Accuracy evaluation of glacier area information extracted from Landsat-5 TM remote sensing image

【Objective】This study aimed to evaluate the accuracy of glacier area information extracted from Landsat-5TM remote sensing image.【Method】The Glacier No.12in Laohugou valley and the glacier No.3in Ningchan river located in the north slope of Qilian mountain were selected research objectives,and GPS-RTK technology was used to obtain high-precision glacier boundary data in 2009.Interpretation results of Landsat-5TM remote sensing image at the same period using geodetic method,visual interpretation method,normalized difference snow index(NDSI)method,and supervised classification method were compared.【Result】Areas of the glacier No.12in Laohugou valley and the glacier No.3in Ningchan river extracted from Landsat-5TM remote sensing image were less than ground measurements,and the errors of different methods were different.Visual interpretation method had the minimum errors of-0.79% and-1.83%for extraction of glaciers No.12and No.3followed by NDSI with errors of-1.19% and-3.82%,respectively.The supervised classification method had the maximum errors of-1.79% and-5.99%,respectively.【Conclusion】It is simple and effective to monitor glacier change using Landsat-5 TM image.The visual interpretation method is the best extraction method with errors of less than±2%. The extraction methods performed better on larger glaciers compared to smaller glaciers.