Ocean Constraints on the Atmospheric Inverse Problem: The contribution of Forward and Inverse Models

Transcription

1 Ocean Constraints on the Atmospheric Inverse Problem: The contribution of Forward and Inverse Models Nicolas Gruber Institute of Geophysics and Planetary Physics & Department of Atmospheric Sciences, University of California, Los Angeles Ackowledgements Manuel Gloor 1, J. L. Sarmiento 2, C. Sabine 3, R. F. Feely 4, J. C. Orr 5 and the OCMIP members (1) Max Planck Institute for Biogeochemistry, Jena, Germany (2) Program in Oceanic and Atmospheric Sciences, Princeton University (3) JISAO, University of Washington, Seattle (4) Pacific Marine Environmental Laboratory, Seattle (5) Lab. des Sciences du Climat de l Environment, Gif-sur-Yvette, France

2 Outline 1. Introduction: A short historical overview 2. Observational approaches 3. Forward Modeling 4. Inverse Modeling 5. Regionalization 6. Summary and Conclusion

3 INTERHEMISPHERIC GRADIENT OF ATMOSPHERIC CO 2 Fan et al. (1999)

4 Interhemispheric ocean CO 2 transport [Keeling et al., 1989]

5 difference [ppm] Mauna Loa - South Pole CO PREINDUSTRIAL INTERHEMISPHERIC CO 2 GRADIENT -0.8 ppm at zero emission Global Fossil CO 2 Emission [PgC/yr] adapted and updated from Keeling et al. (1989)

6 2

7 OBSERVED AND PREDICTED ATMOSPHERIC CO 2 Tans et al. (1990) a) fit to data b) predicted CO2 concentration with fossil fuel emission, sesonal vegetation tropical deforestation, and ocean flux computed using LM gas transfer coefficient c) as b, except for ocean flux computed using Tans et al. gas transfer coefficient d) as b, except for ocean flux computed using previous estimates

8

9

10 OCMIP-1: CO 2 FLUXES AND OCEAN TRANSPORT Sarmiento et al. (2000)

11 2 AND A! " $#&%'('*)

12 OCEAN CO 2 TRANSPORT DIENT!#" $$&%

13 WOCE/JGOFS CO 2 survey 80N 40E 80E 120E 160E 160W 120W 80W 40W 0 o 80N 40N US WOCE NOAA Foreign 40N 0 o 0 o 40S 40S 80S 40E 80E 120E 160E 160W 120W 80W 40W 0 o 80S

14 DISSOLVED INORGANIC CARBON [µmol/kg] ATLANTIC SOUTHERN OC. PACIFIC 60 N 30 N Eq 30 S 60 S 60 S 30 S Eq 30 N 60 N Iceland Mid-Atl. Ridge Hawaii Aleutian Isl.

15 What C gas ex can tell us about the air-sea gas exchange of CO 2 Definition of C gas ex : C gas ex = So S Γ( C gas ex ) = 0. ( DIC r C:P P 1 ) 2 (Alk + r N:PP) C ant, Explanation and Interpretation : The normalization to constant Phosphate (P) and Alkalinity (Alk) removes the contribution of the soft-tissue and carbonate pumps. The remaining variability is due to the exchange of natural CO 2 between the ocean and atmosphere as well as the uptake of anthropogenic CO 2 ( C ant ). After removing C ant as well, we end up with a tracer just reflecting the air-sea exchange of natural (i.e. preindustrial) CO 2.

16 GAS-EXCHANGE COMPONENT ( Cgas-ex) [µmol/kg] ATLANTIC SOUTHERN OC. PACIFIC

17 ANTHROPOGENIC CO 2 [µmol/kg] ATLANTIC SOUTHERN OC. PACIFIC 60 N 30 N Eq 30 S 60 S 60 S 30 S Eq 30 N 60 N Iceland Mid-Atl. Ridge Hawaii Aleutian Isl.

18 Principle of Oceanic Inversion The ocean surface is partitioned into n regions. Basis functions Steady-State Inversion In a OGCM, constant fluxes of dye tracers (Φ) are imposed in each of the n regions, and the model is run until the spatial patterns of the dyes attain a quasi steady-state. Transient-tracer Inversion Analogous to the steady-state inversion except that the dye fluxes are time-varying, i.e. for CO 2, Φ(t) = Φ(t = 0) (pco 2 (t) pco 2 (t = 0)) The model predictions of the dye concentrations are sampled at the observation stations and arranged as a vector χ OGCM. The model therefore provides us with a transport matrix A OGCM that relates the fluxes to the distribution, χ OGCM = A OGCMΦ. Modeled distributions at the observations stations are substituted with observed ones and the matrix A is inverted to get an estimate of the surface fluxes ( Φ est ) : Φ est = A 1 OGCM χ obs.

19 The Models coarse resolution model: 3.75 meridionally, 4.5 zonally and 24 layers vertically seasonal forcing at the surface with observed wind-stress, [Hellermann and Rosenstein, 1983] and heat and freshwater fluxes [DaSilva et al., 1994] with weak restoring towards observed temperature and salinity fields from Levitus et al. [1994]. Sub-grid scale parameterization of eddies according to Gent and McWilliams [1990] KvLOW-AiLOW: Low vertical diffusivity everywhere; KvHISouth- AiLOW: Enhanced vertical diffusivity in the Southern Ocean. Surface ocean has been divided into 15 regions and unit fluxes have been applied for a total runtime of 3000 years. For the time-dependent inversion, the unit fluxes were scaled in time according to the evolution of the atmospheric CO 2 perturbation.

20 OCEAN REGIONS AND NETWORK N N Atl N N Pac Temp N Pac Temp N Atl Eq 12Indian Eq Pac Eq Atl Temp S Indian Temp S Pac Temp S Atl SubPol S Pac & SubPol S Ind SubPol S Atl Southern Ocean WOCE/JGOFS/OACES SURVEY

21 Basis Functions: Color Tracer Distribution

22 INVERSE AIR-SEA CO 2 -FLUXES Pre-industrial CO 2 flux Anthropogenic CO 2 Flux Positive: Flux out of ocean CO 2 Fluxes [Pg C/yr] Southern Ocean SPol S Atl Total CO 2 Flux SPol S Pac & SPol S Ind Temp S Atl Temp S Pac Temp S Indian Eq Atl Eq Pac Eq Indian Temp N Atl Temp N Pac N N ATl N N Pac <58 S 58 S - 36 S 36 S - 13 S 13 S - 13 N 13 N - 53 N 13 N - 36 N 53 N - 90 N 36 N - 62 N Anthropogenic CO 2 Flux: 1.8 PgC yr -1 Gloor et al. (submitted), Gruber et al. (in prep.)

23 Contemporary CO 2 Fluxes [Pg C/yr] PRESENT-DAY AIR-SEA CO2-FLUXES (1990) Inversion (KVLOW(h)) Forward Model KVLOW (1990) Forward Model KVHIGH (1990) TRANSCOM mean T02(u2) Wanninkhof et al. (2001) T02(u3) Wanninkhof et al. (2001) Positive: Flux out of ocean Southern Ocean SPol S SPol S Pac Atl SPol S Ind Temp S Atl Temp S Pac Temp S Indian Temp N Atl Temp N Pac N N ATl N N Pac Eq Atl Eq Pac Eq Indian Present-day Fluxes <58 S 58 S - 36 S 36 S - 13 S 13 S - 13 N 13 N - 53 N 13 N - 36 N 53 N - 90 N 36 N - 62 N Gloor et al. (submitted) Gruber et al. (in prep.)

24 PRE-INDUSTRIAL AIR-SEA FLUXES AND TRANSPORT Eq Ind Temp S Ind Temp N Pac N N Pac Eq Pac Temp S Pac SPol S Pac & SPol S Ind Red numbers: Air-Sea Fluxes in [Pg C/yr] Blue numbers: Ocean Transport in [Pg C/yr] NNAtl Temp N Atl Temp S Atl SPol S Atl Eq Atl Southern Ocean Transport across Equator: 0.4 PgC yr -1 Transport vanishes at about 3 S

25 OCEAN INVERSION CO 2 Fluxes [Mmol C yr -1 m -1 ] Preindustrial Ocean Fluxes Inversion KVLOW(h) AILOW+KVLOW AIHIGH+KVHIGH Inv. KVLOW(h) 20 region a atmospheric CO 2 [ppm] Preindustrial Atmospheric CO 2 c TM3 GCTM 80 o S 60 o S 40 o S 20 o S Eq 20 o N 40 o N 60 o N 80 o N -1 80S 60S 40S 20S 0 20N 40N 60N 80N Northward C Transport [PgC yr -1 ] Preindustrial Ocean Transport Atlantic Indian&Pacific Global 80 o S 60 o S 40 o S 20 o S Eq 20 o N 40 o N 60 o N 80 o N b atmospheric CO 2 [ppm] Contemporary Atmos- pheric CO 2 d Inversion Takahashi et al. (1999) TRANSCOM Keeling et al. (1989) Tans et al. (1990) scenario 8 80S 60S 40S 20S 0 20N 40N 60N 80N Latitude Latitude Gloor et al. (submitted) Gruber et al. (in prep.)

26 atmospheric CO 2 [ppm] OCEAN INVERSION: ATMOSPHERIC RESPONSE Contemporary Atmospheric CO 2 Inversion (KVLOW(h)/GCTM) Takahashi et al. (1999) a S 60S 40S 20S 0 20N 40N 60N 80N Latitude OCEAN INVERSION: IMPACT ON LAND FLUXES Posterior Estimate: Inversion prior Posterior Estimate: Takahashi prior b 1.00 Fluxes [PgC/yr] N America Eurasia Tropical Land SH land Gruber et al. (in prep.)

27 pco 2 VARIABILITY NEAR MONTEREY BAY pco 2 [µatm] (a) J S N (b) pco 2 Temperature J M M J S N J M M J S N J M M J S N J M M J S Temperature [ C] partial pressure of CO 2 [µatm] Spring Summer CALIFORNIA COAST Fall Winter atmospheric CO 2 is about 370ppm OFFSHORE Offshore distance [km] Friederich et al. (in press) Chavez et al. (in prep.)

28 Regional Ocean Modeling System (ROMS) Gruber et al. (in prep.)

29 Summary and Outlook Forward and inversely based estimates of the pre-industrial and anthropogenic CO 2 fluxes show overall very similar patterns and magnitudes, which are generally similar to observationally based estimates. However, this agreement breaks down in the Southern Hemisphere. In particular, the inversely based estimates indicate a substantially smaller CO 2 uptake in the subpolar regions than indicated by pco 2 based estimates. Atmospheric inversions tend to come to similar conclusions. OCMIP forward models and our inverse models find a southward CO 2 transport of the order of 0.4 Pg C yr 1 across the equator. This is considerably smaller than has been proposed by Keeling et al. [1989], and Broecker and Peng [1992]. We nevertheless find an interhemispheric gradient in atmospheric CO 2, but primarily driven by the within hemisphere asymmetry in the fluxes. We find that our inversely estimated CO 2 fluxes, when used as priors in an atmospheric CO 2 inversion instead of those by Takahashi et al. [1999], lead to a relatively small change in the northern hemisphere terrestrial fluxes, but large changes in the tropical and southern hemisphere land regions.

30 Regional ocean modeling advances have made it possible to address also the highly variable CO 2 flux dynamics of the coastal ocean, which might be important for regional atmospheric CO 2 inversions. There is large potential in further exploring how inverse techniques can be used to fuse at the same time oceanic and atmospheric data with models.