地表温度是区域和全球尺度陆面过程研究中的一个关键参数,利用遥感卫星资料反演得到的地表温度数据在气象、水文和生态领域研究中有重要作用.本文基于改进后的针对MODIS 数据的分裂窗口算法,对MODIS L1B 卫星数据进行实用而简便的云检测处理,并根据青藏高原陆地、水体和冰雪等常见下垫面状况的遥感影像分类结果,反演得到了2007-01-03 、04-18 、06-12 和10-02 四日的无云下垫面地表温度.最后,将Sobrino 结果在青藏高原那曲地区与MODIS 日地表温度产品及CAMP/Tibet 观测站地表温度数据进行了对比验证分析.结果表明,该方法得到的地表温度结果与MODIS 数据产品具有较好的一致性,并且地表温度结果与地面观测数据(去除可疑点后)的平均误差仅为1.435 K .

Using split window algorithms developed by Sobrino, added a suff icient and simple cloud detection process for MODIS L1B data. Then based on land surface classif ication (land, water, snow & ice), retrieved four land surface temperature images for clear sky in Naqu area over Tibetan Plateau in January, April, June and October in 2007, respectively. After that the derived LST was compared with the MODIS daily land surface temperature product and in-situ data from the Coordinated Enhanced Observing Period (CEOP) Asia-Australia Monsoon Project (CAMP) on Tibetan Plateau (CAMP/Tibet). The results showed that the LST derived from the improved Sobrino method was in good accordance with the MODIS product and the average difference between LST and in-situ measurements was only 1.435 K.