Empirically Constrained Estimates of Alaskan Regional Net Ecosystem Exchange of CO2, 2012-2014

We present data-driven estimates of the regional NEE of CO₂ across Alaska for three years (2012-2014) derived from CARVE (Carbon in the Arctic Reservoirs Vulnerability Experiment) aircraft measurements. We couple the Polar Weather Research and Forecasting (PWRF) model with the Stochastic Time-Inverted Lagrangian Transport (STILT) model, to produce footprints of surface influence that we convolve with a remote-sensing driven model of NEE across Alaska, the Polar Vegetation Photosynthesis and Respiration Model (PolarVPRM). For each month we calculate a spatially explicit additive flux (ΔF) by minimizing the difference between the measured profiles and the modeled profiles, using a framework that combines a uniform correction at regional scales and a Bayesian inversion of residuals at smaller scales. A rigorous estimate of total uncertainty (including atmospheric transport, measurement error, etc.) was made with a combination of maximum likelihood estimation and Monte Carlo error propagation. Integrating the optimized estimate of NEE (PolarVPRM + ΔF) over the domain for each year, we find that the Alaskan region was a small sink of CO₂ during 2012 and 2014, but a significant source of CO₂ in 2013, even before including emissions from the large forest fire season during 2013. Strong emissions in late spring and fall in 2013 accounted for much of the increased emission. Our optimized fluxes are consistent with other measurements on multiple spatial scales, including CO₂ mixing ratios from the CARVE Tower near Fairbanks and eddy covariance flux towers in both boreal and tundra ecosystems across Alaska. Our empirically constrained estimates of regional Alaskan NEE provide a baseline for comparison with future measurements of Alaskan NEE and to help assess changes in the high latitude carbon cycle due to climate change.