基于Landsat影像的敦煌西湖土地利用变化研究

发布时间：2018-03-14 21:26

本文选题：决策树　切入点：面向对象　出处：《中国地质大学(北京)》2017年硕士论文　论文类型：学位论文

【摘要】：土地作为人类活动的最基本要素,受人类活动深刻影响,随着经济水平增长和人口数量增加,土地利用类型频繁变化,一系列生态环境问题随之而至,生态环境逐渐恶化。在我国西北地区,生态环境恶化等问题尤为突出。同时随着遥感技术和GIS技术的快速发展,基于传统方法下的土地利用类型调查逐渐被其替代。本文选择位于我国生态环境脆弱带上的甘肃敦煌西湖湿地为研究区,以1986-2015年之间的Landsat系列影像(TM和OLI)为主要数据源,采用决策树和面向对象的遥感影像分类方法,借助ENVI、Arcgis、eCognition等软件,对研究区土地利同类型进行提取。其后通过面积转移矩阵和自然因素对土地利用类型变化进行综合分析。本文的主要研究包括:(1)通过对决策树和面向对象分类方法对比,面向对象分类方法的总体精度为91.9996%,Kappa系数为0.8946,高于决策树分类方法89.4705%的总体精度和0.8751的Kappa系数,证明基于中低分辨率影像的面向对象分类也具有较好的分类精度。(2)通过面向对象分类方法,对研究区多时段遥感影像进行土地信息提取。研究区内主要土地利用类型中流动沙丘、裸地和戈壁占研究区总面积70%以上,植被覆盖地区占总面积约10%,盐碱地占总面积约5%,沼泽湿地所占研究区总面积仅为0.63%。(3)对研究区多时段土地利用类型信息进行转移矩阵计算,沼泽湿地面积逐渐减少,面积由41.69km2减少到14.43km2;盐碱地面积逐渐增加,面积203.07km2减少到344.64km2;戈壁、干枯河床等类型面积保持稳定,植被覆盖区和裸地面积呈动态变化。盐碱地同裸地和低植被覆盖区面积转移较多,沼泽湿地同高植被覆盖区域和裸地面积转移较多,高、中和中、低植被覆盖区面积转移较多。(4)对研究区土地利用类型变化同降水量和温度进行性相关性分析,根据其之间的相关性系数表明,戈壁、山地、流动沙丘等未利用地同降水、温度的相关性较小,沼泽湿地面积减少、盐碱地面积增加主要受人类活动的影响。高植被覆盖区域受温度影响较大,中、低植被覆盖区域受降水影响较大。
[Abstract]:As the most basic element of human activities, land is deeply influenced by human activities. With the increase of economic level and population, land use types change frequently, and a series of ecological and environmental problems come along. The ecological environment is deteriorating gradually. In northwest China, the deterioration of ecological environment is especially prominent. At the same time, with the rapid development of remote sensing technology and GIS technology, Based on the traditional method, the land use type survey is gradually replaced by it. In this paper, the wetland of Dunhuang West Lake, Gansu Province, which is located in the fragile ecological environment of our country, is chosen as the research area, and the main data sources are the Landsat series images of TM and OLI from 1986 to 2015. With the help of decision tree and object oriented remote sensing image classification, and with the help of ENVIX ArcGIS Cognition and other software, The paper extracts the same type of land interest in the study area, and then makes a comprehensive analysis of the change of land use type through the area transfer matrix and natural factors. The main research in this paper includes the comparison of decision tree and object-oriented classification method. The overall accuracy of the object-oriented classification method is 91.9996 and the Kappa coefficient is 0.8946, which is higher than the total precision of decision tree classification method 89.4705% and the Kappa coefficient of 0.8751. It is proved that the object oriented classification based on low and medium resolution images also has good classification accuracy. Land information was extracted from multi-time remote sensing images of the study area. Among the main types of land use in the study area, mobile sand dunes, bare land and Gobi accounted for more than 70% of the total area of the study area. The total area of vegetation cover area is about 10, salt alkali land is about 5, the total area of the study area of marshland is only 0.63. 3) the information of land use type in various periods of study area is calculated by transfer matrix, and the area of marsh wetland decreases gradually. The area decreased from 41.69 km ~ 2 to 14.43 km ~ 2, the area of saline-alkali soil increased gradually, and the area of 203.07 km ~ 2 decreased to 344.64 km ~ 2. The vegetation cover area and bare land area showed dynamic change. The area transfer of saline-alkali land and low vegetation cover area was more than that of marsh wetland with high vegetation cover area and bare land area. The change of land use type in the study area was analyzed with precipitation and temperature. According to the correlation coefficient, it was found that Gobi, mountainous area, moving sand dune and other unused land had the same precipitation. The correlation of temperature is small, the area of marsh wetland decreases, the area of saline-alkali land increases mainly by human activities, the area of high vegetation cover is affected by temperature more, and the area of low vegetation cover is affected by precipitation.
【学位授予单位】：中国地质大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P237;F301.2

【相似文献】