Extraction Of Glacier Area Using Optical Remote Sensing Images

The global glaciers are undergoing accelerated retreat and increasingly negative mass balances with the ongoing climate warming. Monitoring of glacier changes is urgently required for a wide range of purposes, including the modeling of contribution of glacier to sea-level rise, the predicting of future climate change, and assessment water resources on a regional scale. Glacier area is a main parameter to study glacier change.Remote sensing based on multi-temporal and multispectral images has been proved to be a particularly tool for mapping glacier extent and monitoring changes. this paper presents a synergistic approach to produce glacier outlines with NDWI (Normalized Difference Water Index) and band ratio; Using the synergistic approach developed, we investigated glaciers area change over the20years in Geladandong mountain, Tibetan Plateau and the correlation between glacier and climate variation; By analyzing the remote sensing data, the HJ-1A/1B CCD image is firstly used to extract glacier area in this paper. The main research achievements consist of:(1) Present a synergistic approach to produce glacier outlines with band ratio and NDWI (Normalized Difference Water Index). By comparing the presented method and band ratio result with reference data from visual image interpretation, we think the presented method extracted glacier outline more accurately and eliminated the misclassified water bodies, which makes the glacier mapping more efficient.(2) Use the synergistic approach of band ratio and NDWI to extract the glaciers and glacial lakes area of Geladandong for year1992,1999,2006and2009, respectively. Results show that the glacier variation is dominated by intensive recession, while the glacial lake had grown continuously during the period1992-2009. In addition, the variability of glaciers area and glacial lake area exhibited negative no-linear relationship. According to the meteorological data, we think glacier area loss mainly resulted from rising summer temperature.(3) Use HJ-1A/1B CCD image firstly to extract glacier area with three remote sensing classification methods, including maximum likelihood classifier, Support Vector Machine and Decision Tree based of HJ-1A/1B CCD image; the experiment results show that the Support Vector Machine method achieved the highest accuracies.