Contribution to the annual report of the office of climate observations

Observing the Global Oceanic Carbon Cycle

Rik Wanninkhof, AOML, & Richard Feely, PMEL

The Global Carbon Cycle: Inventories, Sources and Sinks

Carbon dioxide is one of the major greenhouse gases, contributing about 60 % of the total change in radiative forcing due to human perturbations (Houghton et al. IPCC, 2001). The total emission due to fossil fuel use and cement production averaged about 6.3 ± 0.4 Pg C per year in the 1990s (1 Pg C=1 peta gram carbon = 10¹⁵ gram =1 gigaton). Although this annual CO₂ release has an appreciable effect on the earth radiation balance, it is a small fraction of the reservoir sizes comprising less than 1 % of the CO₂ in the atmosphere, 0.3 % of the labile terrestrial carbon pool; and 0.02 % of the total carbon content of the ocean. The annual addition is also much smaller than the natural exchanges between the reservoirs comprising less than 10 % of the natural annual exchanges between ocean and atmosphere, and between the terrestrial biosphere and atmosphere. The reservoir sizes and exchanges between reservoirs on annual basis are shown in Figure 1.

Figure 1: Fluxes and inventories of the Global Carbon System

Figure 1: Fluxes and inventories of the Global Carbon System

Figure 1. Cartoon of fluxes (arrows) and inventories (number in boxes) of the labile components of the global carbon system for the 1980's. The red arrows are the perturbation fluxes resulting from emissions of anthropogenic CO₂. From Sabine et al. (2003).

Although the anthropogenic perturbation seems small compared to the natural cycling of carbon between ocean, atmosphere and terrestrial systems, models and observations suggest that the increasing CO₂ levels in the atmosphere are causing an increase in global temperature (Fig. 2). While the evidence is rapidly growing for a causal relationship, it has not been unambiguously established yet. The perturbation is also showing effects on terrestrial and oceanic ecosystems.

Figure 2: Trends of atmospheric CO2 levels and global sfc. temp.

Figure 2. Trends of atmospheric carbon dioxide levels (white dashed line, right axis) and global surface temperature (red dashed line).

Based on carbon and carbon isotopic records in ice cores and tree rings we know that the atmospheric CO₂ levels remained very constant at 280 ± 5 parts per million (ppm) for the millennium prior to the industrial revolution. The remarkably constancy of atmospheric CO₂ despite large exchanges between the major reservoirs suggests a well-balanced global carbon cycle prior to the industrial revolution. However, we also know from the paleo-records that atmospheric CO₂ levels varied in proportion to global temperatures between ice ages and warmer periods. Therefore it seems quite possible that the current dramatic atmospheric CO₂ level rise will have a significant effect on climate and ecosystems. Moreover, although we have good paleo-records of climate in low CO₂ environments we have very limited information of how the earth responds to the current unprecedented high CO₂ levels and anticipated increases in the next century.

Our current knowledge of the fate of the anthropogenic CO₂ released to the atmosphere is based on models; atmospheric observations of CO₂ , carbon isotopes and small decreases in oxygen levels; terrestrial measurement of biomass inventories and primary productivity; and oceanic measurements of CO₂ inventories and fluxes between air and ocean. Less than a decade ago there were significant discrepancies between estimates leading to the popular notion of the "missing carbon sink" there now is a broad agreement that the "missing sink" is uptake by the terrestrial ecosystems based on disparate methods as summarized in Table 1. As the table indicates, our level of confidence in different observations ranges from a general good knowledge of the annual changes in some reservoirs, to highly uncertain estimates in others. Annual releases due to fossil fuel burning and cement production, and annual atmospheric CO₂ increases are the most constrained. Decadal changes in the ocean carbon inventory have recently been established with reasonable confidence. Changes in the terrestrial biosphere have been more difficult to pinpoint. From a variety of observations we now have a reasonable estimate of the partitioning of the fossil fuel carbon between reservoirs over the last two centuries with roughly 50 % ending up in the ocean. The terrestrial systems released CO₂ over this same period. Over the last two decades, however, the terrestrial systems appear to have taken up CO₂ but the magnitude, cause, and particularly the longevity of this sink remains in great doubt. Significant efforts, such as those proposed in the North American Carbon Plan (NACP), are underway to directly determine CO₂ sources and sinks in the terrestrial system. However, in the foreseeable future the best approach for constraining the net terrestrial flux will be from the difference between atmospheric and oceanic observations and model calculations.

Table 1. Global inventory of anthropogenic CO₂ for the past 200 and 20 years
CO₂ Sources	1800-1994 [Pg C]	1980-1999 [Pg C]
Constrained sources and sinks
(1) Emissions from fossil fuel and cement production	244 ± 20	117 ± 5
(2) Storage in the atmosphere	165 ± 4	-65 ± 1
(3) Uptake and storage in the ocean	-118 ± 19	-37 ± 8
Inferred net terrestrial balance
(4) Net terrestrial balance = [-(1)-(2)-(3)]	39 ± 28	-15 ± 9
Terrestrial balance
(5) Emissions from land use change	100 to 180	24 ± 12
(6) Terrestrial biosphere sink = [-(1)-(2)-(3)]-(5)	-61 to -141	-39 ± 18
From Sabine et al., 2004

The need for an integrated investigation of the carbon cycle has been well articulated in the US Carbon Cycle Science Plan (Sarmiento and Wofsy, 1999). Through efforts of the Interagency Carbon Working Group and the Scientific Advisory Committee, science and implementation plans have been developed for subcomponents of the program including the NACP Science Plan, the NACP Implementation Strategy, the Ocean Carbon and Climate Change Implementation Strategy, and the Large Scale Carbon Observing Plan (LSCOP)(Bender et al. 2001). The LSCOP plan in particular focuses on the implementation and justification for sustained ocean observations. All of the plans address the central tenets of the Carbon Cycle Science Plan which focuses on the "excess carbon", that is the carbon produced by fossil fuel burning and other activities of mankind releasing CO₂ such as land use change:

Where has the excess carbon gone to over the last two centuries?
Where will the excess carbon go to in the future?
What processes are involved in sequestration of the excess carbon?
Can the future sinks be managed and increased?

Because of the sensitivity of the global economy to terrestrial and oceanic ecosystems, and regional climate, the issue of carbon accounting transgresses the usual stakeholders of scientific information. Like emissions of pollutants, carbon emissions now have an economic value. A number $40 per metric ton carbon sequestered is often used in estimates. Improved constraints on the carbon sources and sinks can now be directly translated into a currency equivalent. For instance the global uptake of carbon by the ocean of about 1.6 Pg C yr^-1 (Table 2) translates into a $64 billion service to the global economy. As shown in Table 2, the uncertainty in the ocean sink is significant translating into an uncertainty in the value of this commodity. Knowledge of the future sink strength of the ocean is thus a critical from scientific and economic perspective.

Table 2: Summary of estimated global CO₂ fluxes using different gas transfer velocities but the same ΔpCO₂ climatology
Parameterization	Uptake (Pg C/yr)

Wanninkhof, 1992	-1.6
Wanninkhof&McGillis, 1999	-1.9
Nightingale, 2000	-1.2
Liss and Merlivat, 1983	-1.0
All these values were obtained using the ΔpCO₂ climatology of Takahashi et al. 2002 and 41-year climatological 6-hour winds from the NCAR/NCEP reanalysis project. The divergence of values illustrates that besides determining seasonal ΔpCO2 fields the gas transfer velocity needs to be better constrained. References for the relationships: Liss, P.S., and L. Merlivat, Air-sea gas exchange rates: Introduction and synthesis, in The Role of Air-Sea Exchange in Geochemical Cycling, edited by P. Buat-Menard, pp. 113-129, Reidel, Boston, 1986. Nightingale, P.D., G. Malin, C.S. Law, A.J. Watson, P.S. Liss, M.I. Liddicoat, J. Boutin, and R.C. Upstill-Goddard, In situ evaluation of air-sea gas exchange parameterizations using novel conservative and volatile tracers, Global Biogeochem. Cycles, 14, 373-387, 2000. Wanninkhof, R., Relationship between gas exchange and wind speed over the ocean., J. Geophys. Res., 97, 7373-7381, 1992. Wanninkhof, R., and W.M. McGillis, A cubic relationship between gas transfer and wind speed, Geophys. Res. Let., 26, 1889-1893, 1999.

The Sustained Ocean Component of the Carbon Cycle Science Plan

The oceanic carbon-observing program addresses two important subcomponents of the determination of the fate of the excess CO₂ in the ocean:

Determining oceanic carbon inventories and attributing the cause of the variations in inventories over time
Quantifying the air-sea CO₂ fluxes and creating of seasonal flux maps

Ocean inventories

As a result of the measurements during the global CO₂ survey in the 1990s and improved methods of quantifying the anthropogenic CO₂ signal above the large natural background, we now have the first measurement based inventory of anthropogenic CO₂ in the ocean. The excess CO₂ has been gridded at 1 degree spacing and 33 levels so it can be compared directly with model outputs. The observations show that surface waters are in near equilibrium with the atmospheric rise with a perturbation of the total carbon content of about 3 % (60 µmol kg^-1 out of a natural background of 2000 µmol kg^-1). The anthropogenic inventory decreases rapidly with depth for most parts of the ocean. Characteristic cross sections for the Atlantic, Indian and Pacific basins are shown in Figure 3. The distribution closely follows the known ventilation pathways of the ocean with deep penetration in the North Atlantic and storage of much of the carbon in the mid-latitude convergence zones. The total uptake over the past 200 years shown in Table 1 validates the model estimates. The total inventory is similar to models but the regional inventory is quite different suggesting that most of the models do not adequately capture the processes responsible for uptake at regional scales.

Figure 3: Sections of anthro.CO2 from the 3 ocean basins

Figure 3: Sections of anthro.CO2 from the 3 ocean basins

Figure 3. Representative sections of anthropogenic CO2 (µmol kg^-1) from the Atlantic (a), Pacific (b) and Indian (c) oceans. Grey hatched regions and numbers indicate amount of anthropogenic carbon stored (Pg C) in the intermediate water masses The two heavy lines on each section give the characteristic potential density contours for the near surface water and intermediate water. Much of the penetration of anthropogenic carbon into the ocean follow isopycnal surfaces. From Sabine et al. (2004).

Decadal inventory changes

The measurement based total inventory of anthropogenic carbon in the ocean is a critical constraint for models and for our understanding of the role of the ocean in the sequestration of excess carbon. However, information on shorter timescales is essential to determine any feedbacks of oceanic carbon sequestration due to climate change, and to determine the role of natural variability on the oceanic carbon system. Therefore the COSP has started, in collaboration with NSF and NASA, a repeat hydrography program. The main objective of the repeat hydrography component of the sustained ocean observing system for climate is to document long-term trends in carbon storage and transport in the global oceans. This program will provide composite global ocean observing system large-scale observations that include: 1) detailed basin-wide observations of CO₂ , hydrography, and tracer measurements; and 2) data delivery and management. This repeat hydrography program will provide the critical and timely information needed for climate research and assessments, as well as long-term, climate quality, and global data sets.

Figure 4. Distribution of the apparent oxygen utilization (AOU) difference between 2003 – 1988 (µmol kg-1) in the North Atlantic Ocean along 20° W. The large differences between 400 – 800m in the water column corresponds to changes in oxygen content of over 20 % at these depths (preliminary data provided by J. Bullister, PMEL).

The first three cruises of the repeat hydrography program were completed in 2003 focusing on the North Atlantic to provide a constraint for the NACP program. The initial highlights are that the ventilation pattern/circulation in the North Atlantic thermocline has changed based on a significant change in oxygen content (Fig. 4). Also, we have been able to unambiguously determine an increase in total carbon content in the upper ocean over 6 to 10 years suggesting that uptake of anthropogenic CO₂ continues unabated and that we can detect anthropogenic carbon increase in the ocean on decadal timescales. (Fig. 5)

Figure 5. Comparison of the total dissolved inorganic carbon concentrations, DIC concentrations a) near the surface (in the density range of 26.0 – 26.6); and b) for the 27.80 – 27.93 isopycnals for 1997 and 2003 in the western North Atlantic. (based on preliminary data from M. Roberts-Lamb, PMEL)

Atmosphere-Ocean CO₂ Fluxes

Background
Changes in carbon inventory are the most robust means of assessing sources and sinks but for the oceans these methods are limited to changes over decadal timescales. On average the total dissolved inorganic carbon content (DIC) of the surface ocean increases by about 1 µmol kg^-1 per year or about 0.05 % over the background. While the accuracy of DIC measurements is about 2 µmol kg^-1 making detection of the anthropogenic signal in principle possible on shorter time scales, the surface ocean DIC changes by 20 to 50 µmol kg^-1 seasonally masking changes less than 5 to 10 µmol kg^-1.

To assess changes in exchanges between reservoirs on sub-decadal timescale we have to determine the fluxes. The fluxes can be determined from measuring the partial pressure differences of CO₂ between surface ocean and lower atmosphere, Δ pCO₂ , and a quantity referred to as the gas transfer velocity that is related to physical forcing and often parameterized with wind speed. Thus, if ΔpCO₂ fields can be determined and used in combination with wind fields, regional fluxes can be obtained.

Creation of flux maps This approach has been applied successfully using a global climatology of Δ pCO₂ painstakingly developed based on 40-years of ΔpCO₂ data from many investigators (Takahashi et al., 2002). Uptakes based on this climatology range from 1 to 1.9 Pg C yr^-1 depending on the relationship between gas exchange velocity and wind speed (Table 2). This approach will be used to quantify regional fluxes on seasonal timescale. The implementation will require a significant increase in ΔpCO₂ observations, development of methods to interpolate ΔpCO₂ in time and space, and improvement of algorithms to quantify the gas transfer from wind or other relevant parameters, such as surface roughness, that can be directly observed from remote sensing.

Following a recommendation in the LSCOP plan a surface ocean flux observing system is being put in place with autonomous instrumentation on volunteer observing ships VOS, research ships, and buoys. The LSCOP plan lays out an observing strategy based on scaling analysis that involves sampling of the ocean roughly at 10 degree spacing and monthly intervals. By coordinating efforts with international and national partners this goal will be attainable in the next decade for the North Atlantic, North Pacific and Equatorial Pacific, particularly if we develop methods to increase time and space scales of observation through use of remotely sensed observations. The scheme of implementing such a system utilizing in situ and remotely sensed data is outlined in Figure 6.

Figure 6: Flow diagram of data and procedures to produce pCO2 flux maps

Figure 6. Flow diagram of data and procedures to produce pCO₂ flux maps.

Determining and attributing changes in ΔpCO₂
The approach of utilizing remote sensing, algorithms of ΔpCO₂ and gas exchange with remotely sensed products has been utilized in test beds in the Equatorial Pacific and Caribbean Sea. Flux map products for these regions are shown in figures 7 and 8. For the Equatorial Pacific work the algorithms are used in a retrospective fashion to determine the large variations in air-sea flux due to the ENSO Cycle.

Figure 7: pCO2 and CO2 flux in the Pacific: 1997 to 2003

Figure 7. Maps of ΔpCO₂ (left) and CO₂ fluxes in moles m^-2 yr^-1 in the equatorial Pacific from November 1997 thru April 2003 based on in situ observations and remotely sensed winds and sea surface temperature. The higher pCO₂ values and normal winds in the eastern Pacific during the 2002-03 El Niño event led to unusually high sea-to-air CO₂ fluxes for an ENSO event. After Feely et al. (2002).

Figure 8. Production of pCO₂ maps in the Caribbean. Empirical algorithms are being developed with parameters that are measured at higher density/frequency (e.g. through remote sensing.). The close correspondence of temperature (left panel) trends and pCO₂ (right panel) along the cruise track (bottom) facilitates robust algorithms to extrapolate the pCO₂ to regional scales. From Olsen et al. (2004).

Limited time series records of surface water pCO₂ levels have shown that for much of the ocean the surface water pCO₂ rises roughly at the same rate as the atmospheric increase implying that the global air-sea flux remains the same. However, changes in the rate of increase are a sensitive indicator of changes in the uptake of the ocean and perturbations in the biogeochemical cycles. Using a historical database of ΔpCO₂ for the Equatorial Pacific Takahashi et al. (2003) determined significantly slower increases in the 80-ties than in the 90-ties that were attributed to a climatic re-organization in the North and Equatorial Pacific referred to as the Pacific Decadal Oscillation (PDO).

Future plans and milestones

The observational efforts to detect changes in water column inventories and to attribute the causes, and the development of regional CO₂ flux maps are part of well documented and justified integrated carbon plans. The CO₂ /CLIVAR Repeat Hydrography Program has a series of cruises planned for the next decade that will yield sequential basin wide inventory changes for the Atlantic, Pacific, Southern and Indian oceans. The cruise sequence is listed in Table 3. NOAA/COSP has the lead on the cruises for A16S, A16N, P16N, P18 and I8. NOAA participants will perform DIC and pCO₂ measurements on all cruises. Operational Milestones are provided in Table 4.

Table 4. Operational milestones of the CO₂ - CLIVAR Repeat Hydrography Program
Summer 2003	Organize and complete the A16N cruise in the North Atlantic and provide leadership (chief scientist), CTD, oxygen, nutrient, total carbon and pCO₂ analysis.
Winter 2003/2004	Provide final CO₂, oxygen, CTD data to the repeat hydrography data center at Scripps.
Summer 2004	Analyze total inorganic carbon on the P2 cruise
Winter 2004/2005	Provide final total CO₂ data to the repeat hydrography data center at Scripps
Winter 2004/2005	Organize and complete the A16N cruise in the North Atlantic and provide leadership (chief scientist), CTD, Oxygen, nutrient, total carbon and pCO₂ analysis.
Winter 2004/2005	Analyze total inorganic carbon on the P16S cruise
Spring 2006	Organize and complete the P16N cruise in the Pacific and provide leadership (chief scientist), CTD, Oxygen, nutrient, total carbon and pCO₂ analysis.

The COSP CO₂ flux map effort focuses on the ΔpCO₂ observations needed to create the seasonal maps. The initial lines in the North Atlantic are shown on figure 9. The implementation schedule is presented in Table 5 with the italicized text that will be proposed in FY-05. The effort is starting to incorporate time series on moorings that are critical to determine the higher frequency (< 1-month) temporal variability. Particularly in the coastal oceans and Equatorial Pacific large changes can occur on weekly timescales. The exact balance and number of fixed pCO₂ observing sites vs. ship based (moving) observing platform has not been firmly established. Analysis of the results of the initial surface pCO₂ observing system will be used to optimize spacing, frequency, and mix of observing methods. Optimizing the observing system requires inclusion of measurements of biogeochemical and physical parameters that influence pCO₂ as well in order to investigate extrapolation routines. The added benefit will be that these parameters yield mechanistic information that can be used in prognostic models and interpolation schemes utilizing satellite data.. An to end-to end iterative effort starting from observations to interpretation and analysis feeding into improved observing system design and assessing the state of the ocean carbon cycle is critical at this point and attainable within national and international frameworks.

Figure 9: VOS pCO2 lines in the North Atlantic

Figure 9. VOS pCO₂ lines in the North Atlantic. The blue anf green lines are those funded by COSP and are part of the GOOS XBT observing network. The dashed lines are the routes outfitted by our European partners as part of the proposed European Carbo-Oceans project.

National and International linkages

The COSP carbon program is an integral part of national and international programs in carbon cycle research. NOAA's contribution is unique as it is the only program that has the sustained observational effort necessary to constrain sources and sinks and provide input for prognostic models to predict future trends. The international connection for the repeat hydrography effort is through WCRP/CLIVAR and the IGBP/IMBER programs. The former is focused on the physical aspects of climate variability while the latter is geared to the ecological and biogeochemical components. The flux map effort is connected to the SOLAS effort theme 3: Air-Sea Flux of CO₂ and Other Long-Lived Radiatively-Active Gases. International coordination for both aspects of CO₂ COSP will occur through the International Ocean Carbon Co-ordination Project (IOCCP). International ties between the ocean carbon programs and the atmospheric, terrestrial, and human dimension carbon cycle research are provided through the IGBP/WCRP/IHDP Global Carbon Project (GCP).

At a national level the CO₂ COSP effort is part of the US Carbon Cycle Science Plan. Its critical role in the overall US ocean science effort is outlined in the multi-agency implementation plan, the Ocean Carbon and Climate Change plan (Doney, 2004). Information about the programs linked to, or which are a part of COSP-CO₂ can be found in Table 6.

Table 6: Web sites of the CO2/COSP program and program partners: Links open new browser window
Data sites for pCO2 data from ships:
AOML	http://www.aoml.noaa.gov/ocd/gcc
PMEL	http://www.pmel.noaa.gov/co2/uwpco2/
LDEO	http://www.ldeo.columbia.edu/res/pi/CO2/
Program sites
CLIVAR:	Climate Variability and Predictability: http://www.clivar.org
SOLAS:	Surface-Ocean Lower Atmosphere Study: http://www.uea.ac.uk/env/solas/
IOCCP:	International Ocean Carbon Coordination Project http://ioc.unesco.org/ioccp
IGBP:	International Geosphere-Biosphere Program: http://www.igbp.kva.se/
IMBER:	Integrated Marine Biogeochemistry and Ecosystem Research http://www.igbp.kva.se/obe/recentupdates.html
WCRP:	World climate research Program: http://www.wmo.ch/web/wcrp/wcrp-home.html
GCP:	Global Carbon Project: http://www.globalcarbonproject.org/

Key References

Bender, M., S. Doney, R.A. Feely, I.Y. Fung, N. Gruber, D.E. Harrison, R. Keeling, J.K. Moore, J. Sarmiento, E. Sarachik, B. Stephens, T. Takahashi, P.P. Tans, and R. Wanninkhof, A large Scale Carbon Observing plan: In Situ Oceans and Atmosphere (LSCOP), pp. 201, Nat. Tech. Info. Services, Springfield, 2002.
Feely, R.A., J. Boutin, C. E. Cosca, Y. Dandonneau, J. Etcheto, H. Y. Inoue, M.Ishii, C. Le Quere, D. Mackey, M. McPhaden, N. Metzl, A. Poisson, and R. Wanninkhof., Seasonal and interannual variability of CO₂ in the Equatorial Pacific, Deep Sea Res. II, 49, 2443-2469, 2002.
Doney 2004, The Ocean Carbon and Climate Change Report, UCAR, Boulder 2004
Houghton, J.T., Y. Ding, D.J. Griggs, M. Noguer, P.J.v.d. Linden, and D. Xiaosu, Climate Change 2001: The Scientific Basis: Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), pp. 944, Cambridge University Press, Cambridge, England, 2001.
Sabine, C. L., M. Heimann, P. Artaxo, D,C. E. Bakker, C-T. A. Chen, C.B. Field, N. Gruber, C. Le Quéré, R.G. Prinn, J. E. Richey, P.R. Lankao, J.A. Sathaye, and R. Valentini, Current Status and Past Trends of the Global Carbon Cycle. Scope, 2003 in press
Sabine, C.L., R.A. Feely, N. Gruber, R. Key, K. Lee, J.L. Bullister, R.Wanninkhof, C.S. Wong, D.W.R. Wallace, B.Tilbrook, F.J. Millero, T.-H. Peng, A. Kozyr, T. Ono, and A.F. Rios, The oceanic sink for anthropogenic CO₂ , Science, submitted, 2004.
Sarmiento, J.L., and N. Gruber, Sinks for anthropogenic carbon, Physics Today, August, 30-36, 2002.
Sarmiento, J.L., and S.C. Wofsy, A U. S. carbon cycle plan, pp. 69, UCAR, Boulder, 1999.
Takahashi, T., S.G. Sutherland, C. Sweeney, A.P. Poisson, N. Metzl, B. Tilbrook, N.R. Bates, R. Wanninkhof, R.A. Feely, C.L. Sabine, J. Olafsson, and Y. Nojjiri, Global sea-air CO₂ flux based on climatological surface ocean pCO₂ , and seasonal biological and temperature effects, Deep-Sea Res. II, 49, 1601-1622, 2002.
Takahashi, T., S.C. Sutherland, R.A. Feely, and C.E. Cosca., Decadal variation of surface water pCO₂ in the Western and Central Equatorial Pacific, Science, 302, 852-856, 2003.