1. Core Theme 1: surface fluxes WP 4: The North Atlantic observing system WP5: The Southern Ocean WP6: modelling and synthesis (WP7: mooring development; discontinued)

2. WP4: North Atlantic observing system Ongoing data collection and synthesis of fluxes

3. Jan-Feb Mar-Apr May-Jun Jul-Aug Nov-Dec Sep-Oct Data coverage in two-month intervals, 2005

4. Constructing basin-wide fluxes For each pCO 2 measurement, assign SST, mixed layer depth, (chlorophyll, position) Find pCO 2 as a function of these variables using –multiple linear regression –Neural network Interpolate to generate a pCO2 map for the whole area Calculate fluxes

5. Distance (degrees lat) to Nearest data point Jan-Feb Sep-Oct

6. MLR estimates of pCO2 field for each month using independent variables SST, MLD and longitude, in 20° latitude bands Jan MarApr Jun May Jul Aug Sep Oct Nov Dec JanFeb

7. 10 to 20 N 30 to 40 N 50 to 65 N JFM AMJ JAS OND -0.02 0 0.02 0.04 0.06 Flux (Gt/ yr) 2005 fluxes by season and latitude band

8. CO 2 fluxes into the North Atlantic – 1994/5 to 2005 (Wanninkhof 1992 gas exchange) Takahashi 1995 climatology Carbo-ocean 2005 Cavassoo – era (approx)

9. What is the uncertainty on pCO 2 sw ? Calculate semi-variograms of the residuals from the multiple linear regressions Where –r are the residuals from the model –N(h) is the set of pairs of observations i, j, that are a separated by a distance h.

10. pCO2 variance (μatm 2 ) Uncertainty continued… Separation (Degrees of latitude) Decorrelation length scale Variance of decorrelated data

11. Uncertainty continued…. The number of degrees of freedom in the estimation problem is n ~A/x 2 where A is the area of the North Atlantic and x is the decorrelation length obtained from the semi-variograms. An upper limit on the variance, σ 2, of pCO 2 about the regressions is given by the sill (amplitude) of the variograms. The 1-σ uncertainties on mean pCO 2 for the region as a whole is given by = ~0.8 μatm using the summer variogram, or 1.5 μatm using the winter variogram. These represent ~6% or ~10% errors on ΔpCO 2 (air-sea difference). A more careful calculation gives overall 1-σ uncertainty of 6% on fluxes from this source.

12. N.Atlantic air-sea fluxes with unprecedented accuracy and resolution We obtain pCO 2 for the entire region (10°-65°N) by relating CO 2 to SST, mixed layer depth, using regressions and/or neural nets Fluxes are derived with ~10° spatial resolution and seasonal-to-monthly time resolution. The precision is 6% (1-σ) for the NA annual flux – (calculated using geostats theory or data gap analysis) -- much better than for any other comparable region in the world. In combination with atmospheric inversions should enable greatly improved estimates for adjacent land sinks.

13. WP5: Southern Ocean air-sea CO 2 fluxes: From Weddel sea to open ocean areas N. Metzl, OISO10, Jan 2003

14. CARIOCA measurements Synthesis of 65 months of CARIOCA measurements: - SAZ strong sink (-0.8PgC yr -1 ) contrary to PZ (-0.1PgC yr -1 ) –Seasonal variation of DIC in SAZ (~30  mol kg -1 ) –North-South increase of DIC and fCO2 in winter between Subtropical front and Subantarctic front Boutin et al, submitted, 2007

15. Trend atmosphere: + 1.72 µatm/yr Trend ocean: + 2.11 µatm/yr Ocean sink decreases ? (Metzl, 2007 sub.) Ocean pCO 2 trend between 1991 and 2007 Atm. trend SummerWinter

16. (December 2002-January 2003) AWBAntarctic ice shelf Ridge Rapid transition from ice covered CO 2 –rich waters to a biologically mediated sink in the eastern Weddell Gyre 0°W20°E 24 December 2002 Ice Cover (blue=open water) (Bakker and Hoppema, in prep.)

17. Neural nets: Poster, Maciek Telszewski and colleagues

20. IFM-GEOMAR

21. CO 2 Gas Exchange Rates from Inversion of Water Column Data R. Schlitzer, J. Schneider, Alfred Wegener Institute, Bremerhaven 1. Pre-industrial, annual mean CO 2 Fluxes 2. Seasonal CO 2 Fluxes Approach: Use tracer calibrated model (CFC, radiocarbon, etc.; Schlitzer, 2007)Use tracer calibrated model (CFC, radiocarbon, etc.; Schlitzer, 2007) Determine CO 2 fluxes by fitting to ocean interior DIC dataDetermine CO 2 fluxes by fitting to ocean interior DIC dataResults: Outgasing in tropics, ingassing in subtropics. As in Takahashi (1999) and other studies.Outgasing in tropics, ingassing in subtropics. As in Takahashi (1999) and other studies. Net outgasing in Southern Ocean (not seen in Takahashi, 1999 or OCMIP II models): 0.6 PgC yr -1 south of 50°S.Net outgasing in Southern Ocean (not seen in Takahashi, 1999 or OCMIP II models): 0.6 PgC yr -1 south of 50°S. Near-zero pre-industrial interhemispheric C transport by ocean.Near-zero pre-industrial interhemispheric C transport by ocean. Takahashi 1999 Investigation of the effect of varying strength and timing of biological production on monthly CO 2 fluxes See poster of Judit Schneider.

22. Assimilation of pCO2 data into an obgc model: progress Assimilation of pCO2 surface data into Met office FOAM model using Hadley centre OBGCM is ongoing. Operational system assimilates SST, SSH, ARGO… Assimilation of ocean colour with biogeochemical code Additional nudge of DIC/ALK using ocean pCO2 data. The system in testing phase but work delayed by other priorities in 2007. It is planned to re-start the work in the first few months of 2008.

23. Surface fluxes: progress High precision estimates of North Atlantic flux. Progress in characterizing the Southern Ocean using measurements and models: observations of a decreasing flux? New techniques for constraining fluxes: neural nets, ocean inversions Progress in assimilating carbon measurements into ocean biogeochemical models.