基于改进型决策树遥感分类的土地利用变化研究

Research on Land Use Change Based on Improved Decision Tree's Remote Sensing Classification

在遥感和GIS技术支持下,以香格里拉建塘镇为研究区,以2000年、2009年Landsat TM影像为基础数据,运用光谱特征和地学辅助知识相结合的改进型决策树分类方法分别对研究区2个年份的基础数据进行土地利用信息提取和变化分析。研究结果显示:(1)基于改进型决策树遥感分类法与最大似然法相比分类精度有明显提高;(2)10 a间,香格里拉建塘镇城镇建筑用地、裸地面积呈上升趋势且变化最大,灌草地面积小幅增加,耕地、林地、水体、雪地冰川呈减少趋势,土地利用变化速度和综合程度越来越快。

Under the supports of remote sensing and GIS technologies,taking Jiantang Town,Shangri-La as the research area,this paper selected Landsat TM images in 2000 and 2009 as basic data,and used the improved decision tree classification method,which combined spectral feature with geoscience auxiliary knowledge,to extract the land use information for the research area's basic data in two years.The results show that ① compared with the maximum likelihood classification,the accuracy of improved decision tree remote sensing classification has improved significantly.② Within 10 years,the areas of urban building land and bare land in Jiantang Town,Shangri-La witnessed an upward trend and the greatest change.The grass land area increases slightly.The areas of farmland,woodland,waters,snowfield and glaciers reduce,and land use's rate of change and comprehensive degree become even faster.