Estimating oceanatmosphere carbon fluxes from atmospheric oxygen measurements

Estimating ocean-atmosphere carbon fluxes from atmospheric oxygen measurements Mark Battle (Bowdoin College) Michael Bender & Nicolas Cassar (Princeton) Roberta Hamme (U BC), Ralph Keeling (SIO) Cindy Nevison (NCAR) UNESCO Surface Ocean CO 2 Variability and Vulnerabilities Workshop April 12, 2007 Funding from: NSF, NOAA GCRP, BP-Amoco, NASA, UNESCO

On the agenda: • Oxygen, O 2/N 2, APO and Carbon Fluxes • What can (did) one do with APO? – – – The ancient past (Keeling, Stephens) The recent past (Gruber, Tohjima, Battle, Naegler) The present: Models (Nevison) The present: Data (Tohjima, Hamme) The present: Data & Models (Rödenbeck)

What determines the amount of O 2 in the atmosphere?

1 st order description of long-term fluxes

More detail on the oceans Seasonality…

1. 4 1. 1

More detail on the oceans Seasonality and secular trends

Atmospheric Potential Oxygen APO O 2 + 1. 1 CO 2 APO reflects air/ocean O 2 & CO 2 fluxes Land biota doesn’t change APO Fossil fuels change APO a little

O 2/N 2 & APO changes are small O 2/N 2 per meg (O 2/N 2 sa – O 2/N 2 st)/(O 2/N 2 st) x 106 1 per meg = 0. 0001% = 0. 001 per mil 1 Pg. C FF 3. 2 per meg O 2/N 2 0. 66 per meg APO 1 Pg. C into oceans 2. 5 per meg APO* *assuming no corresponding O 2 flux

What controls APO? • Ocean biology (light, nutrients, etc. ) • Ocean chemistry (O 2 & CO 2 equilibration) • Ocean temperature (solubility & stratification) • Ocean circulation (shallow & deep) • Atmospheric transport • Fossil fuel (a little)

A brief history of time APO

In the beginning…

Stephens et al. , 1998 Models don’t get interpolar gradient right (physics? ) Equatorial data would be nice.

The next chapter…

p. CO 2, dissolved O 2, PO 4 & heat fluxes of CO 2 and O 2 atmospheric transport atmospheric composition at observing stations

Gruber et al. , 2001 Eliminate OBGC Model. Results seem Independent of Ocean physics Interpolar gradient getting better Equatorial data Would be nice.

New data!

Tohjima, et al. 2005

Still more new data!

Princeton & Scripps data

Battle et al. 2006 Equatorial bulge is confirmed…

Battle et al. 2006 Equatorial bulge is confirmed… and the interpolar gradient looks good too.

…but it’s evolving Battle et al. , 2006

…and more modeling work

Naegler et al. 2006

Work in progress: Fresh ideas and fresh data…

C. Nevison (NCAR) work in progress, with S. Doney & N. Mahawold • Model study with OBGCM & ATM • Seasonal cycle comparison • Annual mean gradient comparison • Emphasis on quantifying transport errors

Ocean Ecosystem Model+OGCM (Doney) CASA land bio (also w/ fire) fossil fuel fluxes of CO 2 and O 2 MATCH (NCEP winds 1979 -2004) atmospheric composition at observing stations

C. Nevison (NCAR) work in progress, with S. Doney & N. Mahawold • • • Ocean O 2 and CO 2 fluxes from WHOI ecosystem model. Set of Carbon/O 2 fluxes with IAV in ocean, land transport, all NCEP driven MATCH has stronger rectifier than ATMs previously used (TM 3, TM 2, GCTM)

Seasonal results from WHOI/MATCH Nevison (in progress)

Fidelity of seasonal cycles Nevison (in progress)

Fidelity of seasonal cycles relative phasing mod/ obs Nevison (in progress)

Fidelity of seasonal cycles Nevison (in progress)

Fidelity of seasonal cycles Palmer Model skill depends on hemisphere Nevison (in progress)

latitudinal gradients in WHOI/MATCH Battle data Gruber ’ 01/MATCH Gruber/TM 3 ATM uncertainties trump OBGCM fluxes again Nevison (in progress)

Tohjima et al. (Tellus B, submitted) • Repeat carbon sink partitioning • Look at APO variability

Atmosphere-ocean partition Tohjima et al. (submitted)

Interannual variability in APO ~20 month smoothing Tohjima et al. (submitted)

Variability reflects O 2 fluxes; not carbon. 12 Pg C/yr? Of course not. ~20 month smoothing Tohjima et al. (submitted)

Atmosphere-ocean partition APO variability Tohjima et al. (submitted)

Further evolution of the interpolar gradient ~2 -year smoothing

• Spatial structure of APO is genuinely timedependent ~2 -year smoothing

• Spatial structure of APO is genuinely timedependent • Neighboring stations move independently ~2 -year smoothing

• Spatial structure of APO is genuinely timedependent • Neighboring stations move independently • Watch out for end-effects ~2 -year smoothing

Hamme, Keeling & Paplawsky (AGU, 2006) • Interhemispheric temporal variability • Mechanisms

Work of Hamme et al. will be available upon publication (expected in late 2007)

Broad conclusions: • APO reflects (too) many oceanic properties • Dataset is good but not great • Interpreting ocean models complicated by atmospheric transport

More detailed conclusions • A model combo can get seasonality right, but still miss annual averages • Some indications that winter ventilation plays a big role • Apparent global signal of NAM • May be an El Niño signal too

O 2 and CO 2 fluxes are related, but not intimately. The degree of linkage depends on temporal and spatial scale.