Characterization of aboveground biomass in an unmanaged boreal forest using Landsat temporal segmentation metrics

摘要

Canada is dominated by forested ecosystems which are subject to various inventory and management practices, with more northern boreal forests subject to neither. Our objectives were to measure the capacity of temporal trajectory metrics for estimating selected forest attributes in a northern Canadian boreal forest context using Landsat imagery and investigate the importance of different types of temporal trajectory metrics. Results indicated that Wetness was the best Tasseled Cap (TC) component for above-ground biomass estimation (R~2 = 50%, RMSE% =56%), and the combination of simple and complex metrics from all TC components produced the highest R~2 (62%) and lowest RMSE% (49%). Using a similar combination of variables, other forest attributes were estimated equally reliably with lower RMSE% values. The most important temporal trajectory metrics were simple and described TC component values at each point of change in the temporal trajectory, however the most important variables overall were environmental variables.