Inverse Modeling of CO2 Fluxes Using O-Buoys, a Multi-Year Dataset of Surface Observations from the Arctic Ocean

The Arctic is undergoing drastic change due to warming and rapid loss of sea ice, both of which affect biogeochemical cycles. It is unclear if or how these changes will alter the air-sea CO₂ fluxes. At the same time, the current global CO₂ monitoring system has low sensitivity over the Arctic Ocean. Arctic surface measurements are almost all over land and a very small fraction of satellite measurements have high-quality in the Arctic due to frequent cloud cover and other issues. To address these observing challenges, the O-Buoy network monitored CO₂ concentrations over the Arctic sea ice year-round from 2009-2016, providing a new dataset for understanding air-sea exchange of CO₂. We show that CO₂ concentrations simulated at the O-Buoy locations by the GEOS-Chem global atmospheric transport model have discrepancies that could be due to unresolved air-sea exchange or long-range transport. To discriminate between these possibilities, we optimize surface fluxes using 4D-variational data assimilation in GEOS-Chem. We perform an inversion with GOSAT and the ObsPack PROTOTYPE, and another inversion with O-Buoy observations, to assess the additional value and impact of surface observations over the Arctic Ocean.