Buffering-based approach to fluctuation analysis of glacier calving fronts using LANDSAT-7 ETM+, with a case study of Jakobshavn Isbræ.

The annual or seasonal fluctuation of glacier calving fronts has been carefully mapped and measured, due to this property's close correlation with overall glacier behavior. For that purpose, remote sensing data is the most useful tool; however, LANDSAT-7 ETM+ images, a popular dataset, has been losing parts of its data since May 2003, due to a scan line corrector (SLC) failure of the ETM+ sensor. Therefore, calving front mapping with later LANDSAT-7 ETM+ data requires interpolation or pre-processing. To overcome these issues, we present a novel approach based on line-buffering in a GIS (Geographic Information System) environment. In applying this method to a time-series of the Jakobshavn Isbræ's glacier front position in West Greenland, we proved this approach to be more objective and robust than alternative methodologies; its shape similarity measure, moreover, was shown to be highly useful. Additionally, a simulation series was carried out, which established that the buffering-based method successfully estimated, with high objectivity, both mean displacement and shape similarity from a pair of calving fronts delineated from the SLC-off data without any modification. Further analyses of Jakobshavn Isbræ based on the buffering-based approach revealed that the frontal advance during a winter season preceding a period of high rates of retreat fails to balance the recession of the previous summer. Moreover, based on analyses of the shape similarity of sequential calving fronts, it was determined that a rapid retreat is likely accompanied by a significant change in calving front shape. [ABSTRACT FROM AUTHOR]