Vegetation mapping on the Midtre Lovenbreen, Svalbard using remote sensed images

Abstract

With changes in the ecosystem caused by the on-going climate change, biodiversity in the respective regions of the ecosystem have also been subject to change. In particular, as vegetation in the Arctic region has shown extra sensitivity to the effects of climate change, it must be continually monitored as indicators of climate change. However, there are lots of obstacles to conduct continuous vegetation monitoring in detail through field work in the Arctic region. For this reason, satellite data was developed to allow research in the regions inaccessible to field work. The objective of this study is to construct the vegetation map of the Midtre Lovenbreen, in Svalbard, Norway and to observe long-term variation of vegetation using remote sensing images. The area of study is the glacier foreland of Midtre Loven (79°N, 12°W) and covers an area of about 10 km2. 2007 Formosat-2 satellite image and 2013 Pleaides 1A were used in this study. Both of these satellite images are high-resolution and as both carry NIR band, are well-suited to make observations of vegetation states. While moss and lichen are dominant in the Arctic region, observation and analysis using satellite imaging is difficult due to their small and sparse distribution. To improve limits in spatial resolution, several methods were used: first, "Minimum Noise Fraction" (MNF) was used to minimize the dimensionality and noise; second, the Pixel Purity Index (PPI) was used to collect the endmember; last, Spectral Mixture Analysis (SMA) was used to confirm the abundance of vegetation. These results will be confirmed by comparing the field data acquired in July 2014 and a vegetation map from Moreau et al(2005).