Vol. 5, No. 1, 100-110, 2006

Abstract
Ongoing climatic change on the Antarctic Peninsula is well documented in numerous scientific publications. Nevertheless, spatial and temporal distribution of glacier changes on the Antarctic Peninsula remain largely unknown. To analyse regional patterns and trends of changes in snow cover parameters on the Antarctic Peninsula north of 70° S, we propose a geostatistical approach based on the analysis of data derived by remote sensing methods. A time series of radar (ERS-1/2 SAR, Envisat ASAR and Radarsat SAR) and optical data (Landsat TM, Landsat ETM+ and Terra ASTER) enables the compilation of a multi-temporal database of glacier parameters for the last two decades. In this paper, we propose a method to facilitate the pre-processing and classification of remotely sensed data from several sources, and describe a sampling method used for selecting a representative set of glaciers for evaluation of classification algorithms. The so-called centreline approach facilitates pre-processing and classification of remote sensing data for monitoring glacier zones. In addition, the approach enables the integration of data from different sources, and minimizes the need of high resolution digital elevation models (DEM) for terrain correction.

ISSN 1729-3782