基于Landsat影像USGS地表反射率产品的土地利用年际变化成因分析

Analysis on the causes of inter-annual change of land use based on Landsat imaging USGS surface reflectance product

应用支持向量机机器学习方法,对贵阳市及周边地区Landsat影像USGS地表反射率产品进行分类,通过比较2011~2015年五年区域地表覆盖分类结果图,结合区域工业和建筑业产值数据,分析该区域地表覆盖和土地利用年际变化成因,分析结果表明:(1)5年间随着经济社会的发展,呈现出林地和耕地的面积不断减少的土地利用变化格局;(2)研究区5景影像分类精度较高,总体精度范围91.14%~98.60%,Kappa系数范围0.93~0.98,分类结果真实的反应了研究区地表覆盖及其变化情况。从分析结果来看,优化城市发展空间格局,有序开展植被修复并进行持续遥感监测十分必要。

By using the support vectormachine learning method, the Landsat imagery USGS surface reflectance products in the surrounding area of Guiyang were classified. By comparing the classification maps of regional land cover from 2011 to 2015, combining regional industrial and construction industry output data, the causes of the surface cover and the land use interannual variation in the region were analyzed. The results showed:(1) With the development of economy and society in 5 years, the pattern of land use change with declining area of the forest land and the cultivated land was presented;(2) The classification accuracy of 5 scenes was high in the study area, the overall accuracy range was 91.14%~98.60%, Kappa coefficient range was 0.93~0.98, the classification results truly reflected the surface coverage and the changes in the study area. From the analysis results, it was necessary to optimize the spatial pattern of urban development, orderly carry out the vegetation restoration and the continuous remote sensing monitoring.