摘要
以Landsat 8 OLI/TM5数据、DEM数据、ISODATA分类结果等构建多源数据集,采用CART决策树方法提取乌审旗2005年、2010年和2015年三期土地利用数据。基于"3S"技术,利用此三期土地利用数据对研究区2005年~2015年土地利用变化特征进行时空变化分析。结果表明:(1)CART决策树分类图的精度分别为86.5%、88%、91.5%,较监督分类精度有所提高,可用于本次土地利用变化研究分析;(2)乌审旗主要的土地利用类型为草地、未利用地和林地;2005年~2015年水域、林地和建设用地的面积均持续上升;耕地与未利用地均先以较快的速度增加后以较快的速度减少,整体为增加趋势;草地面积则先以较快的速度降低,后以较慢的速度增长;(3)未利用地及草地的转化比例占研究区所有土地利用类型转化比例的80.38%,说明未利用地和草地是乌审旗境内极为重要的土地利用现象;(4)2005年~2010年土地利用综合程度减小,而2010年~2015年土地利用综合程度呈现上升趋势。总体来看,2005年~2015年土地利用程度综合指数整体有所增大,说明乌审旗的土地利用正处于发展时期。
A multi-source data set was built with Landsat 80LI/TM5 data, DEM data, and ISODATA classification re- suits, and the three periods of the land use information of Wushen County in 2005, 2010 and 2015 was extracted with the CART decision tree method. The three periods of land use data were used to analyze the temporal and spa- tial variation of the land ffse of the study area in 2005 - 2015 by using the "3S" technology. The results show that the classification diagram accuracy of the CART decision tree were 86.5%, 88%, 91.5%, respectively, showing that it can be used for the analysis of this land' s use change. The major land use types of Wushen County are grassland, unused land and woodland. The areas of water, woodland and construction land were rising in 2005 ~ 2015. And in the same period, the arable land and the unused land were first increasing at a faster rate and then reducing at a faster rate, but the overall trend was the increasing; The grassland area was first reducing at a faster rate, and then increasing at a slower rate. The conversion area of the unused land and grassland accounted for 80.38% in the whole conversion area in the region, indicating that the unused land and the grassland are the ex- tremely important land use phenomenon of Wushen County. The integrated land use degree decreasea in 2005 2010, while it was rised in 2010 -2015. Overall, the land use degree composite index in 2005 -2015 was in- creased as a whole, indicating the land use of Wushen County was in the development period.
出处
《高技术通讯》
北大核心
2017年第5期479-486,共8页
Chinese High Technology Letters
基金
国家自然科学基金(51169016)资助项目
关键词
多源信息
土地利用
决策树分类
转移矩阵
变化图谱
multi-source data, land use, decision tree classification, transfer matrix, change map