• Bjerknes, J., 1964: Atlantic air-sea interaction. Advances in Geophysics, Vol. 10, Academic Press, 1–82.

  • Church, J. A., , J. M. Gregory, , P. Huybrechts, , M. Kuhn, , K. Lambeck, , M. T. Nhuan, , D. Qin, , and P. L. Woodworth, 2001: Changes in sea level. Climate Change 2001: The Scientific Basis, J. T. Houghton et al., Eds., Cambridge University Press, 639–693.

    • Search Google Scholar
    • Export Citation
  • Collins, M., , B. B. B. Booth, , G. R. Harris, , J. M. Murphy, , D. M. H. Sexton, , and M. J. Webb, 2006: Towards quantifying uncertainty in transient climate change. Climate Dyn., 27 , 127147.

    • Search Google Scholar
    • Export Citation
  • Cox, P. M., , R. A. Betts, , C. B. Bunton, , R. L. H. Essery, , P. R. Rowntree, , and J. Smith, 1999: The impact of new land surface physics on the GCM simulation of climate and climate sensitivity. Climate Dyn., 15 , 183203.

    • Search Google Scholar
    • Export Citation
  • Delworth, T. L., , and M. E. Mann, 2000: Observed and simulated multidecadal variability in the Northern Hemisphere. Climate Dyn., 16 , 661676.

    • Search Google Scholar
    • Export Citation
  • Ganachaud, A., , and C. Wunsch, 2003: Large-scale ocean heat and freshwater transports during the World Ocean Circulation Experiment. J. Climate, 16 , 696705.

    • Search Google Scholar
    • Export Citation
  • Gordon, C., , C. Cooper, , C. A. Senior, , H. Banks, , J. M. Gregory, , T. C. Johns, , J. F. B. Mitchell, , and R. A. Wood, 2000: The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Climate Dyn., 16 , 147168.

    • Search Google Scholar
    • Export Citation
  • Gregory, J. M., , P. Huybrechts, , and S. C. B. Raper, 2004: Climatology: Threatened loss of the Greenland ice-sheet. Nature, 428 , 616.

  • Held, I. M., , and A. Y. Hou, 1980: Nonlinear axially symmetric circulations in a nearly inviscid atmosphere. J. Atmos. Sci., 37 , 515533.

    • Search Google Scholar
    • Export Citation
  • Hewitt, C. D., , R. J. Stouffer, , A. J. Broccoli, , J. F. B. Mitchell, , and P. J. Valdes, 2003: The effect of ocean dynamics in a coupled GCM simulation of the last glacial maximum. Climate Dyn., 20 , 203218.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and D. J. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38 , 11791196.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and P. J. Valdes, 1990: On the existence of storm tracks. J. Atmos. Sci., 47 , 18541864.

  • Hughes, T. M. C., , and A. J. Weaver, 1994: Multiple equilibria of an asymmetric two-basin ocean model. J. Phys. Oceanogr., 24 , 619637.

    • Search Google Scholar
    • Export Citation
  • Inatsu, M., , H. Mukougawa, , and S-P. Xie, 2002: Stationary eddy response to surface boundary forcing: Idealized GCM experiments. J. Atmos. Sci., 59 , 18981915.

    • Search Google Scholar
    • Export Citation
  • Jayne, S. R., , and J. Marotzke, 2002: The oceanic eddy heat transport. J. Phys. Oceanogr., 32 , 33283345.

  • Johns, T. C., and Coauthors, 2003: Anthropogenic climate change for 1860 to 2100 simulated with the HadCM3 model under updated emissions scenarios. Climate Dyn., 20 , 583612.

    • Search Google Scholar
    • Export Citation
  • Jones, G. S., , J. M. Gregory, , P. A. Stott, , S. F. B. Tett, , and R. B. Thorpe, 2005: An AOGCM simulation of the climate response to a volcanic super-eruption. Climate Dyn., 25 , 725738.

    • Search Google Scholar
    • Export Citation
  • Knight, J. R., , R. J. Allan, , C. K. Folland, , M. Vellinga, , and M. E. Mann, 2005: A signature of persistent natural thermohaline circulation cycles in observed climate. Geophys. Res. Lett., 32 .L20708, doi:10.1029/2005GL024233.

    • Search Google Scholar
    • Export Citation
  • Manabe, S., , and R. J. Stouffer, 1988: Two stable equilibria of a coupled ocean–atmosphere model. J. Climate, 1 , 841866.

  • Manabe, S., , and R. J. Stouffer, 1999: The role of thermohaline circulation in climate. Tellus, 51A , 91109.

  • Marsh, R., and Coauthors, 2004: Bistability of the thermohaline circulation identified through comprehensive 2-parameter sweeps of an efficient climate model. Climate Dyn., 23 , 761777.

    • Search Google Scholar
    • Export Citation
  • McCreary Jr., J. P., , and P. Lu, 1994: Interaction between the subtropical and equatorial ocean circulations: The subtropical cell. J. Phys. Oceanogr., 24 , 466497.

    • Search Google Scholar
    • Export Citation
  • Mikolajewicz, U., , M. Gröger, , E. Maier-Reimer, , G. Schurgers, , M. Vizcaíno, , and A. M. E. Winguth, 2007: Long-term effects of anthropogenic CO2 emissions simulated with a complex earth system model. Climate Dyn., 28 , 599633.

    • Search Google Scholar
    • Export Citation
  • Murphy, J. M., , D. M. H. Sexton, , D. N. Barnett, , G. S. Jones, , M. J. Webb, , M. Collins, , and D. A. Stainforth, 2004: Quantification of modelling uncertainties in a large ensemble of climate change simulations. Nature, 430 , 768772.

    • Search Google Scholar
    • Export Citation
  • Neelin, J. D., , and I. M. Held, 1987: Modeling tropical convergence based on the moist static energy budget. Mon. Wea. Rev., 115 , 312.

    • Search Google Scholar
    • Export Citation
  • Peixoto, J. P., , and A. H. Oort, 1992: Physics of Climate. American Institute of Physics, 520 pp.

  • Pope, V. D., , and R. A. Stratton, 2002: The processes governing horizontal resolution sensitivity in a climate model. Climate Dyn., 19 , 211236.

    • Search Google Scholar
    • Export Citation
  • Pope, V. D., , M. L. Gallani, , P. R. Rowntree, , and R. A. Stratton, 2000: The impact of new physical parametrizations in the Hadley Centre climate model: HadAM3. Climate Dyn., 16 , 123146.

    • Search Google Scholar
    • Export Citation
  • Rahmstorf, S., 1996: On the freshwater forcing and transport of the Atlantic thermohaline circulation. Climate Dyn., 12 , 799811.

  • Rahmstorf, S., 2002: Ocean circulation and climate during the past 120 000 years. Nature, 419 , 207214.

  • Roberts, M. J., and Coauthors, 2004: Impact of an eddy-permitting ocean resolution on control and climate change simulations with a global coupled GCM. J. Climate, 17 , 320.

    • Search Google Scholar
    • Export Citation
  • Schiller, A., , U. Mikolajewicz, , and R. Voss, 1997: The stability of the North Atlantic thermohaline circulation in a coupled ocean-atmosphere general circulation model. Climate Dyn., 13 , 325347.

    • Search Google Scholar
    • Export Citation
  • Schmittner, A., , M. Latif, , and B. Schneider, 2005: Model projections of the North Atlantic thermohaline circulation for the 21st century assessed by observations. Geophys. Res. Lett., 32 .L23710, doi:10.1029/2005GL024368.

    • Search Google Scholar
    • Export Citation
  • Schneider von Deimling, T., , H. Held, , A. Ganopolski, , and S. Rahmstorf, 2006: Climate sensitivity estimated from ensemble simulations of glacial climate. Climate Dyn., 27 , 149163.

    • Search Google Scholar
    • Export Citation
  • Seager, R., , D. S. Battisti, , J. Yin, , N. Gordon, , N. Naik, , A. C. Clement, , and M. A. Cane, 2002: Is the Gulf Stream responsible for Europe’s mild winters. Quart. J. Roy. Meteor. Soc., 128 , 25632586.

    • Search Google Scholar
    • Export Citation
  • Shaffrey, L., , and R. Sutton, 2004: The interannual variability of energy transports within and over the Atlantic Ocean in a coupled climate model. J. Climate, 17 , 14331448.

    • Search Google Scholar
    • Export Citation
  • Shaffrey, L., , and R. Sutton, 2006: Bjerknes compensation and the decadal variability of the energy transports in a coupled climate model. J. Climate, 19 , 11671181.

    • Search Google Scholar
    • Export Citation
  • Shepherd, T. G., 1993: Nonlinear saturation of baroclinic instability. Part III: Bounds on the energy. J. Atmos. Sci., 50 , 26972709.

  • Stainforth, D. A., and Coauthors, 2005: Uncertainty in predictions of the climate response to rising levels of greenhouse gases. Nature, 433 , 403406.

    • Search Google Scholar
    • Export Citation
  • Stouffer, R. J., and Coauthors, 2006: Investigating the causes of the response of the thermohaline circulation to past and future climate changes. J. Climate, 19 , 13651387.

    • Search Google Scholar
    • Export Citation
  • Talley, L. D., , J. L. Reid, , and P. E. Robbins, 2003: Data-based meridional overturning streamfunctions for the global ocean. J. Climate, 16 , 32133226.

    • Search Google Scholar
    • Export Citation
  • Thorpe, R. B., , J. M. Gregory, , T. C. Johns, , R. A. Wood, , and J. F. B. Mitchell, 2001: Mechanisms determining the Atlantic thermohaline circulation response to greenhouse gas forcing in a non-flux-adjusted coupled climate model. J. Climate, 14 , 31023116.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , and J. M. Caron, 2001: Estimates of meridional atmosphere and ocean heat transports. J. Climate, 14 , 34333443.

  • Vellinga, M., , and R. A. Wood, 2002: Global climatic impacts of a collapse of the Atlantic thermohaline circulation. Climatic Change, 54 , 251267.

    • Search Google Scholar
    • Export Citation
  • Vellinga, M., , and R. A. Wood, 2007: Impacts of thermohaline circulation shutdown in the twenty-first century. Climatic Change, doi:10.1007/s10584-006-9146-y.

    • Search Google Scholar
    • Export Citation
  • Vellinga, M., , R. A. Wood, , and J. M. Gregory, 2002: Processes governing the recovery of a perturbed thermohaline circulation in HadCM3. J. Climate, 15 , 764780.

    • Search Google Scholar
    • Export Citation
  • Wood, R., , M. Collins, , J. Gregory, , G. Harris, , and M. Vellinga, 2006: Towards a risk assessment for shutdown of the Atlantic thermohaline circulation. Avoiding Dangerous Climate Change, H. J. Schellnhuber et al., Eds., Cambridge University Press, 49–54.

    • Search Google Scholar
    • Export Citation
  • Wunsch, C., 1999: Where do ocean eddy heat fluxes matter. J. Geophys. Res., 104 , 1323513250.

  • Wunsch, C., 2005: The total meridional heat flux and its oceanic and atmospheric partition. J. Climate, 18 , 43744380.

  • Zhang, R., , and T. L. Delworth, 2005: Simulated tropical response to a substantial weakening of the Atlantic thermohaline circulation. J. Climate, 18 , 18531860.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y-C., , and W. B. Rossow, 1997: Estimating meridional energy transports by the atmospheric and oceanic general circulations using boundary fluxes. J. Climate, 10 , 23582373.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Strength of the maximum meridional overturning in the Atlantic Ocean between 40° and 55°N in CON (solid) and HOS (dashed). The northward ocean heat transport in the Atlantic across 30°N for HOS is shown by the dotted line (right axis).

  • View in gallery

    Global-mean net radiative flux at TOA in experiment HOS (heavy solid, decadal averages). Changes in TOA clear-sky longwave (dashed), clear-sky shortwave (dotted), and cloud radiative fluxes (dashed–dotted) are for experiment HOS (decadal averages) minus the time average of CON. Downward is defined as positive.

  • View in gallery

    (a) Difference in net TOA radiative flux (solid) and its separate components (clear-sky longwave, clear-sky shortwave, and cloud radiative fluxes), years 230–240 of HOS relative to CON. Downward is defined as positive. (b) Zonal-mean TOA anomalous cloud radiative fluxes. Significant changes (at the 5% level) in (a) and (b) are shown as heavy lines. (c) Zonal-mean changes in total cloud water content. Significant changes (at the 5% level) are shown by the gray shading (in 10−6 kg kg−1). Cloud diagnostics have the hybrid vertical model coordinate (“Eta”); nominal pressure levels are shown on the right.

  • View in gallery

    (a) Global northward energy transport by the atmosphere (solid) and ocean (dashed). Data from the hosing (control) run (years 230–240) are shown by heavy (thin) lines. (b) Change (HOS minus CON) in energy transport by the atmosphere (solid) and the ocean (dashed). Their sum (dotted line) shows the overall change in meridional transport in the climate system.

  • View in gallery

    Northward global energy transport at the equator by the atmosphere (solid) and ocean (dashed), and their sum (dotted).

  • View in gallery

    (a) Zonally integrated northward atmospheric transports in CON of sensible heat (SH), latent heat (LH), potential energy (PE), and moist static energy (MSE). (b) Anomalies in atmospheric transports of the same quantities for years 230–240 of HOS, relative to CON.

  • View in gallery

    Zonal-mean zonal wind (colors, in m s−1) and streamfunction (contours, in 1010 m3 s−1, positive values clockwise) anomalies in years 230–240 of HOS, relative to CON. The differences shown are significant at the 5% level.

  • View in gallery

    (a) Wavenumber spectrum from CON of meridional sensible heat transport as a function of latitude and wavenumber [units (K m s−1)2 m−1]. This spectrum for the total flow (i.e., time mean plus transient) was calculated by averaging 20 years of daily mean k-spectra. (b) Change in wavenumber spectrum in years 230–240 of HOS relative to spectrum of CON from (a). Negative contours are dotted, zero contour is heavy, and the gray shading indicates where the change exceeds the noise level of CON (estimated as twice the annual-mean standard deviation of the spectra in CON).

  • View in gallery

    (a) Change in total northward ocean heat transport in the various basins during years 230–240 of HOS, compared to CON (bold lines). Thin lines show the change in implied meridional transports between HOS and CON, found by accumulating the surface heat flux difference between HOS and CON. “Indian” refers to the Indian Ocean north of the Indonesian Throughflow, “Pacific” refers to the net heat transport in the Pacific Ocean plus the contribution from the Indian Ocean south of the throughflow. (b) Change in heat transport carried by the meridional overturning.

  • View in gallery

    Regressions against global-mean TOA net radiative flux of (a) zonal-mean TOA radiative flux and (b) meridional energy transport in the atmosphere (heavy solid) and global ocean (dashed). For latitudes north of 30°N changes in ocean heat transport in the Atlantic (dotted) and Pacific (dashed–dotted) are shown. Long-term (50 yr or more) average data were used from all 35 perturbed physics experiments; the gray shading is the estimated standard error of the regression (for clarity, errors for Atlantic and Pacific Ocean heat transport have been omitted).

  • View in gallery

    Regression of global zonal-mean meridional overturning [grayscale, in Sv (W m−2)−1] and global zonal-mean zonal density [white contours, in kg m−3 (W m−2)−1] onto global-mean TOA flux across perturbed physics ensemble. Negative streamfunction contours are dotted and indicate counterclockwise circulation; positive streamfunction contours are solid. Most features of the regressions are statistically significant at the 5% level (significance not shown for sake of clarity).

  • View in gallery

    Meridional energy transports and integrated fluxes in perturbed physics ensemble, expressed as regressions against global-mean TOA net radiative flux [PW (W m−2)−1]. For both figures long-term (50 yr or more) average data were used from all 35 ensemble members.

  • View in gallery

    Global-mean TOA flux (horizontal) and maximum global MOC at 45°N in the perturbed physics ensemble. Observational estimates of MOC and heat transport are shown near the respective ordinates, together with error estimates. Ensemble members for which heat transport and MOC strength are consistent with these observed estimates are shown by solid symbols.

  • View in gallery

    Diagram of changes in energy transports between climate states with and without MOC (years 230–240 of experiment HOS minus CON). Shown are changes in total meridional transports in the ocean (dotted) and atmosphere (white), and area-integrated fluxes across the air–sea interface (dashed) and TOA (solid), all in PW.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 11 11 3
PDF Downloads 5 5 1

Relations between Northward Ocean and Atmosphere Energy Transports in a Coupled Climate Model

View More View Less
  • 1 Hadley Centre, Met Office, Exeter, Devon, United Kingdom
© Get Permissions
Full access

Abstract

The Third Hadley Centre Coupled Ocean–Atmosphere General Circulation Model (HadCM3) is used to analyze the relation between northward energy transports in the ocean and atmosphere at centennial time scales. In a transient water-hosing experiment, where suppressing the Atlantic meridional overturning circulation (MOC) causes a reduction in northward ocean heat transport of up to 0.75 PW (i.e., 75%), the atmosphere compensates by increasing its northward transport of moist static energy. This compensation is very efficient at low latitudes and near complete at the equator throughout the experiment, but is incomplete farther north across the northern midlatitude storm tracks. The change in atmosphere energy transport enables the model to find a new global-mean radiative equilibrium after 240 yr.

In a perturbed physics ensemble of HadCM3 it was found that time-averaged meridional energy transports in ocean and atmosphere can act opposingly. Where model formulation causes an unbalanced mean climate state, for example, an excessive top-of-the-atmosphere radiative surplus at low latitudes, the atmosphere increases its poleward energy transport to disperse this excess. MOC and ocean poleward heat transport tend to be reduced in such model versions, and this offsets the increased poleward atmospheric transport of the low-latitude energy surplus. Model versions that are close to net radiative equilibrium also have ocean heat transport and MOC close to observed values.

Corresponding author address: Michael Vellinga, Hadley Centre, Met Office, FitzRoy Road, Exeter, Devon EX1 3PB, United Kingdom. Email: michael.vellinga@metoffice.gov.uk

Abstract

The Third Hadley Centre Coupled Ocean–Atmosphere General Circulation Model (HadCM3) is used to analyze the relation between northward energy transports in the ocean and atmosphere at centennial time scales. In a transient water-hosing experiment, where suppressing the Atlantic meridional overturning circulation (MOC) causes a reduction in northward ocean heat transport of up to 0.75 PW (i.e., 75%), the atmosphere compensates by increasing its northward transport of moist static energy. This compensation is very efficient at low latitudes and near complete at the equator throughout the experiment, but is incomplete farther north across the northern midlatitude storm tracks. The change in atmosphere energy transport enables the model to find a new global-mean radiative equilibrium after 240 yr.

In a perturbed physics ensemble of HadCM3 it was found that time-averaged meridional energy transports in ocean and atmosphere can act opposingly. Where model formulation causes an unbalanced mean climate state, for example, an excessive top-of-the-atmosphere radiative surplus at low latitudes, the atmosphere increases its poleward energy transport to disperse this excess. MOC and ocean poleward heat transport tend to be reduced in such model versions, and this offsets the increased poleward atmospheric transport of the low-latitude energy surplus. Model versions that are close to net radiative equilibrium also have ocean heat transport and MOC close to observed values.

Corresponding author address: Michael Vellinga, Hadley Centre, Met Office, FitzRoy Road, Exeter, Devon EX1 3PB, United Kingdom. Email: michael.vellinga@metoffice.gov.uk

1. Introduction

To balance the net radiative energy input at low latitudes and net radiative loss at high latitudes, a poleward redistribution of energy is required so that climate equilibrium is achieved. Estimates using observed radiative fluxes at the top of the atmosphere (TOA) quantify this transport to range from near-zero at the equator to a maximum of about 5 PW near 40°N and S (Trenberth and Caron 2001). Quantitative analysis of the energy budget (e.g., the partitioning of the amounts carried by ocean and atmosphere) is hampered by the uncertainty in the observed values of the TOA and surface heat fluxes. But within these error estimates, Wunsch (2005) suggests that oceans and the atmosphere carry comparable amounts of the mean transport at low latitudes. Poleward of about 24°N, the ocean releases much of its heat to the atmosphere, which then accommodates most of the poleward energy transport.

Bjerknes (1964) proposed the hypothesis that, in the Atlantic region at least, ocean and atmosphere transports could well be anticorrrelated. If that were the case, a reduction (say) in the heat transported northward by the ocean would cause an increase in the northward transport by the atmosphere, instead, so as to maintain radiative TOA balance. When Shaffrey and Sutton (2004) tested this hypothesis in a climate GCM, they found little evidence for such a compensation mechanism at interannual time scales. Instead, the ocean and atmosphere seemed to have their own preferred mechanisms of interannual energy transport variability, which indicates an important role for oceanic heat storage. However, when they analyzed decadal variability of the same model, Shaffrey and Sutton (2006) found compensation in the northern extratropics, consistent with the Bjerknes hypothesis, and the Atlantic MOC dominated the decadal fluctuations in global meridional ocean heat transport.

As well as being instrumental in driving multidecadal climate fluctuations in the last century or so (e.g., Delworth and Mann 2000; Knight et al. 2005), the MOC has also been identified as a possible source of rapid climate change in past climate [see Rahmstorf (2002) for an overview]. Whether such transitions between climate states with MOC on or off are likely in the next 100 yr remains an open question (Wood et al. 2006), partially because there remains at present considerable uncertainty as to the strength of feedbacks in climate models (Stouffer et al. 2006). An important, but insufficient, condition for climate to possess a stable equilibrium without MOC and its northward heat transport would be for the energy budget to rearrange itself to equilibrium. The question remains unanswered how climate would readjust its energy budget after a major, large-scale change in the ocean’s circulation. If the northward heat transport in the Atlantic is greatly reduced, will the atmosphere be able to carry the ocean’s heat transport instead? Will surface and TOA fluxes be able to adjust to accommodate a new equilibrium, or not?

Studies into multiplicity of MOC equilibria have mainly relied on experiments with climate models in which key climate interactions have either been neglected (e.g., Rahmstorf 1996) or strongly simplified (e.g., Marsh et al. 2004), or that require flux adjustment to maintain a stable control state (Manabe and Stouffer 1988). Studies of MOC shutdown tend to focus on understanding feedbacks on the MOC (Schiller et al. 1997; Vellinga and Wood 2007) and changes in surface climate (Manabe and Stouffer 1999; Vellinga and Wood 2002). Understanding how key climate properties such as the energy budget are related to large changes in the MOC remains incomplete. In this paper we aim to make some progress in this area, by analyzing experiments with the Met Office’s Third Hadley Centre Coupled Ocean–Atmosphere General Circulation Model (HadCM3). This model does not require flux adjustment to maintain a stable control climate, which makes it suitable for studying the climate’s energy budget. If we can understand the links between energy transports and the MOC, this will in the long-term help in identifying any possible constraints on future MOC changes.

The relation between meridional energy transports in the ocean and atmosphere has wider relevance for climate simulations. Clearly it is desirable that both transports are simulated reliably. To ensure this we need to assess if and how time-mean meridional energy transports affect each other and how they are organized to facilitate a long-term, net radiative equilibrium in a climate model. Recently, so-called perturbed physics experiments have been carried out with HadCM3. In this class of experiments changes within plausible ranges are made to poorly known model parameters. In addition, certain schemes have been either included or excluded altogether. The main motivation for this type of experiment is to quantify how uncertainty in model formulation affects projections of anthropogenic climate change (e.g., Murphy et al. 2004). Anthropogenic climate change is not discussed in this paper: However, by altering model formulation in this way, strengths of climatic feedbacks are changed, and this affects the simulated mean state of oceans and atmosphere. We explore this spread of model states to quantify interdependency of energy transports in ocean and atmosphere and make a comparison to observational estimates of ocean transports.

The layout of this paper is as follows: HadCM3 is briefly described in section 2. The hosing experiment and the change in energy transports are described in section 3. Results from the perturbed physics experiments are described in section 4. Conclusions and a discussion follow in section 5.

2. Model

HadCM3 is a coupled ocean–atmosphere GCM with sea ice and land surface schemes (Gordon et al. 2000; Pope et al. 2000; Cox et al. 1999). Resolution in the ocean is 1.25° × 1.25°; in the atmosphere it is 2.5° × 3.75°. These grid spacings allow the major ocean boundary currents and overflows to be marginally resolved. Midlatitude baroclinic disturbances in the atmosphere are resolved, but not oceanic eddies. The model has been used in numerous studies of past (Hewitt et al. 2003) and future (Johns et al. 2003) climate, as well as in sensitivity studies, for example, to study the climate response to an eruption of a supervolcano (Jones et al. 2005) or a hypothetical shutdown of the ocean’s MOC in the twenty-first century (Vellinga and Wood 2007).

The standard, unperturbed version of the model is used in the next section. The effects on the energy transports of perturbed versions of HadCM3 are described in section 4.

3. Hosing experiment

a. Experimental setup

A stable model state with “MOC off” has not been found in HadCM3. In previous studies (e.g., Vellinga et al. 2002) it was found that a large (instantaneous) freshwater perturbation can temporarily suppress the model’s MOC, but the latter gradually recovers in about 100 yr. In the present study we want the model to be in a state without MOC for long enough that transports and surface fluxes have time to adjust. To achieve this we applied a continuous surface freshwater flux to the North Atlantic (between 45°–90°N, 100°W–45°E) for 240 yr. “Hosing” the ocean with this extra freshwater flux reduces high-latitude density and is a well-established way to disrupt the MOC in models. The flux was applied at a constant rate of 1.1 Sv (Sv ≡ 106 m3 s−1). In our experiment the Greenland ice sheet was held fixed, but a flux of 1.1 Sv is equivalent to a melt rate of around 20 m yr−1 of the Greenland ice sheet, if we assume an initial surface area of 1.7 × 1012 m2. At this vigorous rate, complete melting of the Greenland ice sheet would occur in about 75 yr (using ice sheet data from Church et al. 2001). Even in the most extreme emissions scenario, melting of the Greenland ice sheet takes about 1000 yr (Gregory et al. 2004). This experiment (referred to as “HOS”) is clearly not intended as a realistic scenario, but merely to allow us to analyze the model in an equilibrium state without MOC. The hosing was applied to the model’s control state, “CON.” Greenhouse gas concentrations were held fixed at preindustrial levels in both experiments.

The strong hosing in experiment HOS weakens the model’s North Atlantic overturning circulation from 20 to 5 Sv in about 40 yr (Fig. 1). After that the MOC continues to weaken more slowly, to about 1 Sv after 240 yr. The northward ocean heat transport in CON has a maximum of just over 1 PW near 30°N in the Atlantic; toward the end of HOS the weakening of the MOC has reduced this by 75% to 0.25 PW. The heat transport evolves in a similar way as the circulation strength, highlighting the importance of the MOC for the northward heat transport at this latitude.

The response in surface variables at the end of HOS has similar patterns to those from an experiment analyzed previously by Vellinga and Wood (2002). It is marked by cooling in Northern Hemisphere surface air temperature (2°C on average) and warming in the Southern Hemisphere (0.8°C on average). An important difference with the experiment described by Vellinga and Wood is that in the present experiment HOS the MOC is suppressed for much longer. In the following we discuss how the 0.75-PW reduction of Atlantic heat transport affects the model’s energy budget: TOA radiative flux in section 3b and net global meridional energy transports in section 3c. Separate transports in the atmosphere are discussed in section 3d, and ocean transports in section 3e.

b. TOA radiative response

In experiment HOS there is a net downward (defined as positive) TOA radiative flux that peaks at around 0.5 W m−2 in the first 50 yr (Fig. 2). Gradually the radiative fluxes adjust and by year 240 the net flux has reduced to zero, showing that the model has reattained a global radiative balance. To help interpret the changes to the TOA radiative budget the net flux is partitioned in separate clear-sky and cloud radiative components. Clear-sky longwave (shortwave) radiation is the longwave (shortwave) radiation recalculated after the instantaneous removal of the clouds without any compensating adjustment. The cloud radiative flux is the difference between net and clear-sky TOA flux. In the new global-mean balance there is less outgoing clear-sky longwave radiation than before, and more clear-sky shortwave radiation is reflected into space. These changes are consistent with a colder climate in which surface albedo has increased due to larger coverage in snow and ice. The change in global-mean cloud radiative flux is smaller and acts as a warming.

For the discussion of poleward energy transports that will follow in section 3c it is useful to also consider the latitudinal changes in TOA fluxes (Fig. 3a). The net TOA flux anomaly is roughly antisymmetric around the equator: north of about 40°N where surface cooling is strongest, there is anomalous downward TOA flux. As expected, the largest changes in the clear-sky components are in the extratropical Northern Hemisphere where the strongest reduction in surface air temperature occurs. In the deep tropics, where surface temperature changes are relatively small, changes in water vapor content (drier north of the equator, more humid south of the equator) dominate the clear-sky longwave response. South of 60°S, where there is some surface warming, clear-sky fluxes are opposite to those in the Northern Hemisphere, but smaller.

Whereas the global-mean contribution of the cloud radiative flux to the net TOA flux is smaller than those of the clear-sky fluxes (Fig. 2), its zonal-mean contribution is of comparable magnitude between 30°S and 40°N (Fig. 3a). The changes in the cloud radiative flux (Fig. 3b) are compared to those in cloud water content (Fig. 3c). Changes in cloud shortwave flux dominate the net cloud radiative flux: where there is less cloud there is a positive contribution, and vice versa (e.g., north of 45°N: cooler and drier conditions lead to a reduction in cloud cover, lowering the reflection of shortwave radiation back into space). In the northern tropics colder surface temperatures in HOS mean that there is less deep convection and convective cloud than in CON, resulting in less reflection of solar radiation and stronger radiative cooling (because of the lower, warmer cloud tops). South of the equator, where in HOS the surface is warmer, the opposite is the case. The longwave response dominates here, and there is a net radiative warming throughout the tropics due to clouds.

To summarize, changes in the global-mean clear- sky fluxes are dominated by the large changes in the extratropical Northern Hemisphere. At low latitudes, changes in cloud radiative fluxes are important: in the tropics they act to warm climate, in the northern subtropics to cool climate.

c. Global energy transports

In the atmosphere the northward energy transport can be understood in terms of moist static energy (MSE; e.g., Neelin and Held 1987); see appendix A for details on how this was calculated. In CON the transport of MSE varies between 0 PW at the equator to about 4.5 PW of poleward transport at 40°S and N (Fig. 4a, thin). The global ocean heat transport is also shown (dashed lines). Whereas the energy transport by the atmosphere in CON is near antisymmetric with zero transport at the equator, the ocean transport has a clear asymmetry with nonzero transport at the equator. This is due mainly to the northward heat transport in the South Atlantic, associated with the MOC. The global ocean transport peaks at about 1.8 PW near 16°N. These model transports in the atmosphere and ocean are consistent with observational values within the error estimates as provided by Ganachaud and Wunsch (2003) and Wunsch (2005), as well as those estimated by Trenberth and Caron (2001). Note that in CON the atmospheric transport across the equator is essentially zero but that the ocean carries about 0.4 PW from the Southern to the Northern Hemisphere.

The consequences for the global northward energy transport when the MOC has been suppressed are evident in both ocean and atmosphere (Fig. 4a, heavy). In the final decade of HOS the northward ocean heat transport has decreased at most latitudes. In other words the poleward ocean heat transport has decreased in the Northern Hemisphere, but increased in the Southern Hemisphere. In the atmosphere, northward energy transport has gone up. These changes can be seen more clearly in the difference plot, Fig. 4b. The atmosphere is partially able to compensate for the reduced transport by the ocean. Compensation is very effective in the tropics where, in fact, the atmosphere slightly overcompensates. But in the subtropics and at midlatitudes the atmosphere is unable to fully compensate for the ocean. A similar latitudinal dependency of energy transport anomalies can be seen in Fig. 1b of a hosing experiment with the Geophysical Fluid Dynamics Laboratory Climate Model version 2.0 (GFDL CM2.0) by Zhang and Delworth (2005). We found in our experiment that the compensation by the atmosphere in the tropics not only holds in the final decade, but actually throughout experiment HOS (Fig. 5). As a result, there is a near-constant energy transport from the Southern into the Northern Hemisphere of around 0.5 PW, slightly more than the long-term average in CON of 0.4 PW.

In the following two sections the changes in atmosphere and ocean transports will be examined more closely.

d. Changes in atmosphere transport

The net moist static energy transport in the atmosphere is the sum of sensible heat, latent heat, and potential energy transports. The zonal mean of the individual northward transports have an intricate dependence on latitude (Fig. 6a). Nevertheless, when added, the resulting transport of moist static energy has a relatively simple antisymmetrical structure (Fig. 6a, dashed–dotted line; Fig. 4a). Individual and net transports are largely consistent with observational estimates by Peixoto and Oort (1992), who also provide a detailed interpretation.

To see how the atmosphere’s compensation for the reduced ocean heat transport (Fig. 4b) is achieved, we calculated the differences in each of the transports between the last 10 yr of HOS and CON (Fig. 6b). The increase in northward transport in the tropics is mainly accomplished by an enhanced transport of potential energy and counteracted by anomalous southward transports of latent and sensible heat. The changes in energy transport in the tropics are associated with changes in the Hadley circulation: in the annual mean the anomaly consists of rising in the Southern Hemisphere and sinking in the Northern Hemisphere (Fig. 7). This is a well-established response of the tropical atmosphere to the bipolar SST anomaly near the equator after MOC shutdown, which is seen in many other climate models (e.g., Stouffer et al. 2006) although exact positioning and strength of the anomalous circulation depends on the forcing and model (e.g., Zhang and Delworth 2005). This anomalous circulation transports sensible heat southward across the equator, that is, from the anomalously cold into the anomalously warm hemisphere. It is the (larger) northward transport of potential energy associated with this anomalous circulation that feeds into the Northern Hemisphere. Unfortunately we do not have the model data to investigate the conversion from potential into kinetic energy. But the strong intensification of the subtropical zonal jet in the Northern Hemisphere (Fig. 7) indicates a larger availability of kinetic energy, and its proximity to the sinking branch of the anomalous Hadley circulation is consistent with analysis by Held and Hou (1980).

The change in latent heat transport is largest in the tropics, where the anomalous Hadley cell transports moist air into the Southern Hemisphere and has a dry northward return flow aloft. Elsewhere, changes in zonal-mean latent heat transport are not very significant. Obviously that does not rule out important regional changes to the hydrological cycle (Vellinga et al. 2002), but these are not explored here.

At northern midlatitudes transient and stationary eddies carry most of the meridional sensible heat transport (Peixoto and Oort 1992). When the MOC is suppressed, cooling at high latitudes is stronger than in the tropics, so the north–south surface temperature gradient becomes larger across most of the Northern Hemisphere midlatitudes, in particular over the ocean between 40° and 65°N, that is, the storm-track region (cf. Fig. 3 of Vellinga and Wood 2002). Also, as seen in Fig. 7, the time-average zonal flow in the Northern Hemisphere becomes stronger. Both of these changes suggest the possibility of larger midlatitude baroclinic instability. Consistent with this, the wavenumber spectrum of the total (i.e., time mean plus transient) northward sensible heat transport shows a 10% enhancement of power across most of the Northern Hemisphere (Fig. 8). There is an increase in power at wavenumbers 5–7 around 20°–30°N, which perhaps suggests that baroclinic instability increases farther south, compared to CON. The absence of an increase of power at wavenumbers 5–7 in the northern storm-track region is consistent with the limited increase in atmospheric energy transport at these latitudes (Fig. 4b).

At higher latitudes, between 50° and 70°N, the strongest increase occurs for wavenumber 2. The k-spectra of the 20-yr mean value of υT and 500-hPa height in HOS exhibit a similar increase near 50°N of wavenumber 2 (not shown), which suggests a change in the stationary eddy field. The land–sea temperature contrast in HOS goes down in the Northern Hemisphere (the suppression of the MOC cools the ocean more strongly than the land (Fig. 3 of Vellinga and Wood 2002), so it is unlikely to cause the increase in wavenumber-2 power in this case. Work by Inatsu et al. (2002) suggests that a longitudinal contrast of tropical SST is rather effective in forcing stationary eddies, more so than extratropical land–sea contrast. In HOS the tropical North Atlantic cools relative to the tropical North Pacific by several degrees, consistent with the results from Inatsu et al. (2002). Changes in extratropical surface (Hoskins and Karoly 1981) or diabatic (Hoskins and Valdes 1990) heating also have the potential to affect the stationary wave pattern. A detailed analysis of which of these mechanisms is dominant is beyond the scope of this paper.

The increased power at wavenumber 2 in HOS implies that without MOC the atmosphere circulation in HadCM3 has stronger departures from zonality. This appears to be in contrast with experiments described by Seager et al. (2002), who reported little sensitivity of the stationary eddy distribution to removing all ocean heat transport. But in their experiments, in which an atmosphere GCM was coupled to a mixed layer ocean, removing the ocean heat transport introduced little temperature difference between the tropical North Atlantic and Pacific.

e. Changes in ocean heat transport

Here we will briefly look at the changes in ocean heat transport that occur in HOS. The reduced net northward heat transport in the Atlantic Ocean is partially compensated by an anomalous northward heat transport in the other basins (Fig. 9a). From a global energy budget perspective it is relevant that this change in Indo-Pacific cross-equatorial heat transport compensates almost half of the loss of heat transport into the Northern Hemisphere following the reduction in Atlantic heat transport. About 0.25 PW of this compensation is passed on to the atmosphere in the Northern Hemisphere via increased surface heat loss from the Indo-Pacific Ocean. The differences between actual and implied ocean heat transports are small at most latitudes for the global ocean at this stage in HOS. There is, however, an overall warming (cooling) tendency in the Pacific (Atlantic) between 20°N and 30°S, indicating that the deep ocean has not reached equilibrium, in spite of the near-zero TOA flux by this stage (Fig. 2).

As expected, the reduction in global northward heat transport is dominated by the reduction of the Atlantic overturning component (Fig. 9b). The other components show an increased northward transport, indicating global interdependencies of ocean heat transport. Reduced inflow of cold bottom and intermediate water masses from the Southern Ocean into the Indo-Pacific, and their weaker interior upwelling (not shown) reduce the amount of heat transported southward over much of the southern Indo-Pacific. Changes in the gyre heat transport are mostly smaller than those of the overturning (not shown, but can be inferred from the difference between Figs. 9a and 9b).

4. Perturbed physics experiments

To quantify the range of future anthropogenic climate change, several modeling groups have carried out perturbed physics experiments with ensembles of GCMs in various configurations (Murphy et al. 2004; Stainforth et al. 2005; Schneider von Deimling et al. 2006; Collins et al. 2006). The idea behind these experiments is to quantify the uncertainty inherent to climate modeling by varying model formulation (poorly constrained parameter values, inclusion, or modification of parameterization schemes, etc.) within a plausible range. Here we use 35 experiments in which perturbations have been applied across several sections of the atmosphere and sea ice components of HadCM3. The ocean component is identical in all experiments. These changes of model formulation alter radiative, hydrological, etc., processes and hence the strength of model feedbacks. The fact that the ocean model is the same across the ensemble means that, for a given thermal, haline, or wind stress anomaly, the ocean response would be identical, but due to differences in atmosphere and sea ice models, and hence coupled feedbacks, there is still a spread of ocean and atmosphere states across the ensemble. While the modification of the model formulation falls within plausible ranges, applying multiple perturbations simultaneously may result in an overall better or poorer simulation of climate, for example, net TOA radiative flux. Relevant to the present study are potential interdependencies of the energy budgets of ocean and atmosphere apparent under the applied changes. Any emerging relations from across this ensemble point toward feedbacks within this class of climate models, if not the real climate system. Understanding such interdependencies will aid the interpretation of model behavior and, potentially, model development.

Details of the perturbations and how these were determined are described in appendix B. Our perturbed physics experiments with the climate model HadCM3 are similar to those of Collins et al. (2006); however, we applied different perturbations and did not use flux adjustment. All 35 experiments use constant, preindustrial greenhouse gas concentrations and were run for between 100 and 300 years. We discard years 0–50 when SST and TOA radiative fluxes are adjusting rapidly and use long-term averages of the remainder of each of the simulations.

a. Fluxes at the TOA and ocean surface

The perturbations to the model physics result in a spread of the long-term average global-mean TOA flux across the ensemble, from +2.5 to −1.1 W m−2 (as before, positive is downward). At the centennial time scales for which the ensemble was run the heat capacity of the climate system resides overwhelmingly in the ocean. A net positive (negative) radiative imbalance at the TOA therefore causes a gradual warming (cooling) of the ocean. If the ensemble had been run for long enough the deep ocean would eventually come into equilibrium, a process that could take thousands of years. Here we look at the behavior at the centennial time scale, which is more typical of climate model usage, for example, for model spinup, projections of future climate, or historical simulations. For simplicity we will phrase our analysis below in terms of positive net radiative imbalances, but it is important to remember that for the 14 members with net radiative cooling similar sensitivities exist, but with opposite sign.

The physics perturbations act in such a way that, at these time scales, models with a global-mean net radiative input receive most of this in the tropics and at midlatitudes (solid line, Fig. 10a). In the polar regions (i.e., poleward of 60°N and S) there is a TOA flux in the opposite direction as the global mean, which offsets the TOA flux at lower latitudes. In contrast with the TOA flux, the ocean surface heat flux is strongest at high latitudes (Fig. 10a, dashed line). This is where the deep ocean is ventilated. Here, any modification of ocean temperature is removed from the surface and this reduces the scope for a negative local feedback on ocean surface flux to be established (e.g., increased longwave radiative flux). At lower latitudes, where ocean stratification is stronger, it will be easier to establish a local feedback on surface heat flux. The change in surface heat flux is, indeed, smaller at low latitudes and warms as well as cools the ocean. The local mismatch between the zonal-mean change in TOA radiative and in ocean surface heat flux implies that there is a change in poleward energy transport by the atmosphere.

b. Poleward energy transports

In ensemble members with a net radiative surplus (which we noted arises mostly at lower latitudes) the atmosphere increases its poleward energy transport (Fig. 10b, solid line). The change in atmosphere transport is strongest in the subtropics and midlatitudes, and rather small in the tropics (in contrast with the strong response in the water-hosing experiment where the tropical response is strong; section 3d; Fig. 5). Ocean heat transport changes in the opposite way as in the atmosphere: models with a larger downward TOA flux have a smaller poleward transport (dashed line). The change in the atmosphere dominates the net transport. This sensitivity of the atmosphere transport can be understood as the poleward redistribution of an energy surplus at low latitudes. Why this is accompanied by a reduced poleward heat transport in the ocean is explored below.

First of all we note that the change in ocean heat transport is largely explained by the change in heat carried by the zonal-mean circulation (Fig. 10b, dotted line); contributions from changes in the other components (e.g., ocean gyres) matter less. Changes in the global-mean ocean overturning are consistent with this reduction in poleward ocean heat transport (Fig. 11 gray). In the Northern Hemisphere the deep overturning cell weakens (this occurs exclusively in the Atlantic basin; not shown). In the tropics the shallow, wind-driven subtropical cells also weaken (this occurs mainly in the Indian and Pacific basins). In the Southern Ocean the Deacon cell intensifies.

The weakening of the deep overturning cell can be understood in terms of changes to the ocean density field (Fig. 11, white contours). The meridional density gradient in the upper limb of the MOC Atlantic (between 0 and 2000 m) slackens with increasing TOA because density is reduced more strongly in the north than at other latitudes. This slackening of the north–south density difference has the effect of weakening the Atlantic MOC (e.g., Hughes and Weaver 1994; Thorpe et al. 2001). The strength of the subtropical cells is related to the zonal wind stress at low latitudes (McCreary and Lu 1994). The wind stress indeed weakens in ensemble members with a positive TOA anomaly, in particular over the Indian and Pacific basins (not shown). Why the trades weaken is not explored further, but is perhaps related to a reduced meridional SST gradient off the equator in these basins. Stronger westerlies over the Southern Ocean increase the northward Ekman drift, spinning up the Deacon cell (Fig. 11).

The following picture emerges, summarized in Fig. 12: in model versions where there is a larger net radiative TOA flux at low latitudes, more energy is available to be redistributed poleward. Surface warming associated with a net downward TOA radiative flux is particularly strong at high latitudes. This weakens the ocean overturning circulation and leads to a smaller poleward energy transport by the ocean. Weaker zonal wind stress in the tropics reduces the subtropical cells, also causing a reduction in ocean heat transport. Ocean heat transport thus counteracts the transport by the atmosphere: whereas the atmosphere acts to disperse excessive energy surplus from low to high latitudes, the ocean acts to further increase it.

c. Observational constraints

The strong relation between global-mean TOA radiative flux and ocean heat transport in the ensemble (Fig. 10b) has additional implications. This is illustrated in Fig. 13, where Atlantic MOC and global heat transport at 45°N are plotted against global-mean TOA flux. Consistent with what we have seen before the correlation is negative, and strong. Ensemble members for which ocean heat transport and MOC strength are consistent with observational estimates within the error bars (Ganachaud and Wunsch 2003; Talley et al. 2003) are marked by solid symbols. These are all members that are close to radiative balance, with TOA fluxes ranging from −0.3 to 0.6 W m−2. In the ensemble there is no model that is grossly out of radiative balance, yet has a plausible simulation of MOC and heat transport, and vice versa. It suggests that the model physics (parameter values, etc.) that cause a model to be in near-radiative equilibrium are also instrumental in ensuring a correct simulation of MOC and ocean heat transport—at least at the centennial time scales at which the ensemble was run and adjusted. The relation of Fig. 13 could help to identify suitable observational likelihood weighting in probabilistic climate projections (e.g., of future changes to the MOC), which is not a trivial task (Schmittner et al. 2005).

5. Conclusions and discussion

The energy budget places fundamental constraints on admissible climate states, in the real climate system as well as in models. Here we have used a global, non-flux-adjusted climate GCM to analyze the links between energy transports in ocean and atmosphere at centennial time scales. We investigated these links in a time-dependent context (using a transient “water hosing” experiment in which the MOC was suppressed) and also in a time-independent sense, by exploring a range of mean model states in a perturbed physics ensemble of slightly different model versions. Both sets of experiments give support to the notion that, at centennial time scales, the relation between poleward transports in the ocean and atmosphere is not unstructured but, in many respects, anticorrelated.

In the transient experiment we find that a sustained disruption of the Atlantic meridional overturning circulation (MOC) causes a reduction in northward heat transport in the Atlantic ocean by 75%, around 0.75 PW. The radiative flux at the top of the atmosphere (TOA) adjusts to this change in heat transport. After 240 yr a different net radiative balance is reestablished. Changes in global heat transports and fluxes at key sections are shown in the schematic diagram of Fig. 14.

By lowering the outgoing net radiative flux at the TOA north of 30°N the climate system is able to reduce the “demand” for northward heat transport across 30°N by about 0.15 PW. In addition, the atmosphere is able to take over part of the energy transport in the Northern Hemisphere that was previously carried by the ocean. This compensation by the atmosphere is very efficient at low latitudes, and as a result the net energy transport from the Southern to the Northern Hemisphere by the combined ocean–atmosphere system changes comparatively little. At northern midlatitudes there is also compensating transport by the atmosphere, but to a lesser degree. This occurs mainly in the form of sensible heat transport, which in the subtropics shows an increase in power at wavenumbers 5–7. In the northern storm track region compensation is incomplete. This means that the overall meridional energy transport in the northern extratropics goes down without the MOC heat transport. As mentioned before, a smaller outgoing TOA radiative flux compensates for this.

We have found that the climate system as simulated by HadCM3 has the ability to adjust to a new global radiative equilibrium when the MOC and its heat transport are being suppressed. An important condition for a stable climate state without MOC can thus be met, at least in this model. It is important to realize that, at longer time scales, regional imbalances in the ocean heat budget (section 3e; Fig. 9a) and in the poorly constrained hydrological cycle (Stouffer et al. 2006) could still render an MOC-off state unstable, as indeed is the case in HadCM3 (Vellinga et al. 2002).

The compensating roles between oceans and atmosphere found in the hosing experiment is reminiscent of what was hypothesized by Bjerknes (1964), who assumed that meridional energy transports would compensate so as to maintain a long-term TOA radiative balance. Our experiments suggest a partial compensation could occur, even in the presence of considerable regional TOA changes. In particular, the large increase in atmospheric energy transport in the tropics (up to 0.75 PW), maintaining the energy transport to the Northern Hemisphere after a greatly reduced ocean heat transport, is remarkable. This extra atmospheric energy transport is fed by changes in ocean surface heat and TOA radiative fluxes between 0° and 30°S. The increase in outgoing TOA radiative flux between 20° and 40°N could be linked to the limited ability of the atmosphere to completely transport this extra energy from the equator to the northern midlatitudes—either as cause or effect: this is unclear at this stage (there is little change in the ocean surface heat flux between 0° and 30°N; Fig. 14). These points deserve further attention and need to be looked at in other climate models to assess dependency on model formulation.

Shaffrey and Sutton (2006) analyzed decadal internal variability in HadCM3 and concluded that the Bjerknes compensation hypothesis holds well in the extratropics, but not in the tropics. The latter is seemingly at odds with the present results. However, in our experiment, perturbations to the cross-equatorial transport are in excess of 0.5 PW (Fig. 14), which is very large compared to the internal decadal variability in the transport (about 0.04 PW; Fig. 1c of Shaffrey and Sutton 2006), presumably placing the forced anomaly in a different regime as those of internal variability.

It is tempting to speculate how the present results could depend on resolution. There is evidence that mesoscale variability in the ocean has a significant contribution to net northward heat transport in some areas: boundary currents, the tropical Pacific, and the Southern Ocean (Wunsch 1999; Jayne and Marotzke 2002). Whereas the parameterization of eddy heat transport in HadCM3 is satisfactory in the Southern Ocean, the same cannot be said for the tropical Pacific, where it is the mean current that accommodates the heat transport that, at higher resolution, would be carried by mesoscale features not captured by the eddy parameterization (tropical instability waves) (Roberts et al. 2004). The spatial scale of ocean heat transport divergences affects air–sea interaction, and this could cause a different response of the tropical atmosphere in a hosing experiment in a high-resolution model. The inability of the atmosphere model to fully compensate for the loss of ocean heat transport at northern midlatitudes could reflect a saturation in the level of baroclinic instability in the storm track region. This level may depend on properties of the zonal flow (e.g., Shepherd 1993), on numerical resolution (Pope and Stratton 2002), or both. Midlatitude moisture transport is important for MOC stability to anthropogenic climate change (Mikolajewicz et al. 2007), so the dependency of such transports on resolution in the atmosphere warrants further investigation.

We have also analyzed time-mean transports in a perturbed-phyics ensemble of HadCM3 models. The parameter perturbations applied in these models lead to a range of mean climate states (e.g., MOC strength, TOA fluxes, etc). In this ensemble we observe a similar tendency for global energy transports in the ocean and atmosphere to be anticorrelated, but in a quasi-equilibrium sense. On average, model versions with a net radiative surplus have a stronger poleward energy transport in the atmosphere. This is partly offset by a reduced poleward ocean heat transport (Fig. 12) due to a reduction of heat transport by the Atlantic MOC and shallow subtropical overturning cells. Model versions with heat transport and overturning close to observed values are the ones closest to a net global radiative equilibrium—at least at the centennial scales at which the ensemble was run. At longer, millennial time scales adjustment of the deep ocean could alter that relation.

The anticorrelation at these centennial time scales between energy transports in ocean and atmosphere across slightly different model versions has some relevance for climate model development because that also involves model runs of decades to centuries. Our results suggest that coupled model versions with an excessive radiative surplus at low latitudes may be limited in their ability to remove this energy: although the atmosphere can transport some of this excess energy poleward and radiate it out into space, the ocean absorbs the remainder and transports a sizable fraction of heat back to lower latitudes. The ocean transport can be interpreted as providing a positive feedback on low-latitude radiative imbalance at these centennial scales.

Results from the perturbed physics experiments suggest that zonal-mean biases in ocean and atmosphere components of a coupled model can be interpreted, and their seriousness perhaps ranked, in terms of their impacts on the global energy budget, clearly a key feature of climate models: For example, errors in zonal winds at low latitudes affect the strength of the shallow subtropical cells in the ocean, which has a strong impact on ocean poleward heat transport. Finally, if anticorrelation between energy transports in the oceans and atmosphere also exists in other, structurally different climate models, then that may provide a powerful way to understand and quantify differences between these climate models.

Acknowledgments

Thanks to David Sexton and Glen Harris for their help in setting up the perturbed physics experiments. Suggestions from Anne Pardaens and Richard Wood are thankfully acknowledged, as are stimulating discussions with Carl Wunsch and Jonathan Gregory. Valuable comments from three anonymous reviewers have helped to improve the text. This work was funded by the Joint Defra and MoD Programme (Defra) GA01101 (MoD) CBC/2B/0417_Annex C5.

REFERENCES

  • Bjerknes, J., 1964: Atlantic air-sea interaction. Advances in Geophysics, Vol. 10, Academic Press, 1–82.

  • Church, J. A., , J. M. Gregory, , P. Huybrechts, , M. Kuhn, , K. Lambeck, , M. T. Nhuan, , D. Qin, , and P. L. Woodworth, 2001: Changes in sea level. Climate Change 2001: The Scientific Basis, J. T. Houghton et al., Eds., Cambridge University Press, 639–693.

    • Search Google Scholar
    • Export Citation
  • Collins, M., , B. B. B. Booth, , G. R. Harris, , J. M. Murphy, , D. M. H. Sexton, , and M. J. Webb, 2006: Towards quantifying uncertainty in transient climate change. Climate Dyn., 27 , 127147.

    • Search Google Scholar
    • Export Citation
  • Cox, P. M., , R. A. Betts, , C. B. Bunton, , R. L. H. Essery, , P. R. Rowntree, , and J. Smith, 1999: The impact of new land surface physics on the GCM simulation of climate and climate sensitivity. Climate Dyn., 15 , 183203.

    • Search Google Scholar
    • Export Citation
  • Delworth, T. L., , and M. E. Mann, 2000: Observed and simulated multidecadal variability in the Northern Hemisphere. Climate Dyn., 16 , 661676.

    • Search Google Scholar
    • Export Citation
  • Ganachaud, A., , and C. Wunsch, 2003: Large-scale ocean heat and freshwater transports during the World Ocean Circulation Experiment. J. Climate, 16 , 696705.

    • Search Google Scholar
    • Export Citation
  • Gordon, C., , C. Cooper, , C. A. Senior, , H. Banks, , J. M. Gregory, , T. C. Johns, , J. F. B. Mitchell, , and R. A. Wood, 2000: The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Climate Dyn., 16 , 147168.

    • Search Google Scholar
    • Export Citation
  • Gregory, J. M., , P. Huybrechts, , and S. C. B. Raper, 2004: Climatology: Threatened loss of the Greenland ice-sheet. Nature, 428 , 616.

  • Held, I. M., , and A. Y. Hou, 1980: Nonlinear axially symmetric circulations in a nearly inviscid atmosphere. J. Atmos. Sci., 37 , 515533.

    • Search Google Scholar
    • Export Citation
  • Hewitt, C. D., , R. J. Stouffer, , A. J. Broccoli, , J. F. B. Mitchell, , and P. J. Valdes, 2003: The effect of ocean dynamics in a coupled GCM simulation of the last glacial maximum. Climate Dyn., 20 , 203218.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and D. J. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38 , 11791196.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and P. J. Valdes, 1990: On the existence of storm tracks. J. Atmos. Sci., 47 , 18541864.

  • Hughes, T. M. C., , and A. J. Weaver, 1994: Multiple equilibria of an asymmetric two-basin ocean model. J. Phys. Oceanogr., 24 , 619637.

    • Search Google Scholar
    • Export Citation
  • Inatsu, M., , H. Mukougawa, , and S-P. Xie, 2002: Stationary eddy response to surface boundary forcing: Idealized GCM experiments. J. Atmos. Sci., 59 , 18981915.

    • Search Google Scholar
    • Export Citation
  • Jayne, S. R., , and J. Marotzke, 2002: The oceanic eddy heat transport. J. Phys. Oceanogr., 32 , 33283345.

  • Johns, T. C., and Coauthors, 2003: Anthropogenic climate change for 1860 to 2100 simulated with the HadCM3 model under updated emissions scenarios. Climate Dyn., 20 , 583612.

    • Search Google Scholar
    • Export Citation
  • Jones, G. S., , J. M. Gregory, , P. A. Stott, , S. F. B. Tett, , and R. B. Thorpe, 2005: An AOGCM simulation of the climate response to a volcanic super-eruption. Climate Dyn., 25 , 725738.

    • Search Google Scholar
    • Export Citation
  • Knight, J. R., , R. J. Allan, , C. K. Folland, , M. Vellinga, , and M. E. Mann, 2005: A signature of persistent natural thermohaline circulation cycles in observed climate. Geophys. Res. Lett., 32 .L20708, doi:10.1029/2005GL024233.

    • Search Google Scholar
    • Export Citation
  • Manabe, S., , and R. J. Stouffer, 1988: Two stable equilibria of a coupled ocean–atmosphere model. J. Climate, 1 , 841866.

  • Manabe, S., , and R. J. Stouffer, 1999: The role of thermohaline circulation in climate. Tellus, 51A , 91109.

  • Marsh, R., and Coauthors, 2004: Bistability of the thermohaline circulation identified through comprehensive 2-parameter sweeps of an efficient climate model. Climate Dyn., 23 , 761777.

    • Search Google Scholar
    • Export Citation
  • McCreary Jr., J. P., , and P. Lu, 1994: Interaction between the subtropical and equatorial ocean circulations: The subtropical cell. J. Phys. Oceanogr., 24 , 466497.

    • Search Google Scholar
    • Export Citation
  • Mikolajewicz, U., , M. Gröger, , E. Maier-Reimer, , G. Schurgers, , M. Vizcaíno, , and A. M. E. Winguth, 2007: Long-term effects of anthropogenic CO2 emissions simulated with a complex earth system model. Climate Dyn., 28 , 599633.

    • Search Google Scholar
    • Export Citation
  • Murphy, J. M., , D. M. H. Sexton, , D. N. Barnett, , G. S. Jones, , M. J. Webb, , M. Collins, , and D. A. Stainforth, 2004: Quantification of modelling uncertainties in a large ensemble of climate change simulations. Nature, 430 , 768772.

    • Search Google Scholar
    • Export Citation
  • Neelin, J. D., , and I. M. Held, 1987: Modeling tropical convergence based on the moist static energy budget. Mon. Wea. Rev., 115 , 312.

    • Search Google Scholar
    • Export Citation
  • Peixoto, J. P., , and A. H. Oort, 1992: Physics of Climate. American Institute of Physics, 520 pp.

  • Pope, V. D., , and R. A. Stratton, 2002: The processes governing horizontal resolution sensitivity in a climate model. Climate Dyn., 19 , 211236.

    • Search Google Scholar
    • Export Citation
  • Pope, V. D., , M. L. Gallani, , P. R. Rowntree, , and R. A. Stratton, 2000: The impact of new physical parametrizations in the Hadley Centre climate model: HadAM3. Climate Dyn., 16 , 123146.

    • Search Google Scholar
    • Export Citation
  • Rahmstorf, S., 1996: On the freshwater forcing and transport of the Atlantic thermohaline circulation. Climate Dyn., 12 , 799811.

  • Rahmstorf, S., 2002: Ocean circulation and climate during the past 120 000 years. Nature, 419 , 207214.

  • Roberts, M. J., and Coauthors, 2004: Impact of an eddy-permitting ocean resolution on control and climate change simulations with a global coupled GCM. J. Climate, 17 , 320.

    • Search Google Scholar
    • Export Citation
  • Schiller, A., , U. Mikolajewicz, , and R. Voss, 1997: The stability of the North Atlantic thermohaline circulation in a coupled ocean-atmosphere general circulation model. Climate Dyn., 13 , 325347.

    • Search Google Scholar
    • Export Citation
  • Schmittner, A., , M. Latif, , and B. Schneider, 2005: Model projections of the North Atlantic thermohaline circulation for the 21st century assessed by observations. Geophys. Res. Lett., 32 .L23710, doi:10.1029/2005GL024368.

    • Search Google Scholar
    • Export Citation
  • Schneider von Deimling, T., , H. Held, , A. Ganopolski, , and S. Rahmstorf, 2006: Climate sensitivity estimated from ensemble simulations of glacial climate. Climate Dyn., 27 , 149163.

    • Search Google Scholar
    • Export Citation
  • Seager, R., , D. S. Battisti, , J. Yin, , N. Gordon, , N. Naik, , A. C. Clement, , and M. A. Cane, 2002: Is the Gulf Stream responsible for Europe’s mild winters. Quart. J. Roy. Meteor. Soc., 128 , 25632586.

    • Search Google Scholar
    • Export Citation
  • Shaffrey, L., , and R. Sutton, 2004: The interannual variability of energy transports within and over the Atlantic Ocean in a coupled climate model. J. Climate, 17 , 14331448.

    • Search Google Scholar
    • Export Citation
  • Shaffrey, L., , and R. Sutton, 2006: Bjerknes compensation and the decadal variability of the energy transports in a coupled climate model. J. Climate, 19 , 11671181.

    • Search Google Scholar
    • Export Citation
  • Shepherd, T. G., 1993: Nonlinear saturation of baroclinic instability. Part III: Bounds on the energy. J. Atmos. Sci., 50 , 26972709.

  • Stainforth, D. A., and Coauthors, 2005: Uncertainty in predictions of the climate response to rising levels of greenhouse gases. Nature, 433 , 403406.

    • Search Google Scholar
    • Export Citation
  • Stouffer, R. J., and Coauthors, 2006: Investigating the causes of the response of the thermohaline circulation to past and future climate changes. J. Climate, 19 , 13651387.

    • Search Google Scholar
    • Export Citation
  • Talley, L. D., , J. L. Reid, , and P. E. Robbins, 2003: Data-based meridional overturning streamfunctions for the global ocean. J. Climate, 16 , 32133226.

    • Search Google Scholar
    • Export Citation
  • Thorpe, R. B., , J. M. Gregory, , T. C. Johns, , R. A. Wood, , and J. F. B. Mitchell, 2001: Mechanisms determining the Atlantic thermohaline circulation response to greenhouse gas forcing in a non-flux-adjusted coupled climate model. J. Climate, 14 , 31023116.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , and J. M. Caron, 2001: Estimates of meridional atmosphere and ocean heat transports. J. Climate, 14 , 34333443.

  • Vellinga, M., , and R. A. Wood, 2002: Global climatic impacts of a collapse of the Atlantic thermohaline circulation. Climatic Change, 54 , 251267.

    • Search Google Scholar
    • Export Citation
  • Vellinga, M., , and R. A. Wood, 2007: Impacts of thermohaline circulation shutdown in the twenty-first century. Climatic Change, doi:10.1007/s10584-006-9146-y.

    • Search Google Scholar
    • Export Citation
  • Vellinga, M., , R. A. Wood, , and J. M. Gregory, 2002: Processes governing the recovery of a perturbed thermohaline circulation in HadCM3. J. Climate, 15 , 764780.

    • Search Google Scholar
    • Export Citation
  • Wood, R., , M. Collins, , J. Gregory, , G. Harris, , and M. Vellinga, 2006: Towards a risk assessment for shutdown of the Atlantic thermohaline circulation. Avoiding Dangerous Climate Change, H. J. Schellnhuber et al., Eds., Cambridge University Press, 49–54.

    • Search Google Scholar
    • Export Citation
  • Wunsch, C., 1999: Where do ocean eddy heat fluxes matter. J. Geophys. Res., 104 , 1323513250.

  • Wunsch, C., 2005: The total meridional heat flux and its oceanic and atmospheric partition. J. Climate, 18 , 43744380.

  • Zhang, R., , and T. L. Delworth, 2005: Simulated tropical response to a substantial weakening of the Atlantic thermohaline circulation. J. Climate, 18 , 18531860.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y-C., , and W. B. Rossow, 1997: Estimating meridional energy transports by the atmospheric and oceanic general circulations using boundary fluxes. J. Climate, 10 , 23582373.

    • Search Google Scholar
    • Export Citation

APPENDIX A

Calculation of Energy Transport in the Atmosphere

In the atmosphere energy transport occurs mainly in the form of sensible heat, latent heat, and potential energy, which can be combined into moist static energy (e.g., Neelin and Held 1987). The proportion of kinetic energy is very small compared to these other forms (Peixoto and Oort 1992) and will be neglected. For the long model integrations of CON and HOS and the perturbed physics experiments in section 4, we do not have the individual contributions to the MSE transport available. But, if we assume that the atmosphere has a negligible heat capacity, we can calculate the implied total meridional transport M(y) of MSE at each latitude y from the heat fluxes at the surface (Fsurf) and the TOA (FTOA) (e.g., Zhang and Rossow 1997):
i1520-0442-21-3-561-ea1
(the angled brackets indicate the zonal integral). We also reran the model for 20 yr of CON and years 220–240 of HOS, respectively, storing the actual values of M and of transports of latent and sensible heat and potential energy as the runs progressed. This enabled us to confirm the accuracy of Eq. (1) at decadal time scales, which we subsequently used to calculate M.

APPENDIX B

Perturbed Physics Experiments

Parameters across the atmosphere and sea ice components were varied within boundaries that mark the range of uncertainty, according to experts. Additionally, some parameterization schemes are either switched on or off. The aim was to create an ensemble of non-flux-adjusted coupled model versions with a range of hydrological and climate sensitivity to anthropogenic CO2 increase, based on existing perturbed physics ensembles of atmosphere–ocean mixed layer models (Murphy et al. 2004; Stainforth et al. 2005). The physics perturbations are listed in Tables B1 and B2, and a brief description of the meaning of the parameters is given in Table B3. Tables B1–B3 are presented as an electronic supplement (DOI: 10.1175/2007JCLI1754S.1). Further details about the model parameters and schemes can be found in the supplementary material of Murphy et al. (2004).

Fig. 1.
Fig. 1.

Strength of the maximum meridional overturning in the Atlantic Ocean between 40° and 55°N in CON (solid) and HOS (dashed). The northward ocean heat transport in the Atlantic across 30°N for HOS is shown by the dotted line (right axis).

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 2.
Fig. 2.

Global-mean net radiative flux at TOA in experiment HOS (heavy solid, decadal averages). Changes in TOA clear-sky longwave (dashed), clear-sky shortwave (dotted), and cloud radiative fluxes (dashed–dotted) are for experiment HOS (decadal averages) minus the time average of CON. Downward is defined as positive.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 3.
Fig. 3.

(a) Difference in net TOA radiative flux (solid) and its separate components (clear-sky longwave, clear-sky shortwave, and cloud radiative fluxes), years 230–240 of HOS relative to CON. Downward is defined as positive. (b) Zonal-mean TOA anomalous cloud radiative fluxes. Significant changes (at the 5% level) in (a) and (b) are shown as heavy lines. (c) Zonal-mean changes in total cloud water content. Significant changes (at the 5% level) are shown by the gray shading (in 10−6 kg kg−1). Cloud diagnostics have the hybrid vertical model coordinate (“Eta”); nominal pressure levels are shown on the right.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 4.
Fig. 4.

(a) Global northward energy transport by the atmosphere (solid) and ocean (dashed). Data from the hosing (control) run (years 230–240) are shown by heavy (thin) lines. (b) Change (HOS minus CON) in energy transport by the atmosphere (solid) and the ocean (dashed). Their sum (dotted line) shows the overall change in meridional transport in the climate system.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 5.
Fig. 5.

Northward global energy transport at the equator by the atmosphere (solid) and ocean (dashed), and their sum (dotted).

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 6.
Fig. 6.

(a) Zonally integrated northward atmospheric transports in CON of sensible heat (SH), latent heat (LH), potential energy (PE), and moist static energy (MSE). (b) Anomalies in atmospheric transports of the same quantities for years 230–240 of HOS, relative to CON.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 7.
Fig. 7.

Zonal-mean zonal wind (colors, in m s−1) and streamfunction (contours, in 1010 m3 s−1, positive values clockwise) anomalies in years 230–240 of HOS, relative to CON. The differences shown are significant at the 5% level.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 8.
Fig. 8.

(a) Wavenumber spectrum from CON of meridional sensible heat transport as a function of latitude and wavenumber [units (K m s−1)2 m−1]. This spectrum for the total flow (i.e., time mean plus transient) was calculated by averaging 20 years of daily mean k-spectra. (b) Change in wavenumber spectrum in years 230–240 of HOS relative to spectrum of CON from (a). Negative contours are dotted, zero contour is heavy, and the gray shading indicates where the change exceeds the noise level of CON (estimated as twice the annual-mean standard deviation of the spectra in CON).

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 9.
Fig. 9.

(a) Change in total northward ocean heat transport in the various basins during years 230–240 of HOS, compared to CON (bold lines). Thin lines show the change in implied meridional transports between HOS and CON, found by accumulating the surface heat flux difference between HOS and CON. “Indian” refers to the Indian Ocean north of the Indonesian Throughflow, “Pacific” refers to the net heat transport in the Pacific Ocean plus the contribution from the Indian Ocean south of the throughflow. (b) Change in heat transport carried by the meridional overturning.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 10.
Fig. 10.

Regressions against global-mean TOA net radiative flux of (a) zonal-mean TOA radiative flux and (b) meridional energy transport in the atmosphere (heavy solid) and global ocean (dashed). For latitudes north of 30°N changes in ocean heat transport in the Atlantic (dotted) and Pacific (dashed–dotted) are shown. Long-term (50 yr or more) average data were used from all 35 perturbed physics experiments; the gray shading is the estimated standard error of the regression (for clarity, errors for Atlantic and Pacific Ocean heat transport have been omitted).

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 11.
Fig. 11.

Regression of global zonal-mean meridional overturning [grayscale, in Sv (W m−2)−1] and global zonal-mean zonal density [white contours, in kg m−3 (W m−2)−1] onto global-mean TOA flux across perturbed physics ensemble. Negative streamfunction contours are dotted and indicate counterclockwise circulation; positive streamfunction contours are solid. Most features of the regressions are statistically significant at the 5% level (significance not shown for sake of clarity).

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 12.
Fig. 12.

Meridional energy transports and integrated fluxes in perturbed physics ensemble, expressed as regressions against global-mean TOA net radiative flux [PW (W m−2)−1]. For both figures long-term (50 yr or more) average data were used from all 35 ensemble members.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 13.
Fig. 13.

Global-mean TOA flux (horizontal) and maximum global MOC at 45°N in the perturbed physics ensemble. Observational estimates of MOC and heat transport are shown near the respective ordinates, together with error estimates. Ensemble members for which heat transport and MOC strength are consistent with these observed estimates are shown by solid symbols.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Fig. 14.
Fig. 14.

Diagram of changes in energy transports between climate states with and without MOC (years 230–240 of experiment HOS minus CON). Shown are changes in total meridional transports in the ocean (dotted) and atmosphere (white), and area-integrated fluxes across the air–sea interface (dashed) and TOA (solid), all in PW.

Citation: Journal of Climate 21, 3; 10.1175/2007JCLI1754.1

Supplementary Materials

Save