闽江口湿地信息提取方法及动态变化研究

发布时间：2018-11-06 19:07

【摘要】：闽江是福建的母亲河,闽江口范围涵盖福州市的6个辖区及长乐市、闽侯县、连江县,是福建省的政治、经济、文化中心,对福建的发展意义重大。闽江口湿地的保护是该区经济可持续发展的前提条件,也是后备保证。及时准确的获取闽江口湿地资源的现状及动态变化情况,对于湿地资源的开发与保护、经济的可持续发展具有指导意义。遥感(RS)技术具有观测范围大、信息获取速度快、可重复性的观测、信息量大且获取条件限制少的特点,可为湿地资源调查提供大量数据,成为湿地调查中的新手段。本文结合以往的闽江口湿地分类体系以及闽江口的实际情况,利用Landsat TM影像,将闽江口湿地分为河海、滩涂、水田、坑塘、湖泊水库。以闽江口的亭江至口外海滨段的部分为样区,以2010年10月31日的Landsat TM影像为数据源,分别使用传统的最大似然监督分类、决策树分类与面向对象分类提取样区的湿地分布信息。将三种方法提取的湿地与目视解译得到的湿地进行比较发现,面向对象分类的总体分类精度达到了86.30%,Kappa系数为0.8181,优于监督分类(总精度77.28%,Kappa系数0.6983)与决策树分类(总精度83.20%,Kappa系数0.7744)。故研究使用面向对象分类法提取闽江口的湿地信息。研究使用了一种新的水体提取方法,即SRWI法。并在已有研究成果的基础上探讨了SRWI在大气校正后的TM影像上的水体提取效果,结果显示大气校正能够提高SRWI的水体提取精度。故研究利用SRWI法,使用大气校正后的TM影像提取闽江口的水体信息,结合面积、形状指数等实现水体的分类。研究在ENVI EX的支持下,使用面向对象分类法提取了四个时期的闽江口非水体类湿地分布信息。这四个时期分别是1996年、2001年、2007年和2010年,使用的数据均是在低潮位时成像的Landsat TM影像。将水体提取结果与非水体类湿地相叠加完成了闽江口湿地的提取。对4期湿地的分布分析发现,1996-2010年间,闽江口湿地总面积呈逐年减少的趋势。湿地的减少主要表现为水田的减少,由于影像潮位的不完全相同,河海的面积有增加,滩涂的面积有减少。研究期间内,坑塘的空间变化是最剧烈的,原因是坑塘中用于灌溉的水池的存在状态受季节的影响较强。湖泊水库的分布较稳定,受人工的影响,其边缘会发生少许的变化,总面积也保持相对稳定。从湿地的变化模式上来看,保持稳定的湿地类型主要是河海、滩涂、水田,这些类型均表现为总体面积较大的特征。退缩型湿地的类型主要是水田,主要分布在南台岛、闽侯县的部分地区。波动型湿地的面积也占了相当大的比例,这些波动的湿地主要是由于季节的变化造成的水田与裸地的相互转化。
[Abstract]:Minjiang River is the mother river of Fujian Province. The Minjiang River Estuary covers 6 districts of Fuzhou and Changle City, Minhou County and Lianjiang County. Minjiang River is the political, economic and cultural center of Fujian Province, which is of great significance to the development of Fujian Province. The protection of Minjiang Estuary Wetland is a prerequisite for sustainable economic development and a reserve guarantee. Timely and accurate acquisition of the status quo and dynamic changes of wetland resources in Minjiang Estuary is of guiding significance for the development and protection of wetland resources and the sustainable development of economy. Remote sensing (RS) technology has the characteristics of large observation range, fast information acquisition, repeatable observation, large amount of information and limited access conditions. It can provide a large amount of data for wetland resource investigation and become a new method in wetland survey. Based on the previous classification system of the Minjiang Estuary wetland and the actual situation of the Minjiang Estuary, this paper divides the Minjiang Estuary wetland into rivers and seas, beaches, paddy fields, ponds, lakes and reservoirs by using Landsat TM images. Based on the Landsat TM image of October 31, 2010, the traditional maximum likelihood supervised classification, decision tree classification and object oriented classification were used to extract the wetland distribution information of the sample area, taking the part of the waterfront section from the estuary of Minjiang River to outside the mouth as the sample area and the Landsat TM image of October 31, 2010 as the data source. The comparison between the wetland extracted by the three methods and the wetland obtained by visual interpretation shows that the overall classification accuracy of the object-oriented classification is 86.30 and the Kappa coefficient is 0.8181, which is superior to the supervised classification (the total accuracy is 77.28). Kappa coefficient 0.6983) and decision tree classification (total accuracy 83.20% Kappa coefficient 0.7744). Therefore, the object-oriented classification method is used to extract wetland information from Minjiang Estuary. In this paper, a new water extraction method, SRWI method, is used. On the basis of the existing research results, the extraction effect of SRWI on TM image after atmospheric correction is discussed. The results show that atmospheric correction can improve the precision of SRWI water body extraction. Therefore, the SRWI method is used to extract the water information from Minjiang Estuary by using the TM image after atmospheric correction, and the area and shape index are combined to realize the classification of water body. With the support of ENVI EX, the distribution information of non-water wetland in Minjiang Estuary was extracted by object-oriented classification method. These four periods are 1996, 2001, 2007 and 2010. The data used are Landsat TM images at low tide level. The extraction of Minjiang Estuary wetland was completed by superposing the results of water extraction with non-water wetland. It is found that the total area of the wetland in Minjiang Estuary decreased year by year from 1996 to 2010. The decrease of wetland is mainly the decrease of paddy field. The area of river and sea increases and the area of tidal flat decreases because of the different tidal level of image. During the study period, the spatial variation of the potholes was the most severe, because the existence of the potholes used for irrigation was strongly influenced by the seasons. The distribution of lake and reservoir is relatively stable, and the edge of lake and reservoir will change slightly and the total area will remain relatively stable. From the point of view of wetland change pattern, the main wetland types to maintain stability are river and sea, tidal flat and paddy field, all of which are characterized by larger total area. The type of shrinking wetland is mainly paddy field, mainly distributed in Nantai Island, part of Minhou County. The area of fluctuating wetland also accounts for a large proportion. These fluctuating wetlands are mainly caused by the transformation of paddy field and bare land caused by seasonal changes.
【学位授予单位】：福建师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：X171.1

【参考文献】