Explaining the eventual transient saturation of climate-carbon cycle feedback
© Mokhov and Eliseev; licensee BioMed Central Ltd. 2008
Received: 19 December 2007
Accepted: 28 April 2008
Published: 28 April 2008
Coupled climate-carbon cycle simulations generally show that climate feedbacks amplify the buildup of CO2 under respective anthropogenic emission. The effect of climate-carbon cycle feedback is characterised by the feedback gain: the relative increase in CO2 increment as compared to uncoupled simulations. According to the results of the recent Coupled Climate-Carbon Cycle Model Intercomparison Project (C4MIP), the gain is expected to increase during the 21st century. This conclusion is not supported by the climate model developed at the A.M. Obukhov Institute of Atmospheric Physics at the Russian Academy of Sciences (IAP RAS CM). The latter model shows an eventual transient saturation of the feedback gain. This saturation is manifested in a change of climate-carbon cycle feedback gain which grows initially, attains a maximum, and then decreases, eventually tending to unity.
Numerical experiments with the IAP RAS CM as well as an analysis of the conceptual framework demonstrate that this eventual transient saturation results from the fact that transient climate sensitivity decreases with time.
One may conclude that the eventual transient saturation of the climate-carbon cycle feedback is a fundamental property of the coupled climate-carbon system that manifests itself on a relevant time scale.
In , the following expression for climate-carbon cycle feedback gain g = (f - 1)/f is derived
g = -α (γ l + γ oc )/(c0 + β l + β oc ) = -αG (2)
where t stands for time.
In turn, a non-monotonic change of f in the simulation with the IAP RAS climate model of intermediate complexity (IAP RAS CM) under the same scenario SRES A2 was obtained [11, 15]. In this simulation, parameter f grows during the most part of the run but starts to decrease late in the 21st century. Futher, based on a conceptual climate-carbon cycle model, it has been demonstarted that plausible physical reason for this eventual transient saturation is a weak, logarithmic dependence of the CO2 radiative forcing on the atmospheric concentration of carbon dioxide . This leads to small influence of an additional (due to climate-carbon cycle interactions) build up of the carbon dioxide on climate state when C is large. Given a length of integration, the latter condition is fulfilled only if emissions are aggressive enough.
However, the build up of the carbon dioxide in the atmosphere in the uncoupled simulation was prescribed in  and it is unclear how to relate directly time scales associated with this build up with time scales associated with changes in CO2 emissions. In the present paper, ensemble simulations with an Earth system model IAP RAS CM [10–12, 15, 17] forced by idealised emission scenarios are performed. In this, it is demonstrated that an eventual transient climate-carbon cycle saturation may be exhibited also under moderate emission scenarios if respective integration proceeds for a sufficient time. The results obtained in  are supported and it is shown that the above-mentioned hypothesis is proved and even weak but continuing emissions lead to eventual saturation of the climate-carbon cycle feedback. In addition, the IAP RAS CM simulations are supported by integrations with a conceptual coupled model. The latter model is similar, but not identical to that used in . In particular, in the model used here, emissions of CO2 are used instead of prescribing the atmospheric carbon dioxide build up in the uncoupled simulation. It is also demonstrated that eventual transient saturation of climate-carbon cycle feedback in the IAP RAS CM is consistent with that derived from the earlier version of conceptual model.
Results and Discussion
Analysis of a numerical coupled model
A series of numerical experiments with two IAP RAS CM versions were performed. In one version, oceanic uptake of anthropogenic carbon is formulated as a bilinear function of time derivatives of atmospheric concentration of CO2 and globally averaged annual mean sea surface temperature. Hereafter, this version is denoted as FocL. In other version, oceanic uptake of carbon F oc is determined employing a nonlinear model  but with chemical constants computed as functions of temperature in accordance to . In particular, this version takes into account carbonate dissolution in the ocean. Hereafter, this version is denoted as FocNL. Terrestrial uptakes for both model versions are determined based on a zero-dimensional module taking into account direct plants fertilisation by CO2 and influence of climate (expressed via temperature anomaly from the model's preindustrial state employing Q10-relationships) on biospheric productivity and autotrophic and heterotrophic respiration. Half-saturation point for the direct plant fertilisation was set to 150 ppmv in the version FocL and to 460 ppmv in the FocNL. Performance and extensive comparison of these two model versions between each other and with observational estimates are described in [10–12, 15, 17]. There, it is shown that both versions behave realistically in the 20th century and their projections for the 21st century are basically in line with those obtained in the C4MIP project. The simulations performed here were forced by idealised emission scenarios. In these scenarios, fossil fuel emissions increase exponentially in time with a time scale t p from 25 yr to 250 yr starting from the small initial value E0 = 0.1 PgC/yr. These emissions are not meant to represent any historical or future projected emissions. However, one notes that the combined fossil fuel emissions taken as historical for the 19–20th centiries  and adopted from the Special Report on Emission Scenarios  for the 21st century (except the scenario B1 where fossil fuel emissions decline in the late 21st century) may be approximated by an exponential curve with t p changing from 50 yr to 200 yr depending on scenario. One notes that an emission intensity depends both on E0 and t p . Thus, strictly speaking, quantitative results presented below are only valid for this particular choice of E0. However, qualitative conclusions are unlikely to be changed if other sufficiently small value of E0 is selected. In some sense, larger t p may be partly compensated by larger E0. In the performed simulations, land use emissions are neglected. Every simulation starts from the model's preindustrial equilibrated state and proceeds for 1, 500 yr.
As shown in Fig. 1, climate-carbon cycle feedback parameter attains maximum and than eventualy falls down tending to unity. This behaviour is in general agreement with those obtained in the earlier IAP RAS CM simulations [11, 15] and employing a conceptual climate-carbon cycle model . There is a difference in timing t m when f is at maximum between the two employed here IAP RAS CM versions. For the studied here range of t p , t m ranges from 3 × t p to 5 × t p in simulations with the FocL and from 5 × t p to 8 × t p in simulations with the FocNL. For both model versions, the larger t p the earlier (in units of t p ) this maximum of f occurs. Difference in responses of the globally and annually averaged surface air temperatures between coupled and uncoupled simulation also increases early in the simulations, then attains a maximum (which occurs later than t m ) and diminishes afterwards tending to zero (not shown). To diagnose the behavior of the model, an approach by  has been adopted. In this, coefficients γ X and β X (X = l, oc) entering (2) were computed for the running 100-yr segments. For both model versions, β l increases till the plants fertilisation half-saturation point is reached and then decreases to zero (see Fig. 2a as an example). This behaviour reflects the Michaelis-Menten-type dependence of terrestrial productivity on as it is implemented in the IAP RAS CM. Coefficient β oc behaves similarly in the version FocNL while changes in time only slightly in the version FocL. This is expected from the formulations of F oc in these model versions. In contrast, the magnitudes of γ l and γ oc increse during the course of integration (Fig. 2b).
Analysis of a conceptual model
with κ = -G = (γ l + γ oc )/(c0 + β l + β oc ), see Eq. (2). For the C4MIP models under the emission scenario SRES A2, from  one derives κ = (8 – 48) ppmv/K for the 21st century. Coefficient β = β l + β oc varies from 1.7 PgC/ppmv to 3.7 PgC/ppmv with a mean value 2.5 PgC/ppmv . For the the IAP RAS CM ensemble simulation, κ depends on t p . Here, at a time short before t m , κ varies from very small values for t p = 25 yr up to 25 ppmv/K (35 ppmv/K) for the version FocNL (FocL) if t p = 250 yr. Coefficient β depends on t p as well. At time of a few t p , it again attains very small values for t p = 25 yr and amounts 1.6 PgC/ppmv (2.5 PgC/ppmv) for the version FocL (FocNL) for t p = 250 yr.
with the initial atmospheric carbon dioxide concentration C0.
Eqs. (6), (7) and (9) have been numerically integrated subject to the above-mentioned exponential emissions (expressed via (7)) and to initial condition . An ensemble of conceptual model integrations is performed varying t p in the same range as it was for IAP RAS CM and varying β in the range from 0 to 5.0 PgC/ppmv (this range is wider than the corresponding C4MIP range, see above). Equilibrium model sensitivity to doubling of the carbon dioxide in the atmosphere was varied between 1 K and 9 K roughly corresponding to combined range from [25–28] (in , a narrower range from 2.0 K to 4.5 K is figured). Parameter κ was varied from 8 ppmv/K to 48 ppmv/K (see above).
where again the threshold value for f is set to 1.05. This equation has been solved numerically appling the dichotomy method. In general, the larger κ and/or the larger . For = 2.2 K, the values obtained for the conceptual model are similar, but slightly larger, to those estimated from the IAP RAS CM runs.
The value of E0 selected in this paper, 0.1 PgC/yr, fulfills this condition. If (11) is satisfied then qualitative conclusion of the present paper and the respective conclusions related to the order of magnitudes of basic characteristics are expected to remain unchanged.
An eventual transient saturation, as studied in the present framework, implies that initially (at a time ≪ t p ) and eventually (at a time ≫ t p ), the state of the coupled system is close to the state of the uncoupled one. However, during an intermediate period when t' ≃ t p , f deviates substantinally from unity and climate-carbon cycle interaction does matter. Taking into account typical time scales of CO2 emission growth for the SRES emission, t p = 50 – 200 yr, this intermediate period extends for the next several centuries.
While it is beyond the scope of the present paper to concern CO2 emissions changing non-exponentially in time, it is possible to make a note with this respect. In particular, if the future emissions would change in time slower than the exponential ones, it would be expressed in apparent increase of t p during the course of simulation. In turn, the timing t m of maximum of f and an eventual transient saturation would be delayed in comparison to the case with constant t p . On the other hand, for faster-than-exponential emissions this eventual transient saturation would come closer to the present day.
The long-term fate of anthropogenic carbon is basically governed by calcium carbonate dissolution in the ocean [30, 31]. This effect is taken into account only in the version FocNL and neglected in the version FocL and in the conceptual model. However, carbonate dissolution is important only for time scales larger than ≈ 5, 000 yr  which is well above the length of every individual simulation performed in the present paper (1, 500 yr). As a result, the version FocL and the conceptual climate-carbon cycle model are applicable for the problem considered here as well.
Returning to C4MIP simulations, one notes that it is possible to find some hints for this eventual transient saturation in these simulations  as well (see Background). However, in the C4MIP integrations, an eventual climate-carbon cycle carbon saturation is less marked. The most probable reason for this is the length of the C4MIP simulations which is too small to make this saturation visible given an emission intensity in these simulations. One expects that if these simulations would extended to the future, eventually, parameter f would converge to unity in the C4MIP ensemble as well. Some complications for this convergence could arise from a specific behaviour of a for particular models, e.g., due to behaviour of the oceanic heat uptake in the course of integration and its interrelation with equilibrium temperature sensitivity [32–35].
As a final note, one may distinguish manifestation of a climate-carbon cycle feedback eventual transient saturation in terms of feedback parameter, on one hand, and in terms of feedback gain, on the other one. In particular, as dg/dt = f-2df/dt and, generally, f > 1 in the course of simulations, this saturation is somewhat masked for g in comparison to f. This also adds to the masking of the eventual transient saturation of the climate-carbon cycle feedback in the C4MIP simulations.
In this paper, simulations with the IAP RAS climate model of intermediate complexity have been performed to study temporal variations of the climate-carbon cycle feedback parameter. Two model versions were considered differing between each other by the formulation of the oceanic uptake of carbon dioxide and by the governing parameters of the terrestrial carbon uptake module. Both versions were forced by idealised scenarios of fossil fuel emissions. In these scenarios, emissions grow exponentially in time (with a characteristic timescale t p = 25 – 250 yr) starting from a small initial emission value. Land use emissions were set to zero. In all simulations with both model versions, climate-carbon cycle feedback parameter f grows initially, attains maximum, and then decreases eventually tending to unity. The timing of this maximum is of the order of a few t p for the selected small value of initial emission. This general behaviour is consistent to that obtained in a earlier simulations with the same model.
In particular, coefficients of climate-carbon cycle interaction diagnosed for 100-yr running segments lead to increase of climate-carbon cycle gain g during the course of integration. This increase, however, is overcompensated by decrease of transient climate sensitivity α. The latter leads to the above-mentioned eventual transient saturation of climate-carbon cycle feedback.
The IAP RAS CM simulations are supported by an analysis of a conceptual model with linear dependence of carbon sinks from the atmosphere on changes in atmospheric carbon dioxide concentration and global temperature rise. This simple model exhibits an eventual transient saturation of the climate-carbon cycle feedback which is similar to the model of intermediate complexity.
The authors are indebted to L.L. Golubyatnikov and A.M. Tarko for useful discussions on the topic of the manuscript. Comments by anonymous reviewers has led to improved presentation of the results. This work has been supported by the Russian Foundation for Basic Research (grants 05-05-64907, 05-05-65167, and 07-05-00273), the President of Russia grant 4166.2005.5, and by the Programs of the Russian Ministry for Science and Education, Russian Federal Agency for Science and Innovations and the Russian Academy of Sciences (contracts 02.515.11.5031 and 02.515.11.5046).
- Cox P, Betts R, Jones C, Spall S, Totterdell I: Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model. Nature 2000, 408: 184–187. 10.1038/35041539View ArticleGoogle Scholar
- Friedlingstein P, Bopp L, Ciais P, Dufresne JL, Fairhead L, Le Treut H, Monfray P, Orr J: Positive feedback between future climate change and the carbon cycle. Geophys Res Lett 2001,28(8):1543–1546. 10.1029/2000GL012015View ArticleGoogle Scholar
- Brovkin V, Bendtsen J, Claussen M, Ganopolski A, Kubatzki C, Petoukhov V, Andreev A: Carbon cycle, vegetation, and climate dynamics in the Holocene: Experiments with the CLIMBER-2 model. Glob Biogeochem Cycles 2002,16(4):1139. 10.1029/2001GB001662View ArticleGoogle Scholar
- Brovkin V, Sitch S, von Bloh W, Claussen M, Bauer E, Cramer W: Role of land cover changes for atmospheric CO 2 increase and climate change during the last 150 years. Glob Change Biol 2004, 10: 1253–1266. 10.1111/j.1365-2486.2004.00812.xView ArticleGoogle Scholar
- Friedlingstein P, Dufresne JL, Cox P, Rayner P: How positive is the feedback between climate change and the carbon cycle? Tellus 2003,55B(2):692–700.View ArticleGoogle Scholar
- Jones C, Cox P, Essery R, Roberts D, Woodage M: Strong carbon cycle feedbacks in a climate model with interactive CO 2 and sulphate aerosols. Geophys Res Lett 2003,30(9):1479. 10.1029/2003GL016867View ArticleGoogle Scholar
- Matthews H, Weaver A, Meissner K, Gillett N, Eby M: Natural and anthropogenic climate change: incorporating historical land cover change, vegetation dynamics and the global carbon cycle. Clim Dyn 2004, 22: 461–479. 10.1007/s00382-004-0392-2View ArticleGoogle Scholar
- Matthews H, Weaver A, Meissner K: Terrestrial carbon cycle dynamics under recent and future climate change. J Climate 2005,18(10):1609–1628. 10.1175/JCLI3359.1View ArticleGoogle Scholar
- Govindasamy B, Thompson S, Mirin A, Wickett M, Caldeira K, Delire C: Increase of carbon cycle feedback with climate sensitivity: results from a coupled climate and carbon cycle model. Tellus 2005,57B(2):153–163.View ArticleGoogle Scholar
- Mokhov I, Eliseev A, Karpenko A: Sensitivity of the IAP RAS Global Climatic Model with an interactive carbon cycle to anthropogenic influence. Doklady Earth Sci 2006,407(3):424–428. 10.1134/S1028334X06030172View ArticleGoogle Scholar
- Eliseev A, Mokhov I, Karpenko A: Climate and carbon cycle variations in the 20th and 21st centuries in a model ofintermediate complexity. Izvestiya, Atmos Ocean Phys 2007, 43: 1–14. 10.1134/S000143380701001XView ArticleGoogle Scholar
- Eliseev A, Mokhov I: Carbon cycle-climate feedback sensitivity to parameter changes of a zero-dimensional terrestrial carbon cycle scheme in a climate model of intermediate complexity. Theor Appl Climatol 2007,89(1–2):9–24. 10.1007/s00704-006-0260-6View ArticleGoogle Scholar
- Bala G, Caldeira K, Mirin A, Wickett M, Delire C: Multicentury changes to the global climate and carbon cycle: Results from a coupled climate and carbon cycle model. J Climate 2005,18(21):4531–4544. 10.1175/JCLI3542.1View ArticleGoogle Scholar
- Friedlingstein P, Cox P, Betts R, Bopp L, von Bloh W, Brovkin V, Doney S, Eby M, Fung I, Govindasamy B, John J, Jones C, Joos F, Kato T, Kawamiya M, Knorr W, Lindsay K, Matthews H, Raddatz T, Rayner P, Reick C, Roeckner E, Schnitzler KG, Schnur R, Strassmann K, Weaver A, Yoshikawa C, Zeng N: Climate-carbon cycle feedback analysis: Results from the C 4 MIP model intercomparison. J Climate 2006,19(22):3337–3353. 10.1175/JCLI3800.1View ArticleGoogle Scholar
- Mokhov I, Eliseev A, Karpenko A: Decadal-to-centennial scale climate-carbon cycle interactions from global climate models simulations forced by anthropogenic emissions. In Climate Change Research Progress. Edited by: Peretz L. Hauppauge, NY: Nova Publishers; 2008:in press.Google Scholar
- Eliseev A, Mokhov I: Eventual saturation of the climate-carbon cycle feedback studied with a conceptual model. Ecol Mod 2008, 213: 127–132. 10.1016/j.ecolmodel.2007.11.015View ArticleGoogle Scholar
- Mokhov I, Eliseev A, Demchenko P, Khon V, Akperov M, Arzhanov M, Karpenko A, Tikhonov V, Chernokulsky A, Sigaeva E: Climate changes and their assessment based on the IAP RAS global model simulations. Doklady Earth Sci 2005,402(4):591–595.Google Scholar
- Bacastow R: Numerical evaluation of the evasion factor. In Carbon Cycle Modelling, SCOPE-16. Edited by: Bolin B. N.Y.: J. Wiley and Sons; 1981:95–101.Google Scholar
- Millero F: Thermodynamics of carbon dioxide system in the ocean. Geophys Cosmophys Acta 1995,59(4):661–677.Google Scholar
- Marland G, Boden T, Andres R: Global, regional, and national CO2 emissions. In Trends: A Compendium of Data on Global Change. Oak Ridge, Tenn.: Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy; 2005.Google Scholar
- Houghton J, Ding Y, Griggs D, Noguer M, Linden P, Dai X, Maskell K, Johnson C, (Eds): Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge/New York: Cambridge University Press; 2001.Google Scholar
- Gregory J, Stouffer R, Raper S, Stott P, Rayner N: An observationally based estimate of the climate sensitivity. J Climate 2002,15(22):3117–3121. Publisher Full Text 10.1175/1520-0442(2002)015<3117:AOBEOT>2.0.CO;2View ArticleGoogle Scholar
- Myhre G, Highwood E, Shine K, Stordal F: New estimates of radiative forcing due to well mixed greenhouse gases. Geophys Res Lett 1998,25(14):2715–2718. 10.1029/98GL01908View ArticleGoogle Scholar
- Andreae M, Jones C, Cox P: Strong present-day aerosol cooling implies a hot future. Nature 2005, 435: 1187–1190. 10.1038/nature03671View ArticleGoogle Scholar
- Harvey L, Kaufmann R: Simultaneously constraining climate sensitivity and aerosol radiative forcing. J Climate 2002,15(20):2837–2861. Publisher Full Text 10.1175/1520-0442(2002)015<2837:SCCSAA>2.0.CO;2View ArticleGoogle Scholar
- Stainforth D, Aina T, Christensen C, Collins M, Faull N, Frame D, Kettleborough J, Knight S, Martin A, Murphy J, Piani C, Sexton D, Smith L, Spicer R, Thorpe A, Allen M: Uncertainty in predictions of the climate response to rising levels of greenhouse gases. Nature 2005, 433: 403–406. 10.1038/nature03301View ArticleGoogle Scholar
- Forest C, Stone P, Sokolov A: Estimated PDFs of climate system properties including natural and anthropogenic forcings. Geophys Res Lett 2006, 33: L01705. 10.1029/2005GL023977View ArticleGoogle Scholar
- Hegerl G, Crowley T, Hyde W, DJ F: Climate sensitivity constrained by temperature reconstructions over the past seven centuries. Nature 2006, 440: 1029–1032. 10.1038/nature04679View ArticleGoogle Scholar
- Solomon S, Qin D, Manning M, Marquis M, Averyt K, Tignor M, LeRoy Miller H, Chen Z, (Eds): Climate Change 2007: The Physical Science Basis. Cambridge/New York: Cambridge University Press; 2007.Google Scholar
- Archer D, Martin P, Buffett B, Brovkin V, Rahmstorf S, Ganopolski A: The importance of ocean temperature to global biogeochemistry. Earth Planet Sci Lett 2004,222(2):333–348. 10.1016/j.epsl.2004.03.011View ArticleGoogle Scholar
- Scheffer M, Brovkin V, Cox P: Positive feedback between global warming and atmospheric CO 2 concentration inferred from past climate change. Geophys Res Lett 2006,33(10):L10702. 10.1029/2005GL025044View ArticleGoogle Scholar
- Hansen J, Russell G, Lacis A, Fung I, Rind D, Stone P: Climate response times: dependence on climate sensitivity and ocean mixing. Science 1985,229(4716):857–859. 10.1126/science.229.4716.857View ArticleGoogle Scholar
- Harvey L: Effect of ocean mixing on the transient climate response to a CO 2 increase: analysis of recent model results. J Geophys Res 1986, 90: 2709–2718. 10.1029/JD091iD02p02709View ArticleGoogle Scholar
- Schlesinger M, Jiang X: Simple model representation of atmosphere-ocean GCMs and estimation of the time scale of CO 2 -induced climate change. J Climate 1990,3(12):1297–1315. Publisher Full Text 10.1175/1520-0442(1990)003<1297:SMROAO>2.0.CO;2View ArticleGoogle Scholar
- Raper S, Gregory J, Stouffer R: The role of climate sensitivity and ocean heat uptake on AOGCM transient temperature and thermal expansion response. J Climate 2002, 15: 124–130. Publisher Full Text 10.1175/1520-0442(2002)015<0124:TROCSA>2.0.CO;2View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.