Edinburgh Geography: Dissertation Abstract

University of Edinburgh
School of GeoSciences

M.Sc. Geographical Information Science

KINNEAR, D. A. (2015)

"Automated detection and tracking of crevasses on a calving glacier from TerraSAR-X imagery"

Keywords: crevasse calving glacier svalbard feature tracking detection automated tunabreen TerraSAR-X


The formation and development of crevasses provides great insight into the overall dynamics of a calving glacier, in particular because the formation of crevasses reflects glacier stretching, the same process that leads to calving. If a crevasse opens to the water-line then a calving event normally occurs. Thus outlining where and when crevasses form and how fast they open up, can be very important in detecting pre-conditions for calving events. Detecting crevasses from satellite imagery is an efficient way of tracking their movement, but manual techniques for identifying individual features and monitoring their evolution are time-consuming. In addition, most optical data is subject to inconsistent illumination due to time of day or season, long periods of darkness during winter, and cloud cover. High-resolution radar imagery provides an alternative that does not rely on illumination by the sun and is independent of weather constraints. This study uses a series of TerraSAR-X images of Tunabreen, a grounded calving glacier in Svalbard, taken between February 2013 and July 2014, to automatically detect and delineate crevasses close to the calving front. Feature tracking techniques are used to measure the relative strain taken up by these crevasses in relation to the ‘inter-crevasse’ ice between them, as the glacier approaches the calving margin. Strain rates follow a surface velocity gradient that increases as the glacier approaches the calving margin, but both velocity and strain rates level off and, in some cases, decrease in the final ~300m before the terminus. Seasonal fluctuations are also observed.

View Web Showcase for this Dissertation

Search Edinburgh Research Archive for this Dissertation