1. Introduction
The spread of the equilibrium or transient surface temperature response to a CO2 doubling as predicted by atmosphere–ocean coupled models is still large (Meehl et al. 2007), and an open question is to identify the primary sources of this spread. Global warming estimates depend on radiative forcing, feedback processes that may either amplify or dampen the climate response and, in the transient case, ocean heat uptake. For individual models, it has been suggested that atmospheric processes were the most critical factors for estimating global temperature changes in transient simulations (e.g., Williams et al. 2001; Meehl et al. 2004; Collins et al. 2007). Here our purpose is to investigate whether these results extend to multimodel ensembles, and how much the various feedbacks and the ocean heat uptake contribute to the multimodel mean and spread of global warming estimates. The main radiative feedbacks are associated with changes in water vapor (WV), temperature lapse rate (LR), clouds, and surface albedo. The associated feedback parameters have been diagnosed for some multimodel ensembles (e.g., Colman 2003; Soden and Held 2006; Webb et al. 2006), but they have not been translated into temperature changes. This makes it difficult to compare the temperature change associated with each feedback with that from other processes, such as the ocean heat uptake.
In this paper we show that it is possible to decompose, and thus to compare, the contributions of the different climate feedbacks, and eventually of the ocean heat uptake, to the global temperature response to a specified forcing. After a brief presentation of the feedback analysis framework (section 2), the decomposition methodology is presented (section 3) and, after gathering the required data (feedback parameters, radiative forcing, and ocean heat uptake; section 4), this methodology is applied to an ensemble of models that participated in the World Climate Research Programme’s (WCRP’s) third phase of the Coupled Model Intercomparison Project (CMIP3) in support of the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4; section 5). There is very little in this paper that is entirely new. Rather, we propose a new presentation of existing results that allows us to quantify in a more straightforward way the relative contribution of different processes to intermodel differences in global mean temperature changes.
2. The feedback analysis framework
3. Relative contribution of each feedback to the global temperature change
a. Equilibrium temperature change
b. Transient temperature change
4. CMIP3/AR4 atmosphere–ocean GCMs
We now apply the above decomposition to the global surface temperature response to a CO2 doubling predicted by an ensemble of 12 coupled atmosphere–ocean GCMs (AOGCMs) participating in the CMIP3/AR4 (Meehl et al. 2005; Randall et al. 2007a). For this purpose, for each model we need the global mean values of the radiative forcing, the climate feedback parameters, and the ocean heat uptake efficiency.
a. 2 × CO2 radiative forcing
In this study we use the radiative forcing for a CO2 doubling reported by Forster and Taylor (2006) and Randall et al. (2007b). These forcings have been computed after stratospheric adjustment in all-sky conditions and are averaged over the globe for a year (Table 1). For the 12 GCMs considered here, the multimodel average of the net radiative forcing (3.71W m−2) is very close to previous Myhre et al. (1998) results, and the relative intermodel standard deviation is about 6% (Table 2).
In another intercomparison study, for 16 GCMs Collins et al. (2006) obtained an intermodel spread of the net radiative forcing as large as 15% (Randall et al. 2007a). These forcing have been computed at 200 hPa for a unique atmospheric profile (midlatitude summer climatological conditions) in clear-sky conditions, without any stratospheric adjustment. When compared with Forster and Taylor (2006) results, the relative values of the intermodel standard deviation of the longwave (LW) forcing are similar in both studies (8%, see Table 1). This is not the case in the shortwave (SW) domain, and the difference is even larger for the net radiative forcing. In the results of Collins et al. (2006), as reported by Randall et al. (2007a), the standard deviation of the net forcing is larger than the quadratic sum of the standard deviation of the SW and LW forcings, which indicates that the SW and LW intermodel differences are positively correlated. The opposite is found in Forster and Taylor (2006), which indicates that the error in the SW and LW domains are anticorrelated, and that stratospheric adjustment can explain part of it. We believe that the intermodel spread of the forcing reported by Forster and Taylor (2006) is the most relevant for our study because the global warming estimates are derived from global simulations, including clouds and a stratospheric temperature response.
b. Feedback parameters
As reviewed by different authors (e.g., Soden et al. 2004; Stephens 2005; Bony et al. 2006), several approaches have been followed to decompose the total feedback parameter into its several components (water vapor, clouds, surface albedo, etc.), with each method having its own strengths and weaknesses. Soden and Held (2006) computed these feedback parameters for 12 CMIP3/AR4 models (Table 1), using the Special Report on Emissions Scenarios (SRES) A1B simulations, and their results are fairly consistent with previous results obtained by Colman (2003) with older GCMs (cf. Bony et al. 2006). The multimodel mean and standard deviation of the total feedback parameters (
c. Ocean heat uptake efficiency
We computed the ocean heat uptake efficiency κ using Eq. (12). For each model, the TOA flux Ft and the surface air temperature Ts were averaged over the 20-yr period centered at the time of CO2 doubling, that is, year 70 for the 1% yr−1 simulation. The differences with the corresponding period of the control simulation were performed and the values of κ reported in Table 1.
d. Representativity of the ensemble of models considered
Using the values reported in Table 1, the equilibrium and transient temperature changes are computed for each of the 12 models as ΔTes = −ΔQt/λ and ΔTts = −ΔQt/(λ + κ), respectively. This leads to a multimodel mean ±1 standard deviation of the equilibrium temperature change of 3.1° ± 0.7°C. These numbers are comparable with those of the AR4 equilibrium climate sensitivity estimates derived from 18 atmospheric GCMs coupled to slab oceans (3.3° ± 0.7°C; see Meehl et al. 2007). For the transient temperature change, we obtain 2.0° ± 0.3°C, which is closed to the AR4 values reported on the basis of 19 coupled atmosphere–ocean GCMs (1.8° ± 0.3°C; see Meehl et al. 2007). As far as global temperature change is concerned, the subset of 12 models considered here is therefore representative of the larger set of CMIP3/AR4 models.
5. Results
a. Decomposition of equilibrium temperature changes
The multimodel mean of the equilibrium temperature change and the contributions associated with the Planck response [Eq. (4)] and each feedback [Eq. (10)], computed for a reference radiative forcing, are shown in Fig. 1a and reported in Table 3. On average, for the set of models considered here, the Planck response represents about a third of the total temperature response (1.2° versus 3.1°C), while climate feedbacks account for two-thirds of it. The increase of water vapor with warming enhances the absorption of longwave radiation and enhances the warming by 1.7°C. Lapse-rate changes are associated with a negative feedback, resulting from the moist-adiabatic structure of the tropical atmosphere. Because of the strong anticorrelation between these two feedbacks, it is convenient to consider the sum of both of them (WV + LR; Soden and Held 2006). This combined feedback increases the temperature by 0.9°C, which is slightly less than the Planck response. The cloud feedback’s contribution to the warming is, on average, slightly weaker than that of the WV + LR feedback, and the surface albedo feedback’s contribution is the smallest.
However, Fig. 2 shows that for each feedback there are some intermodel differences, especially for the cloud feedback contribution, and that the amplitude of the equilibrium temperature change is primarily driven by the cloud feedback component. This appears also clearly when considering the intermodel standard deviation of the temperature change resulting from each feedback normalized by the intermodel standard deviation of the total temperature change (Fig. 1b). The standard deviation resulting from cloud feedback represents nearly 70% the standard deviation of the total temperature change. The temperature spread resulting from the radiative forcing is comparable to the spread resulting from the WV + LR feedback and the spread resulting from the surface albedo feedback is the smallest.
b. Decomposition of transient temperature changes
The transient temperature changes (or TCR) from individual GCMs, as well as the contribution of the various feeedbacks, are displayed in Fig. 3. The multimodel mean and standard deviation are displayed in Fig. 4 and reported in Table 3. The temperature damping resulting from the ocean heat uptake is about −0.4°C, and its absolute value is comparable to the multimodel contributions of the WV + LR (0.6°C) and cloud (0.4°C) feedback. The mean transient temperature change is nearly 2/3 of that at equilibrium; therefore, the transient temperature changes associated with each feedback scale with it [cf. Eq. (14)]. The intermodel standard deviation of the temperature change resulting from cloud feedback represents nearly 90% of the standard deviation of the total temperature change (Fig. 4b). Similarly for the equilibrium case, cloud feedbacks thus constitute the main source of spread of the transient temperature response among GCMs. The WV + LR feedback, the ocean heat uptake, and the radiative forcing constitute secondary and roughly comparable sources of spread, and the surface albedo feedback constitutes the smallest one.
The intermodel standard deviation of the global temperature change may also be normalized with the multimodel mean global temperature change. This relative standard deviation is comparable in both equilibrium and transient conditions; the spread in equilibrium is slightly larger (23% versus 16%). The same holds for the relative standard deviation of the temperature change associated with each feedback. Therefore, the contribution of the various feedbacks to the total spread is, in relative terms, as important in the transient case as in the equilibrium case.
6. Summary and conclusions
In this note we propose a simple decomposition of the equilibrium and transient global temperature responses to an external forcing into a sum of contributions associated with the Planck response, the different climate feedbacks, and, eventually, the ocean heat uptake. This allows us to quantify how the various processes contribute to the multimodel mean and intermodel spread of the global temperature change. This is illustrated (Figs. 1 –4) using published results for the feedback parameters and the radiative forcings (Soden and Held 2006; Forster and Taylor 2006; Randall et al. 2007b), and by diagnosing the ocean heat uptake efficiency from model outputs. In transient simulations, the absolute values of the contributions of the WV + LR feedback, the cloud feedbacks, and the ocean heat uptake to the global temperature response appears to be comparable (Fig. 4a). However, for the ensemble of models considered here, the spread of the transient temperature change resulting from intermodel differences appears to be primarily due to cloud feedback. The spread resulting from WV + LR feedback, ocean heat uptake, or radiative forcing appears to be of the same order of magnitude and roughly one-third of the spread resulting from the cloud feedback (Fig. 4b). Note that the radiative forcing associated with non-CO2 greenhouse gases and aerosols is more uncertain than that associated with CO2 (Forster et al. 2007). Therefore, the intermodel spread of radiative forcing estimates might be larger either for twentieth-century simulations or for climate change simulations, based on emission scenarios that include changes in aerosol concentrations, than in this study. This difference is mitigated, however, by the fact that the relative contribution of aerosols versus greenhouse gases is likely to decrease in the future (Dufresne et al. 2005).
Our analysis shows that the contribution of each feedback and of the radiative forcing to intermodel differences in temperature change is roughly similar, in a normalized sense, in equilibrium and transient simulations (Figs. 1b and 4b). In particular, cloud feedbacks appear to be the main source of spread in both cases. Intermodel differences in cloud feedbacks have been shown to arise primarily from the response of low-level clouds (Bony and Dufresne 2005; Webb et al. 2006; Wyant et al. 2006). Understanding and evaluating the physical processes that control these cloud responses thus appears to be of primary importance for better assessing the relative credibility of climate projections from the different models.
Acknowledgments
We thank Jean-Yves Grandpeix for frequent and useful discussions on feedback. We are grateful to Jonathan Gregory and an anonymous reviewer for useful comments on this study, and we acknowledge the modeling groups for making their simulations available for analysis, the Program for Climate Model Diagnosis and Intercomparison (PCMDI) for collecting and archiving the CMIP3 model output, and the WCRP’s Working Group on Coupled Modelling (WGCM) for organizing the model data analysis activity. The WCRP CMIP3 multimodel dataset is supported by the Office of Science, U.S. Department of Energy. This research was supported by CNRS, byINSU-EVE French Program under the project MissTerre, and by the European Commission under the project ENSEMBLES (Contract GOCE-CT-2003-505539).
REFERENCES
Bony, S., and J-L. Dufresne, 2005: Marine boundary layer clouds at the heart of cloud feedback uncertainties in climate models. Geophys. Res. Lett., 32 .L20806, doi:10.1029/2005GL023851.
Bony, S., and Coauthors, 2006: How well do we understand and evaluate climate change feedback processes? J. Climate, 19 , 3445–3482.
Collins, M., C. M. Brierley, M. Mac Vean, B. B. B. Booth, and G. R. Harris, 2007: The sensitivity of the rate of transient climate change to ocean physics perturbations. J. Climate, 20 , 2315–2320.
Collins, W. D., and Coauthors, 2006: Radiative forcing by well-mixed greenhouse gases: Estimates from climate models in the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4). J. Geophys. Res., 111 .D14317, doi:10.1029/2005JD006713.
Colman, R., 2003: A comparison of climate feedbacks in general circulation models. Climate Dyn., 20 , 865–873. doi:10.1007/s00382-003-0310-z.
Dufresne, J-L., J. Quaas, O. Boucher, S. Denvil, and L. Fairhead, 2005: Contrasts in the effects on climate of anthropogenic sulfate aerosols between the 20th and the 21st century. Geophys. Res. Lett., 32 .L21703, doi:10.1029/2005GL023619.
Forster, P., and Coauthors, 2007: Changes in atmospheric constituents and in radiative forcing. Climate Change 2007: The Scientific Basis, S. Solomon et al., Eds., Cambridge University Press, 129–234.
Forster, P., and K. E. Taylor, 2006: Climate forcings and climate sensitivities diagnosed from coupled climate model integrations. J. Climate, 19 , 6181–6194.
Gregory, J. M., and J. F. B. Mitchell, 1997: The climate response to CO2 of the Hadley Centre coupled AOGCM with and without flux adjustment. Geophys. Res. Lett., 24 , 1943–1946.
Hall, A., and S. Manabe, 1999: The role of water vapor feedback in unperturbed climate variability and global warming. J. Climate, 12 , 2327–2346.
Hallegatte, S., A. Lahellec, and J-Y. Grandpeix, 2006: An elicitation of the dynamic nature of water vapor feedback in climate change using a 1D model. J. Atmos. Sci., 63 , 1878–1894.
Hansen, J., A. Lacis, D. Rind, G. Russell, P. Stone, I. Fung, R. Ruedy, and J. Lerner, 1984: Climate Sensitivity: Analysis of Feedback Mechanisms. Geophys. Monogr., Vol. 29, Amer. Geophys. Union.
Meehl, G. A., W. M. Washington, J. M. Arblaster, and A. Hu, 2004: Factors affecting climate sensitivity in global coupled models. J. Climate, 17 , 1584–1596.
Meehl, G. A., C. Covey, B. McAvaney, M. Latif, and R. J. Stouffer, 2005: Overview of the Coupled Model Intercomparison Project. Bull. Amer. Meteor. Soc., 86 , 89–93.
Meehl, G. A., and Coauthors, 2007: Global climate projections. Climate Change 2007: The Scientific Basis, S. Solomon et al., Eds., Cambridge University Press, 747–846.
Myhre, G., E. J. Highwood, K. P. Shine, and F. Stordal, 1998: New estimates of radiative forcing due to well mixed greenhouse gases. Geophys. Res. Lett., 25 , 2715–2718.
Randall, D. A., and Coauthors, 2007a: Climate models and their evaluation. Climate Change 2007: The Scientific Basis, S. Solomon et al., Eds., Cambridge University Press, 589–662.
Randall, D. A., and Coauthors, 2007b: Climate models and their evaluation—Supplementary materials. Climate Change 2007: The Scientific Basis, S. Solomon et al., Eds., Cambridge University Press. [Available online at http://www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1-chapter8-supp-material.pdf.].
Soden, B. J., and I. M. Held, 2006: An assessment of climate feedbacks in coupled ocean atmosphere models. J. Climate, 19 , 3354–3360.
Soden, B. J., A. J. Broccoli, and R. S. Hemler, 2004: On the use of cloud forcing to estimate cloud feedback. J. Climate, 17 , 3661–3665.
Stephens, G. L., 2005: Cloud feedbacks in the climate system: A critical review. J. Climate, 18 , 237–273.
Webb, M. J., and Coauthors, 2006: On the contribution of local feedback mechanisms to the range of climate sensitivity in two GCM ensembles. Climate Dyn., 27 , 17–38. doi:10.1007/s00382-006-0111-2.
Williams, K. D., C. A. Senior, and J. F. B. Mitchell, 2001: Transient climate change in the Hadley Centre models: The role of physical processes. J. Climate, 14 , 2659–2674.
Wyant, M. C., C. S. Bretherton, J. T. Bacmeister, J. T. Kiehl, I. M. Held, M. Zhao, S. A. Klein, and B. J. Soden, 2006: A comparison of low-latitude cloud properties and their response to climate change in three AGCMs sorted into regimes using mid-tropospheric vertical velocity. Climate Dyn., 27 , 261–279. doi:10.1007/s00382-006-0138-4.
For a CO2 doubling, (a) multimodel mean ±1 standard deviation (thick line) and 5%–95% interval (thin line) of the equilibrium temperature change (ΔTes), and contributions to this temperature change associated with the Planck response, combined water vapor and lapse-rate (WV + LR) feedback, surface albedo feedback, and cloud feedback. (b) Intermodel standard deviation of the temperature change estimates associated with the radiative forcing, the Planck response, and the various feedbacks normalized by the intermodel standard deviation of the equilibrium temperature change ΔTes reported in (a).
Citation: Journal of Climate 21, 19; 10.1175/2008JCLI2239.1
Equilibrium temperature change associated with the Planck response and the various feedbacks, computed for 12 CMIP3/AR4 AOGCMs for a 2 × CO2 forcing of reference (3.71 W m−2). The GCMs are sorted according to ΔTes.
Citation: Journal of Climate 21, 19; 10.1175/2008JCLI2239.1
Transient temperature change (ΔTts or TCR, red line) and contributions to this temperature change associated with the Planck response, the ocean heat uptake (OHU), and the various feedbacks, computed for 12 CMIP3/AR4 AOGCMs for a 2 × CO2 forcing of reference (3.71 W m−2). The GCMs are sorted according to ΔTts.
Citation: Journal of Climate 21, 19; 10.1175/2008JCLI2239.1
For a CO2 doubling, (a) multimodel mean ±1 standard deviation (thick line) and 5%–95% interval (thin line) of the transient temperature change (ΔTts) and contributions to this temperature change associated with the Planck response, OHU, combined water vapor and lapse-rate (WV + LR) feedback, surface albedo feedback, and cloud feedback. (b) Intermodel standard deviation of the transient temperature change estimates associated with intermodel differences in radiative forcing, Planck response, ocean heat uptake, and the various feedbacks normalized by the intermodel standard deviation of the transient temperature change ΔTts.
Citation: Journal of Climate 21, 19; 10.1175/2008JCLI2239.1
The 2 × CO2 radiative forcing ΔQt, total feedback parameter λ, and ocean heat uptake efficiency κ estimates of the 12 CMIP3/AR4 models used in this paper, and their multimodel mean and standard deviation.
Multimodel mean and intermodel standard deviation of the LW, SW, and net radiative forcing (W m−2) for a CO2 doubling computed by GCMs in two intercomparison studies, with two different numerical setups (see text). In parenthesis, the standard deviation is computed relative to the mean.
Multimodel mean and intermodel standard deviation of total feedback parameter λ and its components λx (W m−2 K−1), the ocean heat uptake efficiency κ (W m−2 K−1), the 2 × CO2 radiative forcing ΔQt (W m−2), and their associated equilibrium and transient temperature changes (°C). The multimodel mean and standard deviation of the equilibrium (Δes) and transient (ΔTts) temperature changes (°C) are also given.
