Did a Tree Really Fall in the Forest? Incorporating Uncertainty Into Regional-Scale Monitoring of Forest Growth and Mortality Processes.

Although forests and woodlands are important reservoirs of carbon,  the net carbon flux from these lands at regional scales is highly constrained by disturbance and subsequent regrowth processes.  Earth System Models that do not realistically incorporate these dynamics will not stand the scrutiny of science or the public, and yet current approaches to monitoring carbon dynamics in forests have lacked the spatial or temporal detail needed to train and test models.  Moreover, those maps that do exist to monitor forest dynamics have been hampered by an inability to reliably characterize methodological uncertainities. 

Here, we describe an approach to link forest plot data (FIA;  Forest Inventory and Analysis), environmental data, satellite-imagery, and aircraft-based lidar acquisitions to create a full assessment of carbon dynamics in forests.  We incorporate methodological uncertainties introduced by our satellite processing, by our base multivariate statistical technique, and by the choice of allometric equation used to the convert estimated plot-measurements to site-level biomass and carbon. 

Using the West Cascades province of Oregon and Northern California as a test case, we  show that the combined impacts of methodological uncertainty vary both across forest types and over time, and may be particularly acute in early ecological succession.  These patterns of uncertainty can not only help guide managers in their interpretation of mapped carbon changes on the landscape, but may provide a means of prioritizing future improvements in ground, aircraft, or satellite missions.