This study extracted the local glacier information over the Nianchu river basin in the Tibet in 1996 and 2005 by using ice index,snow index and water index of Landsat TM-5 multi-spectral images,and Synthetic Aperture Radar(SAR)intensity and coherence information of ERS-1/2 and Envisat ASAR.The optimized features were determined by their classification accuracies based on Support Vector Machine (SVM)classifier.The result showed that the composition of multi-spectral and SAR features could effectively discriminate the water and ice from other types,with overall accuracies of 84%and 85%in 1996 and 2005,respectively.Based on the thematic information of these two years,the changes of the local glacial area and boundary were detected.The result showed that the glacial area of the Nianchu river basin was reduced by 154.7 km~2,which mainly caused by the climate warming.