基于面向对象分类方法和多源遥感数据的龙祥岛区域湿地变化检测

Wetland Change Detection in Longxiang Island Area Based on Object-based Classification Method and Multi-source Remote Sensing Images

利用龙祥岛区域2001年ASTER影像、2013年GF-1影像以及Google Earth影像等辅助数据,基于面向对象分类的方法,建立各地物类型的提取规则,提取出龙祥岛区域湿地-非湿地信息,并基于e Cognition软件对龙祥岛区域湿地进行变化检测,最后利用Excel统计软件作出湿地-非湿地信息面积转移矩阵及面积转移率矩阵.结果表明:2001年ASTER影像分类总体精度为87.42%,Kappa系数为0.84;2013年GF-1影像分类总体精度为90.72%,Kappa系数为0.87.研究区2001-2013年间,湿地总面积减少了4.05%,为1.1 km^2,其中天然湿地(包括永久性河流和滩涂)总面积减少了1.65%,为0.42 km^2,人工湿地(包括淡水养殖场和农用池塘)总面积减少了39.76%,为0.68 km2;非湿地总面积增加了4.71%,为1.1 km^2.人工湿地和天然湿地内部变化情况呈现出不同的状态,天然湿地中,约有3.888 km^2的滩涂面积变为永久性河流,主要变化区域位于研究区的东南部和西北部;而在人工湿地中,约有0.663 km^2的淡水养殖场和0.41 km2的农用池塘变为非湿地,主要变化区域位于龙祥岛本岛及研究区的西南角部分,人工湿地变化的主要原因是人类生产生活进程的加快.

Based on Terra ASTER image in 2001, GF-1 PMS image in 2013 and other auxilia- ry data such as Google Earth image of Longxiang island area, this paper builds up extraction rules of each class using the object-based classification method, extracts wetland and non-wetland informa- tion of Longxiang island area and conducts wetland change detection of Longxiang island area based on eCognition software. At last, this paper makes area conversion matrix and area convertible rate matrix of watland and non-wetland information based on Excel software. The result shows that： The overall classification accuracy of Terra ASTER image in 2001 is 87.42% , Kappa coefficient is 0. 84. The overall classification accuracy of GF-1 PMS image in 2013 is 90. 72% , Kappa coeffi- cient is 0. 87. From 2001 to 2013, the total wetland area decreases 4.05% and the decreased area is 1.1 km2. The total natural wetland area which includes permanent river and shallows decreases 1.65% and the decreased area is 0. 42 km2. The total artificial wetland area which includes fresh- water aquaculture farms and agricultural ponds decreases 39. 760/0 and the decreased area is 0. 68 km2. In addition, the total non-wetland area increases 4.71% and the increased area is 1.1 km2.The internal changes of artificial wetland and natural wetland appears different situation. In natural wetland, there is about 3. 888 km2 shallows which turns to permanent river, and the major change area is located in the southeast and northwest of Longxiang island area. However, in artificial wet- land, there is about 0. 663 km2 freshwater aquaculture farms and 0. 41 km2 agricultural ponds which turns into non-wetland and the major change area is located in the Longxiang island and the south- west of the study area, the acceleration of human society and production process is the main reason which causes the drastic changes of artificial wetland.