Anthropogenic carbon in a varying ocean Fortunat Joos, Thomas Fröhlicher Climate and Environmental Physics, Physics Institute, University of Bern CARBOOCEAN Meeting Bremen, December, 2006 Thanks to C. Lo Monaco, A. Velo and co-workers to the MPI and IPSL modelling groups

Data from the past show that anthropogenic climate change is proceeding at high speed

(IPCC, 2007, Fig. TS-2a ) Time (years before present) CO 2 versus Antarctic Temperature The atmospheric concentration of carbon dioxide in 2005 exceeds by far the natural range over the last 650,000 years (180 to 300 ppm) as determined from ice cores (IPCC, SPM, 2007). Natural Range

What about rates of change?

The rate of increase in the combined radiative forcing from CO 2, CH 4 and N 2 O during the industrial era is very likely to have been unprecedented in more than 10,000 years (SPM, 2007) (IPCC, 2007, Fig. TS-2 )

(Joos and Spahni, PNAS, submitted) The age distribution of air enclosed in ice Greenland CH 4 Antarctic (Dome C), today Antarctic (Dome C), Last Glacial Maximum

(Joos and Spahni, PNAS, submitted) The rate of increase in the combined radiative forcing from CO 2, CH 4 and N 2 O during the industrial era is very likely to have been unprecedented in more than 10,000 years (IPCC, SPM, 2007) Rates of Change over the past 22,000 years

Models and system understanding: current carbon emissions will affect the climate for many millennia

Long-term CO 2 and sea level committment in EMICs Year 60% 0 Atmosphere Ocean Land 2000 GtC Thermal Expansion (m) 1 IPCC AR4 EMIC Intercomparison Plattner et al., J.Clim., 2007; Cumulative Emissions 0%

IPCC Scenario Meeting, Sep 2007 A new set of emission mitigation and baseline scenarios Four scenarios to be selected for AOGCM runs → use in CARBOOCEAN

(Van Vuren et al., submitted) Cumulative carbon emission in multi-gas mitigation scenarios

Projected CO2 in 2100 Baseline Mitigation 650 to 950 ppm 380 to 620 ppm (Van Vuren et al., submitted)

A new set of mitigation scenarios is available van Vuren et al., In preparation

(Van Vuren et al., submitted) Baseline Mitigation Radiative Forcing (W m -2 ): 6 to to 5.1 CO2 (ppm): 650 to to 620 Projected Radiative Forcing in 2100

(Van Vuren et al., submitted) Baseline Mitigation 2.6 to 4.6°C 1.1 to 2.4°C Projected Temperature Change (1990 to 2100)

Are there critical thresholds...?

. Ocean Acidification and Aragonite Saturation Observation- based NCAR CSM1.4 supersaturation AA AA (Steinacher, 2007)

. Observation-based Aragonite Saturation in the Atlantic NCAR CSM1.4 supersaturation  CO 3 — [mol/m 3 ] depth [m] 90 o S90 o N (Steinacher, 2007)

Evolution of Aragonite Saturation in the Surface Surface pCO 2 (ppm) Latitude 90 o S 90 o N supersaturation AA (Steinacher, 2007)

Fluxes of calcite and aragonite to depth CaCO3 flux in PISCES (PgC/m2/y) Aragonite Calcite Total CaCO 3 Gangsto, in prep. CaCO 3 Flux (PgC/yr) Depth (m) Total CaCO 3 Calcite Aragonite Will dissolution of shallow aragonite sediments mitigate some of the ocean acidification signal? Magnitude? Time scales?

460 ppm: Arctic Ocean becomes undersaturated with respect to Aragonite 560 ppm: Antarctic surface waters become undersaturated 560 ppm: surface water that is more than 3 times oversaturated dissappears Conclusions: Ocean Acidification

How well do different reconstruction methods of Canth work in the AOGCM model world?

a) Canth simulated by model b) Canth reconstructed from simulated tracers (C, Alk, O 2,...) Simulated and reconstructed Canth should be identical

a) Canth simulated by model

. NCAR CSM1.4 Surface Temperature and CO 2 for SRES A2 and B Year CO 2 (ppm)  T ( o C) A2 B1 A2 B1 Anthropogenic forcing - Fossil and land use CO 2 emissions - CH 4, N 2 O, CFCs - direct sulphate aerosols Natural forcing - solar irradiance - stratospheric volcanic aerosols Year Instrumental Data

. Oxygen in the Atlantic, 20 W depth 90 o N 80 o S O2 (  mol-C/kg), NCAR 80 o N

. Observation-based (GLODAP) Anthropogenic CO 2 NCAR CSM1.4 Canth (mol m -2 )

Change in decadal-mean PO 4 from 1820 to 2000 AD, Atlantic, 20 W Depth (m) 80 o S 60 o N 0 40  PO 4 *117 (  mol-C/kg) No century-scale trends decadal variability in high latitudes of NA

Modelled evolution of DIC and Canth for an individual grid cell (60 N, 20 W) Remove natural variability in DIC by splining to get Canth DIC (mmol/m 3 ) Time Canth DIC

Canth in the Atlantic along 20 W NCAR CSM1.4, 1994 Depth (m) 80 o S60 o N Canth (  mol-C/kg)

b) Canth reconstructed from simulated tracers (Carbon, Alk, O 2,...)

The TrOCA method as an example The usual assumptions - Fixed Redfield ratios to correct for remineralisation - No century-scale trends - time-invariant air-sea disequilibrium „Organic Matter Remineralization“ „CaCO3 dissolution“ Total Carbon „preindustrial Total Carbon“

Canth in the Atlantic along 20 W TrOCA with NCAR output, 1994 Depth (m) 0 Canth (  mol-C/kg) 80 o S60 o N

a) Canth simulated by model b) Canth reconstructed from simulated tracers (T, S, O2,...) Are simulated and reconstructed Canth identical?

TrOCA-NCAR „truth“, NCAR-Model (  mol-C/kg)

Potential Problems

Oxygen in NCAR TrOCA Remineralisation of organic matter does not consume O 2 in oxygen minimum zones (OCMIP Protocoll) Remineralisation and Canth is overestimated in reconstruction

Anoxic remineralisation of organic matter may bias Canth estimates

TrOCA-MPI „truth“, MPI Model (  mol-C/kg)

TrOCA-MPI Century-scale Trend in PO 4 in MPI Model Negative Canth in deep ocean

Century-scale trends may bias Canth estimates

What about interannual and decadal variability?

Internal Variability in AOU sdv of decal averaged AOU (  mol/kg) Frölicher et al., in prep

The impact of volcanic forcing on global mean AOU and O 2 Frölicher et al., in prep Depth (m) -  AOU O2O Year Optical Depth

The impact of volcanic forcing on global meanO 2 and AOU Frölicher et al., in prep Depth (m) -  AOU O2O Year

. Oxygen in the Atlantic, A16N O2 (  mol-C/kg), NCAR

. Oxygen in the Atlantic, 20 W depth 90 o N O2 (  mol-C/kg), NCAR

. Internal Variability in DIC and in  C* from a control run in top 2000 m depth (  mol-C/kg) -4 4 Levine et al., in press 60 o S80 o N 1 , DIC  C* 0 60 o S80 o N

. Difference between modeled and reconstructed Canth for the  C* method Levine et al., in press Modelled increase over 10 year Difference (model-reconstruction) 60 o S80 o N (  mol-C/kg)

Both externally-forced and internal variability may bias Canth estimates

Other potential problems? Parameters of reconstruction method have not been determined with model output Fixed Redfield ratios assumed in model – correct?

How do results from different methods compare with modeled Canth?

Reconstruction methods: TrOCA (Touratier et al.)  C T 0 (Vazquez-Rodriguez; adjusted C* method) C T 0 IPSL: (Lo Monaco; back-calculation method, uses different preformed relationships for southern and northern water)

TrOCA „Truth“: NCAR Model CT0CT0 IPSL (  mol-C/kg)

Difference between simulated and reconstructed Canth, 20 W,1994 (  mol-C/kg) Depth (m) o N80 o S TrOCA

Difference between simulated and reconstructed Canth, 20 W,1994 (  mol-C/kg) Depth (m) o N80 o S CT0CT0

Difference between simulated and reconstructed Canth, 20 W,1994 (  mol-C/kg) Depth (m) o N80 o S IPSL

Conclusions Significant deviations between predicted and reconstructed Canth are found for all methods Potential biases: - externally-forced and internal variability - anoxic remineralisation - century-scale trends in watermasses/tracers - deviations from fixed Redfield ratios - parameters not determined with model output (IPSL Reconstruction Method)

. Oxygen in the Atlantic, 20 W depth 90 o N 80 o S O2 (  mol-C/kg), NCAR 80 o N

. Oxygen in the Atlantic, 20 W depth 90 o N 80 o S O2 (  mol-C/kg), NCAR 80 o N

. Oxygen in the Atlantic, 20 W depth 90 o N 80 o S O2 (  mol-C/kg), NCAR 80 o N

. Oxygen in the Atlantic, 20 W depth 90 o N 80 o S O2 (  mol-C/kg), NCAR 80 o N

. Oxygen in the Atlantic, 20 W depth 90 o N 80 o S O2 (  mol-C/kg), NCAR 80 o N

DIC in the Atlantic,1994, 20 W

PO4 in the Atlantic, 20 W

Canth in the Atlantic, 20 W

in the Atlantic, 24 N

in the Atlantic, 20 W 1994

in the Atlantic, 20 W 1994

PHICT in the Atlantic, 20 W

PHICT in the Atlantic, 24 N

TROCA in the Atlantic, 20 W

TROCA in the Atlantic, 24 N

in the Atlantic, 20 W

What ist the impact of 1. temperature distribution 2. organic matter export 3. CaCO 3 export on DIC and atm. CO 2 ? Apply a model to discriminate the mechanisms

Dead ocean with T=18 o C (present day carbon inventory in ocean-atmosphere system) Uniform distribution of DIC pCO 2 (atm) = 560 ppm AtlanticSouthern OceanPacific Depth DIC (mmol/m 3 )

Temperatures in the World Ocean

1. Solubility and carbon chemistry: cold water holds more DIC than warm water → DIC concentration in the (warm) surface are on average depleted with respect to the (cold) deep ocean → atmospheric CO2 is lower compared to an ocean with a uniform temperature of T=18 o C

Dead ocean with present temperature distribution (present-day carbon inventory in ocean-atmosphere system) Surface water somewhat depleted in DIC pCO 2 (atm) = 439 ppm Depth AtlanticSouthern OceanPacific DIC (mmol/m 3 )

2. Marine biota removes carbon from surface waters and this carbon is exported to the deep → DIC and nutrient concentrations in the surface are on average depleted with respect to the deep ocean → atmospheric CO 2 is lower compared to a dead ocean

The marine organic carbon cycle

Vertical distribution of tracers in the North Pacific Redfield ratio: P:N:C:O2 = 1:16:120:-170

Observed distribution of phosphate Surface water in the Atlantic and Pacific are depleted in nutrients by biological activities Depth AtlanticSouthern OceanPacific (mmol/m 3 )

Ocean with organic matter production and temperature distribution (but no calcite production) Surface water depleted in DIC pCO 2 (atm) = 229 ppm Depth AtlanticSouthern OceanPacific DIC (mmol/m 3 )

3. Production and export of CaCO 3 increases the partial pressure of CO 2 in surface waters → A range of marine organisms form shells made of CaCO 3 Ca HCO 3 - → CaCO 3 + CO 2 + H 2 O → Decrease in DIC, but shift in the ratio between different carbonate species → Alkalinity ~ [HCO 3 - ]+2 [CO 3 -- ] is decreasing → [H 2 CO 3 *] in the surface and atmospheric CO 2 is increased compared to an ocean without calcifying organisms

Observed distribution of potential Alkalinity Surface water is depleted in alkalinity Depth Pot. Alk (mmol/m 3 ) AtlanticSouthern OceanPacific

Observed distribution of DIC Surface water depleted in DIC and in alkalinity pCO 2 (atm) = 278 ppm Depth DIC (mmol/m 3 ) AtlanticSouthern OceanPacific

Regulation of atmospheric CO 2 Dead ocean with uniform T of 18 o C: 560 ppm Realistic temperature distribution:439 ppm + organic matter export229 ppm + CaCO 3 export278 ppm

Greenland and Antarctic temperature, CO 2 and CH 4 over the last transition Stauffer, Monnin, Blunier and co-workers, 2003 Past changes indicate the magnitude of potential future feedbacks

„Surprises in the Climate System“? Projected strenght of the North Atlantic overturning Cubasch et al., 2001

From the past to the future Indermühle et al, 1999 NADW collapse: limited impact on atmospheric CO 2 and ocean uptake CO 2 varies by up to 20 ppm during D/O events WRE1000 Joos et al, 1999; Plattner et al., 2001

How well can ocean carbon cycle model simulate the present state? Selected Results from NCAR CSM1.4-carbon Thanks to S. Doney, I. Fung, K. Lindsay

. Observation-based (GLODAP) Anthropogenic CO 2 NCAR CSM1.4 DIC (mol m -2 )

Canth 2100 in Atlantic CSM1.4-carbon Laws et al. (2000) Total: 9.2 Pg C Total: 11.1 Pg C Figure2: Depth vs. latitude contour plot of annual mean anthropogenic CO 2 in the Atlantic Ocean for simulated with the CSM1.4-carbon.

. Ocean Acidification and Aragonite Saturation Observation- based NCAR CSM1.4 supersaturation AA AA

. Observation-based Aragonite Saturation in the Atlantic NCAR CSM1.4 supersaturation  CO 3 — [mol/m 3 ] depth [m] 90 o S90 o N

. Aragonite Saturation in the Pacific 0 Observation-based depth [m] NCAR CSM1.4 supersaturation  CO 3 — [mol/m 3 ] 0

GLODAP/WOA01 CSM-1.4-carbon Comparison with observations Saturation state Ω A at surface

Δ[CO 3 2- ] (umol/cm 3 ) Atlantic Pacific GLODAP/WOA01 CSM-1.4-carbon Comparison with observations

Phospate: Model versus Observations CSM1.4-carbonWorld Ocean Atlas (2001) Atlantic 20° W Pacific 167° W

Export Production of POC Laws et al. (2000) Global: 9.2 Pg C Export (g-C m -2 ) CSM Pg C

Export Production of POC CSM1.4-carbon Laws et al. (2000) Total: 9.2 Pg C Total: 11.1 Pg C Figure2: Depth vs. latitude contour plot of annual mean anthropogenic CO 2 in the Atlantic Ocean for simulated with the CSM1.4-carbon.

AOU Model vs WOA

. Apparent Oxygen Utilisation in the Pacific Observation-based NCAR CSM1.4 0 AOU (  mol/kg) depth 90 o N90 o S

. Apparent Oxygen Utilisation in the Atlantic Observation-based NCAR CSM AOU [  mol/kg]

. Apparent Oxygen Utilisation in the Indian Observation-based NCAR CSM AOU [  mol/kg]

. NCAR CSM1.4 Surface Temperature and CO 2 for SRES A2 and B Year CO 2 (ppm)  T ( o C) A2 B1 A2 B1 Anthropogenic forcing - Fossil and land use CO 2 emissions - CH 4, N 2 O, CFCs - direct sulphate aerosols Natural forcing - solar irradiance - stratospheric volcanic aerosols Year Instrumental

. NCAR CSM1.4 Decrease in the volume of water oversaturated with respect to aragonite (SRES A2)

Difference Decrease in Export of DOM ∆ ~ 8 % Year

Variability and a decreasing trend in meridional overturning circulation 80 o N Latitude 40 o N 0o0o 40 o S Year Overturning (Sv) 1900

Ocean Acidification: large decrease in the aragonite saturation in the tropics and subtropics Change in saturation state Ω A Difference 2100 – % -100% 0%

Ocean Acidification: large changes in the subtropics Saturation state Ω A at surface Difference 2100 – 2000 Evolution (global zonal mean) Contour lines: Δ[CO 3 2- ] (umol/cm 3 )

Difference Hovmoeller Meridional Overturning Circulation Decrease in North Atlantic Ocean

Long-term CO 2 and sea level committment in EMICs Year 60% 0 Atmosphere Ocean Land 2000 GtC Thermal Expansion (m) 1 IPCC AR4 EMIC Intercomparison Plattner et al., 2006; Cumulative Emissions 0%

1. Inverse methods such as Ensemble Kalman filtering provide powerful tools for improved estimates of biogeochemical quantities. It is time to include CARBOOCEAN measurements. 2. Bomb radiocarbon data suggest that the OCMIP air-sea transfer velocity field must be downscaled by ~26%; global mean: 16 cm/hr (Müller et al., 2006, Sweeney, 2006, Ho et al., 2006, Nägler et al., 2006) It is time to use a downscaled transfer velocity 3. Simulations with forced and internal variability available It is time to analyse internal and externally-forced variability Conclusions

CSM1.4-carbon Fossil only Sabine et al. (2003) CO 2 [mol/m 2 ] Column Inventory of anth. CO

A Historical Perspective Siegenthaler and Oeschger, Science, 1978: “With climate models becoming more and more realistic, a maximum permissible atmospheric CO 2 level might be found which should not be exceeded if the atmospheric radiation balance is not to be disturbed in a dangerous way. … This scenario clearly does not allow us to go on burning fossil fuel at the present growth rate for a long time … Around the turn of the century new technologies would have to take over a substantial part of global energy production.”

(IPCC, 2007, Fig. TS-31) Both past and future anthropogenic carbon dioxide emissions will continue to contribute to warming and sea level rise for more than a millennium, due to the timescales required for removal of this gas from the atmosphere.

Atmospheric Increase and Fossil Emissions IPCC, Chap 7, 2007

Fraction of Fossil Emissions Staying Airborne IPCC, Chap 7, 2007

Evaluating the overall impact in a probabilistic way SRES B1 Knutti et al, 2003 Higher CO 2 under global warming leads to an increased probability for high warming