We propose to continue the activities of the OCO-2 flux inversion theme group global CO2 flux inversion model intercomparison project (MIP) for an additional three years, to help interpret the…

This is a $397,378.80 grant awarded by the National Aeronautics and Space Administration (NASA) to Colorado State University (CSU) to continue the activities of the OCO-2 Flux Inversion Theme Group's global CO2 flux inversion model intercomparison project for an additional three years. The project aims to help interpret the scientific content of new updates to the Orbiting Carbon Observatory 2 (OCO-2) and future OCO-3 CO2 data products, and validate them against in-situ measurements and flux inversions. CSU will manage the work of the OCO-2 Flux Inversion Theme Group and the inversion model intercomparison project, contribute their own global CO2 flux inversion results, and analyze the data to identify and mitigate potential biases between the satellite retrievals and in-situ measurements. This grant does not have a set-aside designation.

We propose to continue the activities of the OCO-2 flux inversion theme group global CO2 flux inversion model intercomparison project (MIP) for an additional three years, to help interpret the scientific content of new updates of the OCO-2 CO2 retrievals (V8 etc.), and the acos gosat and OCO-3 CO2 retrievals when available. Over the past couple years, this MIP has helped to disentangle the scientifically meaningful information in the V7 OCO-2 retrievals from retrieval biases and model errors. IT has provided a􀁆top-down􀁆 approach for validating the OCO-2 XCO2 product that complements the bias correction work of the OCO-2 validation team: in situ CO2 measurements are assimilated by the flux inversions, and the optimized CO2 mixing ratio fields are sampled at the location of the valid OCO-2 retrievals (with the proper vertical weighting) and compared to the retrievals themselves. Any systematic patterns in the differences could indicate new information contained in the satellite data or biases remaining in the retrievals􀁆 the analysis provided by this approach can help distinguish the two. Because the differences could also be due to transport model and inversion method errors, we do the comparisons across the full ensemble of models to mitigate those impacts. This approach has been useful in bias correcting the V7 OCO-2 retrievals and should be particularly useful moving forward as retrievals taken over the oceans are used more: the current approach of using the land-based tccon data to bias correct the oceans may not be adequate. We will generate the data products needed by the MIP: in situ CO2 measurements processed with a reduced lag time, with associated model-data-mismatch uncertainties; OCO-2 and OCO-3 x CO2 data and uncertainties averaged to appropriate scales for use in the flux inversions (e.g. 10-second averages); and tccon and aircraft data packaged for validation purposes. We will manage the work of both the OCO-2 flux inversion theme group and the inversion MIP. We will contribute our own global CO2 flux inversion results to the MIP, updating them as the new data become available. We will analyze our results (and the results of the rest of the group, as available) to look for absolute biases between the satellite retrievals and the in situ data, and relative biases between the different modes of satellite data (land nadir, land glint, ocean glint, etc), and will TRY to characterize and remove these. We will then invert the adjusted satellite retrievals and examine science issues particularly susceptible to retrieval biases: the meridional distribution of the annual mean ocean fluxes, the land/ocean breakdown of the fluxes, especially in the tropics/south, and the interannual variability of both. We will investigate the importance of the prior ocean flux fields to the final estimates. In particular, we will examine the possibility (suggested by recent soccom float data) that the southern ocean may be taking up ~1 PGC/YR less carbon than previously thought. Perhaps we have been constraining our inversions tightly to an incorrect, strong southern ocean uptake and have been underestimating the uptake by the tropical/ southern land biosphere as a result? Finally, we will examine the extent to which, by bias correcting and assimilating the straight pressure-weighted column (XCO2), we are throwing away valuable information in the lower portion of the atmospheric column that is of the most use for constraining regionalscale fluxes. We will calculate new CO2 column averages (scalars like XCO2, but weighted closer to the surface) from the current 20- level retrieved profiles, bias-correct them against our suite of in situ-constrained model fields using the same approach outlined above for XCO2, and use them in our flux inversions to see how the flux distribution (especially land VS. Ocean) changes and whether better agreement with independent CO2 data is obtained