Deriving Land Cover Information over Siberia: Comparing Results from MERIS and MODIS

With the availability of space-borne data from medium resolution optical sensors such as MERIS and MODIS, deriving land cover information of large areas is expected to improve considerably. Both, the higher spatial and spectral resolution compared to previously available data from AVHRR seem to be far more suited for detailed mapping of land cover. Within the Siberia-II project (EU contract EVG2-2001-00008) land cover information is a key parameter in driving vegetation models that help to fulfil the aim of full greenhouse gas accounting over Siberia. Due to the regional approach of the project MERIS and MODIS data have been selected for land cover mapping. An automated processing chain that is very similar for both datasets has been developed and implemented to pre-process and classify the vast amount of imagery available to the project. In this paper focus is given to the comparison of the land cover maps derived from MERIS and MODIS data. As part of this analysis the thematic accuracy achieved will be reviewed with respect to ground data and existing global land cover data (e.g. GLC2000). Furthermore, practical issues related to data processing and product quality as well as experiences gained while processing these two data sets shall be addressed from a user’s point of view.