Data-driven Diagnostics of Terrestrial Carbon Dynamics over North America (Invited)

Accurate quantification of net carbon dioxide exchange at regional to continental scales would improve our understanding of feedbacks between the terrestrial biosphere and the climate system and would be a substantial aid in climate policy-making. There is general agreement that the terrestrial ecosystems in North America provide a large carbon sink, although the size and distribution of this sink remain uncertain. More importantly, the interannual variability of this sink and the dominant controlling factors are not well understood. Here we developed data-driven models to estimate carbon fluxes - gross primary productivity, net ecosystem carbon exchange (NEE), and ecosystem respiration by combining eddy flux measurements, MODIS data (e.g., enhanced vegetation index, land surface temperature, water index), and micrometeorological data. The main innovations in our approach are the use of continuous NEE measurements from eddy covariance flux towers and the use of aboveground biomass as an explanatory variable that accounts for the effects of disturbance and stand age. The addition of aboveground biomass improved the performance of the models by 5-10%. We used the predictive models to produce continuous flux estimates with high spatial (1km) and temporal (8-day) resolutions across the continent over the period 2000-2009. We then used our continuous flux estimates to assess the magnitude, distribution, and interannual variability of ecosystem carbon exchange over North America. Ecosystem carbon exchange exhibited large spatial and temporal variability. The dominant sources of interannual variation of the North American carbon sink include extreme climate events (e.g., droughts) and disturbances (e.g., wildfires, hurricanes). Droughts and disturbances reduced net ecosystem carbon uptake or resulted in carbon release at regional scales.