Mapping moraines and glaciers using multispectral imagery and ancillary elevation data : An approach for Svalbard, Norwegian Arctic

• Main Author: Teuchert, Jonathan
• Other Authors: Andreas Kaab, Jon-Ove Hagen, Jack Kohler, Doug Benn
• Format: Master Thesis
• Language: English
• Published: 2007
• Subjects: geomatikk geofysikk fjernmåling terrengklassifikasjon Svalbard breer morener isbreer; VDP::450; Arctic; Brøggerhalvøya; Norway; Svalbard; Arctic Archipelago; Climate change; glacier; Norwegian Polar Institute; Spitsbergen
• Online Access: http://hdl.handle.net/10852/12432; http://urn.nb.no/URN:NBN:no-16021

Abstract This study uses multispectral Landsat satellite imagery and a high resolution Digital Elevation Model based on 1990/95 aerial stereoimagery (Norwegian Polar Institute) to map moraines and glacier terminus positions on the high Arctic archipelago of Svalbard, Norway. To accomplish the task, this methodological Master thesis investigates different approaches of classification and incorporation of ancillary elevation data. Glaciers and other ice masses are important parts of the climatic system and also eminent terrestrial indicators for climate change in the scale of tenths to thousands of years. In the context of climatic response and global sea-level rise, mass balance studies on glaciers are important means to quantify trends however they are laborious and only for a few glaciers worldwide available. The application of remote sensing may help to determine changes in volume and extent of glaciers with a greater spatial coverage. Furthermore, area-wide mapping of unvegetated moraines on Svalbard may aid to quantify past glacial dynamics, especially since the Little Ice Age (LIA). During this study a thematic map of vegetation- and glacier-cover dating to the period from 1999 to 2002 has been produced by contrast-enhancing-methods and supervised determination of class thresholds, covering almost the complete archipelago. ISODATA classifications of combined terrain statistics and primary topographic attributes are merged with multispectral data to map the extend of moraines. Combinations of VIS, NIR and SWIR spectral bands with elevation range and standard deviation, curvature and slope deliver the most promising results when joined to neighbourhood analysis. Datasets from principal component analysis deliver good results as well. However, since the morphological diversity is high, no method is superior on all moraines. Multitemporal change detection was tested for a small number of glaciers on Brøggerhalvøya, western Spitsbergen, where recent data was acquired in the field during 2006. High variation ...