear that GSL of the vegetation in urban is longer 5 days and the NDVI amplitude is smaller 0.21 than that in area which locates 8-10 kilometers away from city center by comparing the internal difference in the six cities with the average level of GSL and the NDVI amplitude of the six cities.At last Shanghai city is studied as a typical rapidly expanding urban area using Landsat TM images. The urban vegetation change trajectory from 1989 to 2003 is analyzed in this paper. The total vegetation area in Shanghai city appears continued downward trend particularly in the Pudong district. Stable vegetation trajectory in the three periods （1989, 1997 and 2003） occupies more half of total area of Shanghai city and stable non-vegetation trajectory in the three periods accounts for about twenty percent of total area of Shanghai city. Transformation from vegetation to non-vegetation in the first phase （1989 to 1997） usually takes place nearby the urban, whereas the same transformation in the second phase （1997 to 2003） often occurs in area far away from the urban. There is small area of transformation trajectory from non-vegetation to vegetation in two phases. Analysis with the Logistic regression model shows that the distance from road has the strongest influence on vegetation change trajectories than other 13 factors which in descending sort according to the influence degree is the distance from centers of districts, distance from edge between vegetation patch and non-vegetation （1997）, distance from edge between vegetation patch and non-vegetation （1989）, distance from business center, population density 1990, distance from river, distance from expressway, land use diversity within 100 meters, population density different between 2003 and 1990 and distance from the center point of Shanghai city. It is worth mentioning that the influence of distance from expressway on vegetation change trajectory is opposite compared with those of other 13 factors. That is to say that more near the expressway more vegetation distributes and the probability of transformation from vegetation to non-vegetation is less. The results also indicate regression models have a satisfying accuracy and binary logistic model is better than multinomial logistic model.更多还原