Afforestation, Re-, and Deforestation monitoring in Siberia - accuracy requirements and first results

Work on Afforestation, Reforestation and Deforestation (ARD) mapping and monitoring in Siberia is presented. The accuracy requirements of a satellite based ARD data product to be useful for an independent verification of the Kyoto protocol in the first commitment period (2008-2012) are discussed. Mapping of ARD is important for national carbon emission inventories related to land use change as specified in the Kyoto protocol. The methodological requirements for remote sensing derived ARD information is outlined. The overall objective is to create a remote sensing based ARD classification system based on multi-temporal Earth observation data, multi-date forest inventory information and land cover information of test territories in Siberia. Historical (1989) and near present (2000-2003) high resolution optical satellite data (Landsat TM-5 and ETM) are pre-processed and a two-date change detection method using artificial neural networks (ANNs) is tested to derive specific ARD classes. Forest inventory information from Russian forest enterprises is used as ground truth. First results indicate that ANNs can be used successfully for ARD monitoring.