Testing model representation of seasonal air-sea exchange of O2 and CO2

Testing model representation of seasonal air-sea exchange of O2 and CO2

Scientific Motivation

The processes driving air-sea CO2 exchange are not understood well enough to support predictions of future atmospheric CO2 trends. Diagnosing ESM air-sea CO2 fluxes and their drivers is difficult owing to sparse ocean observations and competing thermal and biological influences. Atmospheric measurements provide larger scale constraints and adding O2 enables unique insights into the processes influencing air-sea CO2 exchange.

We now have a wealth of atmospheric O2 and CO2 observations, from the 5 HIPPO and 4 ATom global airborne surveys, as well as the ORCAS intensive Southern Ocean campaign, and shipboard and station records. The seasonal amplitude in column-mean atmospheric potential oxygen (APO = O2 + CO2) from HIPPO and ATom are surprisingly symmetric in each hemisphere, in disagreement with preliminary model comparisons. Further, measurements of O2:CO2 ratios in spatial gradients over the Southern Ocean from stations, ships, and aircraft reveal consistently negative values indicating biological dominance of seasonal CO2 exchange, but in disagreement with a subset of CMIP5 models. Having abiotic CO2 from the models will be particularly useful for evaluating the ability of observed O2:CO2 flux ratios to discern forcings.

Proposed Hacking

From available historical simulations, extract gridded monthly air-sea O2, CO2, and heat fluxes as well as SST and salinity fields. Calculate latitudinal distributions of seasonal net outgassing of APO and compare to similar quantities derived from the HIPPO and ATom data. Calculate O2:CO2 ratios over the Southern Ocean, evaluate spatial patterns and coherence, and calculate spatially integrated timeseries. Regridding all fields to a common 1 x 1 degree grid would be helpful, but may not be necessary. (The heat, SST, and salinity are required to estimate air-sea N2 fluxes to make a small correction to the O2 fluxes to match the observed atmospheric quantity which is deviations in the O2/N2 ratio). Compare bio vs thermal forcing ratio implied by O2:CO2 to that from model abiotic CO2 flux.

Anticipated Data Needs

Air-sea O2, CO2, and heat fluxes as well as SST and salinity fields from historical simulations.The last 20 years of the simulations will be the focus but earlier times may be useful for evaluating trends.

I put a request in for historical + Omon + fgo2, fgco2, fgco2abio, fgco2nat, tos, sos, hfds. Using hist-bgc would eventually probably be better but there is little output available presently. I only think 1 version of each unique ocean model and 1 ensemble member are needed.

Anticipated Software Tools

Jupyterhub, Python, xarray and/or R

What software libraries do you intend to use? Will your project involve developing new software?


Desired Collaborators

Anyone interested in seasonal air-sea exchange of O2 and CO2. It may be possible to team up with one of the other projects evaluating ocean carbon and oxygen.


Hi Britt! Great proposal, and lots of technical and science synergies with the O2 and Carbon/productivity projects! It would be interesting e.g. to see if there are any relations between models specific representation of seasonal carbon fluxes and their longterm projections. @plerner has some interesting ideas on the seasonal cycle:

Also interested in whether HIPPO & ATom can provide constrains on the net mean fluxes and their meridional profile (i.e. the equatorial APO bulge, etc.).

Agreed with Yassir that this is a great idea, and lots of clear synergy with the set of ocean BGC project he proposed! I’ve been thinking about synergies between the ocean BGC proposal and the Time of Emergence proposal from @darothen. Your project seems to me like a place where those intersections would be especially interesting because of the new opportunity to separate out biotic and abiotic air-sea CO2 flux. Scientifically, I think it would be quite interesting to separate out the time of emergence for fgco2, fgco2abio, and fgco2nat, providing some mechanistic explanation for changes in air-sea CO2 flux we could find observationally. Technically, it seems that the framework that the time of emergence team is planning to put together could easily be combined with the analysis you are proposing, providing a nice synergy between the two groups.

Hi Yassir and Hilary - thanks for your comments!

Yassir, indeed that would be a home run if the seasonal metrics observable today correlate with and can discriminate among different projections. Your point re: the mean meridional profile is also a good one. However, because the CMIP models do not carry O2 in their atmospheres, in the context of a 3-day hackathon, and even more generally, I have been trying to think of metrics that can be compared to ocean or land fluxes without the need for separate atmospheric transport simulations. The candidates I was thinking of were the NH:SH ratio of extratropical seasonal APO amplitudes (if N-S mixing in each hemisphere are similar then these should map to ratio of fluxes) and the O2:CO2 ratio of SO atmospheric gradients (as long as land and fossil influences are small, these should map to flux ratios). Can you think of something we can use for the mean meridional gradient? The Resplandy asymmetry metric would be a good one for interhemispheric exchange but I haven’t thought of one for the equatorial bulge. In any case, we can certainly compile a set of fluxes for blowing around an ATM later.

Hilary - I had not thought about time of emergence in this context yet, but I agree this would definitely be worth looking at. If it turned out that an impact from circulation changes could be seen in air-sea flux O2:CO2 ratios before in either dissolved O2 or pCO2, that would be really cool (and help motivate our measurement efforts).

I’ll try to pull some figures together in the next week that encapsulate the observations.

Thanks again,