摘要
以云贵高原典型区域贵州省赤水市及其邻近区域为研究区,以2010年9月该区域先进的陆地观测卫星(advanced land observing satellite,ALOS)多光谱影像为数据源,在充分分析和统计影像光谱特征的基础上,对比8种典型地物在NDVI(normalized difference vegetation index)、NDWI(normalized difference water index)、DEM(digital elevation model)3种数据及影像波段运算后特征值的数值差异,总结出区分8类典型地物的阈值,建立了以阈值为规则的决策树模型,进行土地利用分类。结果表明:该方法的分类总体精度达到89.05%,Kappa系数为0.87,能较好地利用云贵高原典型区域的地物信息,有效提高地物分类精度,在研究区具有良好的适用性。
The advanced land observing satellite(ALOS)muhispectral image gained in September,2010 of the Chishui City area and its adjacent region which were located in the Yunnan-Guizhou Plateau was used in this study. Based on the statistics analysis of the image spectral characteristics, the numerical differences of eight typical land use type~ of the normalized difference vegetation index (NDVI), normalized difference water index ( NDWI ), digital elevation model (DEM) &Lta and image band calculated data were compared. A decision tree model for land use classification was built after the thresholds of different land use types were determined. The results showed that the overall accuracy and the Kappa coefficient of classification result had been improved to 89.05% and 0.87 ,respectively. It indicated that ,the decision tree method was suitable for the study area and could better utilize the ground feature information as well as improve classification precision on the region of the Yunnan-Guizhou Plateau arena which has complex terrain features.
出处
《南京农业大学学报》
CAS
CSCD
北大核心
2013年第6期45-50,共6页
Journal of Nanjing Agricultural University
基金
长江委员会资助项目(2009112803)
江苏省高校优秀中青年教师和校长境外研修资助项目
关键词
ALOS多光谱影像
土地利用分类
决策树模型
云贵高原
ALOS multispectral image
land use classification
decision tree model
Yunnan-Guizhou Plateau