PROSPECTIVE OF HIGH RESOLUTION WORLDVIEW-2 SATELLITE DATA FOR GEOSPATIAL SURFACE FACIES MAPPING OF AN ALPINE GLACIER

Hits: 140
Research areas:
Year:
2017
Type of Publication:
Article
Keywords:
Object Based Classification, Pixel Based Classification, Customized Spectral Indices, Error Matrices
Authors:
Sagar F. Wankhede, Alvarinho J. Luis Shridhar D. Jawak
Abstract:
Alpine glaciers are dynamic in nature; and are thus seen as sensitive indicators of climate change. The study of glaciers with respect to these changes must be carried out by first mapping the glacier surface facies. This study has attempted to map the available range of surface facies of an unnamed glacier from the Chandra Basin, in the Great Himalayan Range, Himachal Pradesh. The classification of these facies has been carried out using object-based and pixel-based approaches. Conventional glacier facies classification has usually utilized data acquired in the melt season. This study however has endeavoured to map facies on data acquired during early winter. WorldView-2 high-resolution imagery has been used to develop customized spectral indices using the new bands in its multispectral range. Error matrices were utilized to assess the classification accuracies. The object based approach has been found to have an overall accuracy of 88.33%. Two pixel based classifiers were utilized, yielding overall accuracies of 81.67% and 78.33% respectively. The highest kappa statistics obtained for the objectbased strategy is 0.86 and the pixel based classifiers delivered Kappa statistics of 0.78 and 0.74 respectively. The results clearly indicate that the object based classification is superior to the pixel based classification methods.
Full text: 177.pdf [Bibtex]
You are here: Home ACRS ACRS Overview Proceedings prospective-of-high-resolution-worldview-2-satellite-data-for-geospatial-surface-facies-mapping-of-an-alpine-glacier