The estimates of above-ground forest biomass (AGB) in Northern Eurasia are highly uncertain, despite the global importance of AGB for ecosystem services and its role as carbon stores. In this paper, we demonstrate the potential of ALOS/PALSAR (Advanced Land Observing Satellite/Phased Array L-band Synthetic Aperture Radar), for the estimation of AGB in the range of 10-190tons (dry matter)/ha in mixed and deciduous forests at the southern edge of the boreal region in Western Siberia. Various regression models were tested to determine the relationship between forest biomass derived from field measurements and radar backscatter. The best results were obtained using HV-polarized backscatter with the Water Cloud model, giving estimation errors in terms of root mean square errors (RMSE) between 25% and 32% of the mean biomass, and coefficient of determination (R2) between 0.35 and 0.49 for the whole range of SAR backscatter used in the analysis. The method displayed a higher prediction accuracy with RMSE of 15%, and the R2 between 0.55 and 0.72 when restricted to SAR backscatter (σ0)<-12.6dB where the model was clearly defined. The SAR-based estimates offer a potential of rapid, high resolution and low cost mapping of the lower biomass woody vegetation (sparse or young forests on shallow peat) in Siberia, the area where more accurate national or large scale forest inventories hardly exist.
ASJC Scopus subject areas