【摘要】 基于环境卫星影像利用支持向量机法提取鄱阳湖近十年枯季湿地信息,并结合各湿地类型的时空变化分析其影响因素,得出结论:①支持向量机法总体分类精度为97.52%,相比于最大似然法、决策树法在鄱阳湖湿地信息提取中有更好的提取效果;②十年间水域面积平均为1209.45 km~2,天然水域主要分布于湖区东部和北部,圩堤湖水域面积几乎不变;枯季以草洲为主,广泛分布于南矶乡地区和入湖河道两侧洲滩;裸滩变化幅度较大,总体与水域呈负相关性;沙地面积最少,最大面积不足50 km~2,大部分分布于松门山岛;③气候、地形地势及城镇建设、人工水利、围垦、采砂、捕捞、排污等人类干预活动均在不同程度上影响鄱阳湖湿地类型的演变.更多还原

【Abstract】 Based on environmental satellite imagery, information of Poyang Lake wetland in the past ten years was extracted by using the support vector machine(SVM) method. Then the information was analyzed while factors affected the spatial and temporal changes of each type of wetland were taken into consideration. After th at, a conclusion was made as follows : ① The SVM method had an overall accuracy rate of 97.52 %, better in extracting Poyang Lake wetland information than the maximum likelihood method and the decision tree method;②The average water area in the past decade was 1209.45 km~2. Natural waters were mainly distributed in the eastern and the northern parts of the lake, and the water area of the Dyke Lake was almost unchanged. Wet meadows were widely scattered in Nanji Town and the beaches on both sides of river channels leading to Poyang Lake, taking up the largest area among other types of wetland during low flow period. The beach without any plant changed a lot and had an overall negative correlation with waters area. The sand land was the smallest, covering less than 50 km~2. Most of the sand land was distributed in Songmenshan Island. ③Climate, topography and human intervention activities such as urban construction, the construction of artificial water conservancy, reclamation, sand mining, fishing, and sewage discharge all affect the evolution of wetland types in Poyang Lake at varying degree levels.更多还原