基于光学和热红外遥感影像的易贡藏布流域冰川识别

摘要

Based on the Landsat-5 TM image,combined with the characteristics of glacier spectrum in the thermal infrared band, the glacier information in Yigong Zangbo River Basin in 2010 is extracted.First,the radiometric calibration is conducted using optical band of TM image.Then,thermal infrared band of TM and MOD05 water vapour products are used to extract surface temperature.Finally,the light and shadow areas are respectively classified by maximum likelihood method and the random point method is used for accuracy assessment.It is showed that the glacier area of Yigong Zangbo River Basin was 2 698 km2 ,the cov-erage was 27.5%,and the glacier distribution was in the range of 3 500-6 000 m in 2010.The overall accuracy of the glacier iden-tification was 75% and Kappa coefficient is 0.68.%基于 Landsat-5 TM 影像，结合冰川光谱在热红外波段的特征，提取了2010年易贡藏布流域冰川的信息。首先将 Land-sat-5 TM 影像进行辐射定标，然后利用热红外波段和 MOD05水汽数据产品进行地表温度反演，最后利用最大似然法对影像的光照区和阴影区分别进行分类，并利用随机点方法进行精度评价。结果表明：2010年易贡藏布流域冰川面积为2698 km2，冰川覆盖度为27．5％，冰川基本分布在3500～6000 m。计算得到冰川识别总体精度为75％，Kappa 系数为0．68，分类精度较好。