Mapping the margin of the Barnes Ice Cap using SAR imagery

Short, Naomi (1999) Mapping the margin of the Barnes Ice Cap using SAR imagery. Masters thesis, Memorial University of Newfoundland.

[English] PDF (Migrated (PDF/A Conversion) from original format: (application/pdf)) - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (9MB) [English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. (Original Version)

Abstract

The limit of active ice, accurate to within 49 m, is extracted from SAR imagery of the Barnes Ice Cap, using a sequence of image speckle filtering, image texture analysis, supervised image classification, image segmentation and edge detection. Overall classification accuracies of the ice marginal environment are between 42 and 53%. Despite misclassification of some proglacial landforms it is possible to detect the limit of active ice as these surfaces are separated from the ice cap by supraglacial debris cover, an elevated ice cored debris ridge and perennial snowbeds. -- Comparisons of opposing look angles reveal that 'downglacier' illumination produces the highest classification accuracies of ice marginal features and debris covered ice surfaces, whereas illumination from an off-ice perspective, looking 'upglacier', is found to produce higher classification accuracies for proglacial surfaces. Comparisons of standard and fine mode imagery conclude that 25 m spatial resolution yields higher classification accuracies than 8 m spatial resolution. Quantitative analysis of surface roughness demonstrates that the dominant grain size of surficial materials is the best method for relating surface cover to radar brightness. -- Second order texture measures mean, homogeneity and correlation are found to be effective variables for maximum likelihood classification of ice marginal SAR imagery. Texture window size should be as large as the smallest feature to be identified, in this case approximately 85 x 85 m for the mean and homogeneity measures and 185 x 185 m for the correlation measure.

Actions (login required)

View Item