The non-steady-state oceanic CO 2 signal : its importance , magnitude and a novel way to detect it

Introduction Conclusions References

For thousands of years before the onset of the industrial revolution (∼ 1800 AD), carbon cycling between atmospheric, land and oceanic biospheres was in relative steady state.Although large gross exchanges of CO 2 were occurring annually between land, atmosphere and oceans, atmospheric CO 2 remained relatively constant at about 280 ± 5 µatm, implying a steady state carbon cycle (Etheridge et al., 1998).Humans, via the burning of fossil fuel carbon, have emitted about 530 Pg C into the atmosphere, perturbing atmospheric CO 2 .
Quantifying the flows, exchanges and storage of this anthropogenic CO 2 in the earth system has been a primary objective for the biogeochemical research community.Due to the heterogeneity in both space and time within the land carbon system, partitioning the global carbon storage between land and ocean has largely relied on our more certain understanding of the oceans storage of anthropogenic CO 2 .Fortunately, a number of different independent methods have allowed researchers to quantify the oceanic anthropogenic CO 2 sink (Quay et al., 1992;Gruber et al., 1996;Keeling et al., 1996;Gruber and Keeling, 2001;McNeil et al., 2003;Sabine et al., 2004;Waugh et al., 2006;Khatiwala et al., 2009).Fundamental to these estimates is the assumption that large-scale natural cycling of carbon through biological and B. I. McNeil and R. J. Matear: The non-steady state oceanic CO 2 signal circulation pathways have remained in steady state throughout the 20th century, with the anthropogenic perturbation acting passively on top of that large natural but unchanging "background" carbon cycle.This steady state assumption was valid during most of the 20th century whereby any climate-related alterations to the oceanic anthropogenic CO 2 sink have been small in comparison to the large emissions signal itself (Sarmiento et al., 1998;Matear and Hirst, 1999).Although the steady state assumption may have been adequate for the 20th century due to the small impact on net oceanic CO 2 uptake, as discussed below, this is no longer the case for the 2st century.

The emerging non-steady state oceanic CO 2 signal
Over recent decades, oceanographers have observed largescale decadal and longer timescale trends in the ocean associated with biological changes, circulation changes and temperature-related solubility changes.The first observational research documenting large-scale decadal changes in the oceans circulation pathways were shown in the 1990s with temperature and salinity alterations in major ventilation pathways of the ocean (Wong et al., 1999).Since then the number of hydrographic measurements have increased, leading to a detailed understanding of the large-scale ocean warming trend (Levitus et al., 2000) as well as salinity changes associated with the amplification of the hydrological cycle (Durack and Wijffels, 2010;Helm et al., 2010).In the case of biogeochemistry, although nutrients were hypothesized to be changing (Pahlow and Riebesell, 2000), it was a suite of studies showing declining and/or changing oxygen concentrations in various parts of the ocean (Emerson et al., 2001;Matear et al., 2000;Whitney et al., 2007;Stendardo and Gruber, 2012), which confirmed that large-scale circulation and/or biological changes were impacting biogeochemical cycles.
A new suite of climate models (Wetzel et al., 2005;Le Quéré et al., 2007;Lovenduski et al., 2007;Matear and Lenton, 2008;Rodgers et al., 2008) driven with observed wind, heat and freshwater fluxes showed that beginning in the late 1980s, net CO 2 uptake started to level off.Oceanic CO 2 uptake was not increasing in the way it should be if the uptake was only a function of increasing atmospheric CO 2 levels and was quite aptly described by Sarmiento et al. (2010) as "somewhat of a surprise".
Le Quéré et al. (2007) combined this modeling result with atmospheric CO 2 measurements over the Southern Ocean to postulate that the net oceanic CO 2 sink was leveling in response to an observed intensification of winds over the Southern Ocean that caused higher outgassing of naturally CO 2 -rich deep waters, partially offsetting a large anthropogenic CO 2 uptake signal.Le Quéré et al. (2007) and a follow-up model study by Lovenduski et al. (2008) showed that the CO 2 outgassing in the Southern Ocean to be up to 35 % of the anthropogenic CO 2 flux.
These non-steady state ocean carbon changes create systematic biases in many tracer-based techniques that attempt to quantify the anthropogenic CO 2 storage in the ocean and these biases will continue to grow as the non-steady state ocean evolves through the 21st century.This new development is critically important from an atmospheric CO 2 perspective, since non-steady state changes in the ocean (at least the present outgassing) are a positive CO 2 feedback on atmospheric CO 2 levels.Recent evidence appears to show that the airborne fraction of CO 2 (the fraction of anthropogenic emissions which remain in the atmosphere) maybe increasing (Raupach et al., 2007;Le Quéré et al., 2009;Gloor et al., 2010;Knorr, 2012), and climate-driven oceanic CO 2 outgassing may be playing a role in those atmospheric CO 2 trends.The non-steady state CO 2 signal is what we seek to detect, since without it, the ability to monitor and predict future atmospheric CO 2 levels will be impeded.
In this manuscript we seek to do two things.First, we present a decomposition of total oceanic CO 2 changes over time into natural and anthropogenic, steady and non-steady state components.By doing this we seek to show the important difference between what often is referred to as anthropogenic CO 2 change in the ocean and the very different total net change in CO 2 in the ocean, since they are sometimes incorrectly used interchangeably.But foremost, this decomposition clarifies the differing steady and non-steady components in the oceanic CO 2 signal for budgetary purposes.Second, we investigate and present a simple data-based method to partition the time-evolving CO 2 sink into a steady state and non-steady state signal.

Decomposing the time evolution of CO in the Ocean
The net oceanic dissolved inorganic carbon change ( DIC Net ) between a period of time (t1 to t2) reflects changes in both natural and anthropogenic carbon dynamics: (1) Natural changes in DIC ( DIC Nat ) occur through temperature, biological and ocean circulation changes via climate/ocean variability and change.On top of these natural DIC changes are also changes in DIC due to the oceanic uptake of anthropogenic CO 2 ( ACO 2 ).
If there were no net changes to the natural DIC concentrations in the ocean over a given time period (i.e.steady state), then the time-evolving net change would be simply equal to the anthropogenic CO 2 uptake by the ocean (i.e.DIC Net = ACO 2 ).
In this steady state world, anthropogenic CO 2 can be treated as a passive solubility tracer such as chlorofluorocarbons (CFCs), whereby its uptake is driven solely by the This definition of anthropogenic CO 2 makes tracer-based approaches very attractive.However, as models have shown (Joos et al., 1999;Friedlingstein et al., 2006;Matear and Hirst, 1999;Sarmiento et al., 1998) and observed CO 2 trends show (Le Quéré et al., 2007;Lenton et al., 2012), the ocean CO 2 system is not in steady state, consistent with physical oceanic properties (temperature, salinity, oxygen) that have showed temporal change.
To account for this time-evolving behavior in CO 2 it is therefore necessary to decompose the time evolution of anthropogenic CO 2 (ACO 2 ) into its steady and non-steady state components as follows: where ACO 2 is the traditional steady state definition of anthropogenic CO 2 used in the literature, whereby circulation and biological carbon changes remain constant with rising CO 2 in the atmosphere.ACO 2 is the non-steady state term identified by climate change models whereby anthropogenic CO 2 is modified by changes in circulation and/or biology from global warming (like stratification or warming's effect on CO 2 solubility).This non-steady state term has been explored in ocean biogeochemical models and shows a relatively small but growing non-steady state anthropogenic CO 2 uptake signal by the end of this century (10-20 %) (Sarmiento et al., 1998;Matear and Hirst, 1999;Plattner et al., 2001;Friedlingstein et al., 2006;Matear and Lenton, 2008).
Natural decadal variability and change alter ocean circulation and biology, therefore impacting the time-evolving DIC signal, requiring a separate set of equations: (3) By definition, globally DIC Nat over time is 0, therefore the total changes in the natural carbon cycle ( DIC Nat ) are equal to the non-steady state changes to the natural carbon cycle ( DIC Nat ).For example, DIC Nat could be those carbon changes resulting from natural variability in the climate system, such as perhaps those driven from decadal-scale intensification of Southern Ocean winds, El Niño-Southern Oscillation events, trends in remineralization stoichiometry or even regional time-evolving movements in circulation pathways over two different periods of time.
What is important to remember here is that correcting for the natural DIC signal in the ocean from back-calculation techniques, such as C * , does not account for either the natural non-steady state signal ( DIC Nat ) or the anthropogenic non-steady state signal ( ACO 2 ).
In summary, the time-evolving net DIC signal is the sum of three terms: To simplify Eq. ( 4), we combine them into steady and nonsteady state signals: 1. ACO 2 ; the steady state change in ocean CO 2 over time due to rising atmospheric CO 2 in a unchanging ocean.
2. ( ACO 2 + DIC Nat ); the combined non-steady state signal that incorporates how a changing ocean alters DIC in the ocean.
The most important term in Eq. ( 4) for atmospheric CO 2 modulation is the combined net CO 2 sink ( DIC Net ); therefore, it is important to investigate and constrain the nonsteady state CO 2 signal.

The anthropogenic non-steady state signal
When introducing the C * method, Gruber et al. (1996) were clear that their technique required a steady state assumption.The global application of the C * method was performed by Sabine et al. (2004) and it is important to understand that they estimated the steady state oceanic anthropogenic CO 2 inventory (i.e.ACO 2 in Eq. 4) and assumed it was equivalent to the total net change in oceanic CO 2 (i.e.DIC Net ).Between 1880 and 1994, Sabine et al. (2004) estimated an anthropogenic CO 2 storage in the ocean of 118 ± 19 Pg C, which was recently increased to ∼ 155 Pg C by 2010 based on a different steady state tracer technique (Khatiwala et al., 2009).
In a commentary to Science, Ralph Keeling highlighted the important but missing non-steady state anthropogenic signal (i.e.ACO 2 ) within the Sabine et al estimate (Keeling, 2005).He noted that "Anthropogenic CO 2 " as it has been used traditionally is "an incomplete measure of the change in the ocean carbon content".He further discussed the necessary steady state assumption that oceanic tracer-based techniques require and made a first attempt to calculate the nonsteady state anthropogenic CO 2 change in association with recent climate change (i.e.ACO 2 ).
Using a box-diffusion model, Keeling (2005) estimated a 13 Pg C outgassing of CO 2 due to the thermodynamic effect of increasing CO 2 in warmer oceanic surface waters up until 1994.With recent climate change until that point, upper ocean density stratification would also have already occurred.Although this stratification likely limits the subduction of water masses into the interior, Keeling (2005) Keeling (2005).This non-steady state term for oceanic anthropogenic CO 2 is still less than 10 % of the final anthropogenic CO 2 inventory estimate and well within the total uncertainty of the technique (±19 Pg C), as noted by Sabine and Gruber (2005).So although ACO 2 has been known to bias the data-based estimates of anthropogenic CO 2 storage in the ocean, the bias is relatively small (∼ 10 %).
However, the most important term, at least over the past twenty years, in Eq. ( 4) is not ACO 2 but DIC Nat as recent discoveries have shown us.

The natural non-steady state signal
Le Quéré et al. (2007) showed that wind-speed intensification in the Southern Ocean was not only causing a large natural outgassing of CO 2 as other models showed (Lovenduski et al., 2007;Lenton et al., 2009), but this trend was detected from atmospheric CO 2 observations, although the atmospheric trend is still the subject of debate (Law et al., 2008).Arguably this Southern Ocean outgassing could be the fingerprint of climate change itself, but it could also be simply a natural decadal response and therefore would be mostly identified as a natural non-steady state signal ( DIC Nat ).In any case, partitioning this signal into ACO 2 , or DIC Nat is not important, since we combine both the natural and anthropogenic non-steady state signals for simplicity.
It is important to emphasize that the above decomposition is relevant for global changes in ocean carbon inventories that can then translated into global air-sea CO 2 fluxes.On a regional basis however, translating carbon inventory changes to regional air-sea CO 2 fluxes could be largely independent.For example a lateral change in the position of a gyre would translate into large changes in DIC Nat within the ocean interior without a corresponding air-sea CO 2 flux signature.
Although research has progressed in understanding the regional natural non-steady state CO 2 signal (Le Quéré et al., 2007;Lovenduski et al., 2007;Levine et al., 2008;Goodkin et al., 2011), how important is it on a global scale?
3 Can we detect the global non-steady state CO 2 signal?Sarmiento et al. (2010) synthesize a suite of different models from earlier studies (Wetzel et al., 2005;Mikaloff-Fletcher et al., 2006;Le Quéré et al., 2007;Lovenduski et al., 2008;Rodgers et al., 2008) to estimate the global magnitude of the combined non-steady state oceanic CO 2 signal between 1989 and 2007.For completeness, here we add the CSIRO biogeo-chemistry model (Matear and Lenton, 2008) to that suite of model results.
To illustrate the divergence of the time-evolving net oceanic CO 2 uptake in recent decades, we combine the five different model predictions that use time-evolving NCEP atmospheric forcings and compare it to the expected steady state uptake just from atmospheric CO 2 increases alone (Fig. 1).
On average, as reported by Sarmiento et al. (2010), timevarying ocean models take up 0.35 Pg C yr −1 less CO 2 between 1989 and 2007 than they would have if ocean circulation and biogeochemistry had remained in steady state.This would equate to a combined non-steady state CO 2 inventory reduction of about 6.3 Pg C between 1989 and 2007 in comparison to the total net CO 2 inventory ( DIC Net ) of about 36 Pg C from these models (Fig. 1).Each model differs in magnitude, but on average the magnitude of the combined non-steady state CO 2 signal between 1989 and 2007 is about 18 % of the total carbon stored in the ocean over this time period.
By combining the Keeling (2005) and Sarmiento et al. ( 2010) estimate, the total non-steady state CO 2 outgassing since the industrial revolution would be about 13 ± 10 Pg C, which is about ∼ 10 % of the steady state anthropogenic CO 2 inventory (∼ 155 Pg C).It is important to emphasize that the Keeling (2005) estimate was made using the Sabine et al. ( 2004) estimate for anthropogenic CO 2 sink up until 1994.The Sarmiento et al. ( 2010) model results were made from 1990 onwards, so that by combining these estimates results in a four year overlap.However, if we take the models from Sarmiento et al. (2010) as a gauge, the largest signal in the non-steady state comes after the mid-1990s, so this overlap would probably result in a small bias (∼ 0.5 Pg C of the estimate).
Despite the relatively small estimated non-steady state CO 2 signal over the entire industrial era, this non-steady state signal has grown to be about 18 % of net oceanic CO 2 storage between 1989-2007.Given the large magnitude of the non-steady state CO 2 signal simulated in the models since 1989, can we use data-based methods to constrain it?

The multi-method approach to estimate the non-steady state signal
In recent years there has been a proliferation of new tracerbased techniques to quantify decadal changes in oceanic CO 2 uptake (Hall et al., 2002(Hall et al., , 2004;;Gloor et al., 2003;Waugh et al., 2006;Khatiwala et al., 2009) that follow on from the C * method (Gruber et al., 1996) and earlier attempts (Chen, 1982).Other techniques using atmospheric observations, i.e.CO 2 inversions or atmospheric O 2 /N 2 methods (Ciais et al., 1995;Keeling et al., 1996;Bousquet et al., 2000;Keeling and Garcia, 2002;Patra et al., 2005;Manning and Keeling, 2006) or a combination (Jacobson et al., 2007)  Illustration of the time-evolving net CO 2 uptake from the ocean since 1970 comparing the expected steady state uptake that assumes constant circulation and biology (black line) with the mean net uptake from five different ocean models that include time-evolving natural forcings (Wetzel et al., 2005;Le Quéré et al., 2007;Lovenduski et al., 2007;Matear and Lenton, 2008;Rodgers et al., 2008).We include the results from the hindcast CSIRO climate model from Matear and Lenton (2008) and the suite of model projections documented in Sarmiento et al. (2010).The expected steady state CO 2 uptake is taken from Mikaloff-Fletcher et al. (2006) as documented by Sarmiento et al. (2010), normalized to the decade of 1970-1980.The average net CO 2 uptake (red dots) was taken as a combined mean between the five different models with minimum and maximum bounds shaded in light red.This is an illustrative figure used to demonstrate the time-evolving changes in the various CO 2 signals within the ocean as described in Eq. 4. The integrated dotted red curve is the total net CO 2 uptake by the ocean ( DIC Net ), which is about 36 Pg C between 1989 and 2007.The black curve is the expected steady state CO 2 uptake or what has been historically called anthropogenic CO 2 uptake ( ACO 2 ), which is about 42 Pg C between 1989 and 2007.The difference between these curves is therefore the combined non-steady state CO 2 loss to the atmosphere ( ACO 2 + DIC Nat ), which comes to about 6.3 Pg C outgassing to the atmosphere between 1989 and 2007 (see Table 2 of Sarmiento et al., 2010).
methods with the assumption they are quantifying the same time-evolving oceanic CO 2 signal, which is not true.Different data-based techniques constrain different oceanic CO 2 signals.Or to put it another way, each databased technique has a different sensitivity towards capturing the non-steady state signal in the ocean.For example, the C * method uses actual carbon measurements, that to some degree will contain some but not all of the non-steady state signal.The multi-method approach here requires the use of methods that capture either the steady-state-state signal or the net oceanic CO 2 signals with the most certainty.A technique which partially captures both the steady state and non-steady state signals is not a method that is helpful when applying the multi-method approach here.To clarify what signal differing data-based techniques are actually constraining, we list them in Table 1.
Most of the ocean-based tracer techniques quantify the steady state anthropogenic CO 2 signal alone ( ACO 2 ), while the atmospheric techniques quantify the total timeevolving net CO 2 oceanic signal ( DIC Net ).Although each method has inherent uncertainties and biases, there is powerful information in treating them as independent, whereby there difference theoretically should constrain the combined non-steady state response.
For an ocean that is changing with climate change and decadal variability, combining steady state methods with total net methods could provide a powerful way to quantify how the oceanic CO 2 sink is actually evolving.For example, the total net CO 2 sink (i.e.DIC Net in Eq. 4) is best captured from two different techniques: the O 2 /N 2 atmospheric technique and surface ocean pCO 2 climatologies.On the other hand, CFC-based and ocean inversion methods are the most accurate techniques to capture the steady state anthropogenic CO 2 inventory between two different periods of time.In theory, by comparing the results of these different techniques should produce a testable signal equivalent to the non-steady state oceanic CO 2 change.
The expected steady state anthropogenic CO 2 uptake in the 1990s from oceanic inversions and CFCs is 2-2.2Pg C yr −1 (see Table 2).The total net CO 2 uptake for the 1990s from oceanic pCO 2 climatology and atmospheric O 2 /N 2 methods is 1.7-1.9Pg C yr −1 .Solving equation 4 implies that the difference between these constraints is the nonsteady state CO 2 signal, implying an outgassing of 0.1 to www.biogeosciences.net/10/2219/2013/Biogeosciences, 10, 2219-2228, 2013 Table 1.Data-based techniques to quantify "anthropogenic" CO 2 storage in the ocean.
The application of the multi-method approach, however, illustrates the potential benefit of using such an approach but also challenges us to obtain better observations to reduce these uncertainties amongst the suite of different oceanic CO 2 uptake data-based techniques.It is interesting, however, that models are suggesting a non-steady state CO 2 outgassing (∼ +0.4 Pg C yr −1 ) which is nearing the uncertainty limits for a multi-method approach (i.e.±0.6-0.8Pg C yr −1 ), implying that into the future, with greater certainty, such a non-steady state CO 2 signal could become observationally statistically significant.
We formulate a revised oceanic carbon budget for the 1989-2007 period (Fig. 2) that takes into account the ∼ +0.4 Pg C yr −1 non-steady state CO 2 outgassing predicted by both the models and somewhat tentatively by the multimethodological constraint illustrated here.

Challenges to reducing uncertainty
The key limitation to this multi-methodological approach today is the current large uncertainty of the different databased techniques to quantify anthropogenic CO 2 uptake in the ocean.It is not necessarily important which combination of technique is used, but rather we have certainty over its assumptions and application.
Many of the techniques suffer from a lack of measurements, which can be rectified in the future.For example, the pCO 2 database, although good coverage exists in the Northern Hemisphere, complex regimes like the equatorial Pacific and Southern Oceans have large gaps in coverage.However, with autonomous CO 2 measurements increasing, this will change.
One complication with pCO 2 climatologies is associated with the natural outgassing of carbon that enters the ocean via rivers and estuaries.This is an uncertain but necessary constraint for the oceanic carbon budget when using pCO 2 climatologies or oceanic inversions (Jacobson et al., 2007;Gruber et al., 2009).From an atmospheric perspective, this riverine CO 2 outgassing is a steady state signal since the coinciding uptake of CO 2 occurs on land through biomass production.However, from the "raw" pCO 2 climatologies, the riverine CO 2 outgassing needs to be added to the final global ocean estimate.This riverine CO 2 outgassing is estimated to be +0.45Pg C yr −1 (Jacobson et al., 2007) with a 50 % uncertainty.The magnitude of this riverine CO 2 outgassing would dampen this techniques ability to detect the non-steady state CO 2 signal, unless that is constrained to a much higher accuracy.
Other techniques like the modern application of CFCtracers (Waugh et al., 2006;Khatiwala et al., 2009)    technique could go one-step further to constrain the nonsteady state circulation based changes to anthropogenic CO 2 , although any changes to the biological pump or natural carbon would be missed.

Conclusions
Here we decompose the time evolution of net CO 2 changes in the ocean to clarify the notion of "anthropogenic" CO 2 uptake.Traditionally, oceanic tracer-based techniques have constrained net oceanic storage of CO 2 associated with elevated CO 2 concentrations in the atmosphere under the as-sumption of a steady state ocean carbon cycle.Both climatechange and decadal changes to the oceanic CO 2 cycle shown recently have clearly marked a new era of non-steady state conditions for CO 2 that was already known from other biogeochemical parameters like oxygen.This global nonsteady state CO 2 signal is estimated to have outgassed about 6.3 Pg C of CO 2 (or ∼ 3 ppm to the atmosphere) between 1989 and 2007, which is ∼ 18 % of the net oceanic CO 2 uptake rate estimated from models.
After illustrating the different components of the timeevolving oceanic CO 2 sink, we present a simple concept to estimate the non-steady state oceanic CO multi-methodological budget approach, we estimate a 0.1-0.5 Pg C yr −1 outgassing over the last two decades, however, the uncertainty across the suite of different data-based techniques is too large at present (∼ 0.4-0.6Pg C yr −1 ) to provide a significant non-zero estimate of the non-steady state signal.
Although the non-steady state CO 2 signal currently provides a positive feedback to atmospheric CO 2 levels, the future direction and magnitude of the signal is not clear, since some models suggest that recent Southern Ocean outgassing will eventually reverse in the future and absorb greater CO 2 than expected from steady state conditions (Zickfeld et al., 2008).
In the end, atmospheric CO 2 levels only change via the total time-evolving CO 2 changes in the ocean, which based on recent evidence is now entering a new non-steady state mode.Given this emerging mode, the challenge for the observational community is to reduce the uncertainty across a suite of independent data-based techniques to enable the clear separation between the net storage of CO 2 in the ocean and the steady state anthropogenic CO 2 signal on decadal time frames.By embracing more accurate and diverse techniques, we can better help detect how a changing ocean is modifying rising atmospheric CO 2 levels.
Fig.1.Illustration of the time-evolving net CO 2 uptake from the ocean since 1970 comparing the expected steady state uptake that assumes constant circulation and biology (black line) with the mean net uptake from five different ocean models that include time-evolving natural forcings(Wetzel et al., 2005;Le Quéré et al., 2007;Lovenduski et al., 2007;Matear and Lenton, 2008; Rodgers et al., 2008).We include the results from the hindcast CSIRO climate model fromMatear and Lenton (2008) and the suite of model projections documented inSarmiento et al. (2010).The expected steady state CO 2 uptake is taken fromMikaloff-Fletcher et al. (2006) as documented bySarmiento et  al. (2010), normalized to the decade of 1970-1980.The average net CO 2 uptake (red dots) was taken as a combined mean between the five different models with minimum and maximum bounds shaded in light red.This is an illustrative figure used to demonstrate the time-evolving changes in the various CO 2 signals within the ocean as described in Eq. 4. The integrated dotted red curve is the total net CO 2 uptake by the ocean ( DIC Net ), which is about 36 Pg C between 1989 and 2007.The black curve is the expected steady state CO 2 uptake or what has been historically called anthropogenic CO 2 uptake ( ACO 2 ), which is about 42 Pg C between 1989 and 2007.The difference between these curves is therefore the combined non-steady state CO 2 loss to the atmosphere ( ACO 2 + DIC Nat ), which comes to about 6.3 Pg C outgassing to the atmosphere between 1989 and 2007 (see Table2ofSarmiento et al., 2010).

Fig. 2 .
Fig. 2. Estimated non-steady state global carbon budget for 1989-2007 combining oceanic steady state observational estimates with the oceanic non-steady state signal as diagnosed from data and models in this study.
2 signal and determine the net change in carbon stored in the ocean.With a and R. J. Matear: The non-steady state oceanic CO 2 signal

www.biogeosciences.net/10/2219/2013/ Biogeosciences, 10, 2219-2228, 2013 the
postulated a stable upper ocean would also result in a more efficient (at least initially) biological drawdown of CO 2 from surface to the interior.Based on a suite of ocean model simulations, he estimated a net carbon drawdown of +6 Pg C up until 1994.Different non-steady state anthropogenic processes (ocean warming and stratification/biological export) partially offset each other, resulting in a final combined estimate of 7 ± 10 Pg C for ACO 2 from

Biogeosciences, 10, 2219-2228, 2013 www
are not as data-limited and provide the most accurate way of constraining the steady state anthropogenic CO 2 signal over decadal timescales.With temporal CFC or tracer measurements, this .biogeosciences.net/10/2219/2013/