Surface CO2 fluxes implied by a full year of OCO-2 column CO2 measurements

Abstract:

Over one year of full-column CO₂ concentration data is now available from the Orbiting Carbon Observatory (OCO-2) satellite, with retrieval biases corrected using upward-looking solar spectrometer data from the TCCON network as well with internal consistency checks. We use this OCO-2 data to estimate weekly surface CO₂ flux corrections at 6.7ºx6.7º resolution with a variational data assimilation technique built around the off-line PCTM atmospheric transport model driven with MERRA 1ºx1.25° winds and mixing parameters.

Since such flux estimates can depend strongly on the prior fluxes assumed (which may remain unchanged in regions of sparse sampling), the initial 3-D concentrations assumed (especially in the upper part of the atmosphere), vertical transport/mixing errors in the model, and un-corrected biases in the satellite data, we invert the OCO-2 data in multiple inversions in which different prior fluxes are used (e.g. SiB4 vs. CASA land bio, Takahashi vs. Doney ocean, FFDAS vs. CDIAC fossil fuel), in which ACOS GOSAT data and NOAA surface in situ and aircraft profile data are used (or not) to correct the prior fluxes and concentration fields, and in which the vertical mixing in the transport model is artificially increased/decreased by a factor of 3, to assess the sensitivity of the OCO-2 flux corrections. These inversions are done in the context of a longer span (2009-2015) to allow the impact of the fluxes and other data sources to fully impact the upper layers of the model.

The bias between the OCO-2 data and the prior forward CO₂ fields is also calculated before doing the inversions, and compared to similar retrieval biases solved for the ACOS GOSAT data (B3.5). The impact of these bias corrections, as well as the standard ones provided by the OCO-2 team, is assessed by comparing the fit of the a posteriori CO₂ fields to independent data (including surface in situ and NOAA aircraft).