Incorporation of landscape disturbance in probabilistic upscaling of CO2 fluxes and associated uncertainties

Background/Question/Methods

The contribution of landscape disturbance, relative to patterns in land cover types or climate, on terrestrial-atmosphere CO₂ exchange is unknown.  New approaches are needed that integrate disturbance patterns into diagnostic models to forecast their impact on landscape carbon fluxes and associated uncertainty.  Here, we leverage previous data-model fusion efforts to map mean CO₂ flux and associated uncertainty by including estimates of forest stand age derived from historical Landsat imagery.  We ask whether inclusion of this data improves yearly to decadal terrestrial CO₂flux hindcasts compared to previous efforts based on land cover type and climate.

Results/Conclusions

Initial results indicate that parameter and prediction uncertainty varies with stand age.  Extrapolated to the landscape scale, maps of mean CO₂ flux and associated uncertainty are critical for meeting climate change mitigation and adaptation efforts that aim to prioritize disturbance activities in the region. However, our work also highlights key challenges.  The characterization of uncertainty reflects only a small slice of this uncertainty spectrum and it is clear that interactions between stand age and disturbance type are likely critical in this landscape; however, towers represented only a subset of conditions present in the broader landscape, hindering model-data fusion efforts. Yet, despite these short-comings, our work represented an important portal for science-management communication of terrestrial carbon science and results highlight the importance of including management activities and observed data into regional terrestrial biosphere models.