Classification of central Siberian forest types by combining interferometric radar remote sensing (ERS Tandem Mission, JERS), topographic data and ecophysiological information

Boreal landscape modelling of hydrological processes and carbon and nitrogen exchange requires an adequate spatial database of topography, edaphic conditions and vegetation cover. Spatial variability in boreal vegetation types and edaphic factors depends largely on topography and geomorphology. One aspect of the authors´ work is the development of an a priori landscape classification scheme which is based on (a) topographic information extracted from interferometric and conventional digital elevation models (DEM) and (b) knowledge of ecophysiological requirements of major vegetation types. Based on this information classes of soil types and correlated potential vegetation will be defined. This a priori landscape classification is then checked by a SAR remote sensing classification based on radar backscatter and coherence which detects structural discrepancies between predicted and current land cover (e.g. open areas vs. forest due to fire disturbance or non-climax vegetation). Both, ERS-Tandem data and JERS data are used. Enhancement of the SAR based vegetation classification is expected from matching the SAR classification with the a priori landscape classification. As a next step the enhanced functional-structural landscape and vegetation classification may be used for upscaling fluxes and assessing pool sizes on a regional and continental scale