Multi-Scale Vegetation and Water Body Mapping of the Northern Latitudes in Siberia with Optical Remote Sensing

Abstract

The Arctic ecosystem is highly vulnerable to modification within the climate system. Increasing greenhouse gas emissions, e.g. from melting permafrost soils, are assumed to result in positive feedback mechanisms within the global climate system. Water bodies and the seasonal dependent freeze/thaw dynamics of the uppermost permafrost layer (active layer) are the major carbon and methane source in the Arctic regions. Impacts induced by climate change are resulting in the transformation of the existing Arctic landforms, such as the destruction of settlements or changes in the terrain, hydrology and vegetation cover. This paper presents the potential of using Earth observation data from various sources and time steps to monitor land cover characteristics and changes in the Arctic regions. Information on vegetation structure types and physiognomy is commonly incorporated into spatial models predicting the permafrost distribution. The MODIS land cover, the GlobCover land cover map, SYNMAP and MODIS VCF (vegetation continuous field) have been combined in a product describing the fractional vegetation cover. The dataset, with a spatial resolution of 1 km, consists of four layers providing percentage cover information for trees, shrubs, herbaceous areas and barren areas. Additional information, such as the CAVM (Circumpolar Arctic Vegetation Map), has been integrated into the harmonization approach. Local land cover and water body changes have been analyzed using high spatial resolution earth observation information from Landsat, RapidEye and Corona Keyhole. This analysis was carried out for a test site in central Yakutia and the Lena river delta system in Siberia, Russia. High-resolution land cover information was mapped using an object-oriented classification approach. Object characteristics, such as the shape, spectral properties and information within different hierarchical object levels, are utilized to identify individual vegetation class properties for assignment to a thematic class. Water body changes are identified using historical earth observation data from the 1960s and recent RapidEye data.