Examinations of the impact of anthropogenic aerosols on oceanic heat content have focused largely on the global average response. Given that aerosol-induced cooling is greater in the Northern Hemisphere (NH) than in the Southern Hemisphere (SH), do aerosols induce a greater impact on NH oceanic heat content? Sea level rise over the past 50 years has shown little hemispheric differentiation. Using a set of global climate model experiments forced with and without anthropogenic aerosols, we show that increasing aerosols in the 20th century induce a pan-oceanic heat redistribution. This leads to a reduction in the SH oceanic heat content comparable to that in the NH oceans. The process includes a strengthening of the northward cross-equatorial heat transport in the Atlantic and Pacific Oceans, with the majority taking place in the Atlantic Ocean via the most effective pathway: the globally interconnected ocean current system associated with the Atlantic overturning.
 The direct and indirect effects of anthropogenic aerosols tend to offset the climatic warming due to greenhouse gases [Mitchell et al., 1995] and also reduce the amount of solar irradiance reaching the Earth's surface [Liepert et al., 2004; Nazarenko and Menon, 2005]. The spatially inhomogeneous distribution of aerosol forcing generates a stronger surface cooling in the Northern Hemisphere (NH) than in the Southern Hemisphere (SH) [Rotstayn and Lohmann, 2002]. Recent studies have also suggested the possibility of large impacts of aerosol forcing on regional climate, such as droughts in the Sahel [Rotstayn and Lohmann, 2002], summertime floods and droughts in China [Menon et al., 2002], a weakening of the South Asian monsoon [Ramanathan et al., 2005], and an increasing rainfall trend over northern Australia (L. D. Rotstayn et al., Have Australian rainfall and cloudiness inceased due to the remote effects of anthropogenic aerosols?, submitted to Journal of Geophysical Research, 2006, hereinafter referred to as Rotstayn et al., submitted manuscript, 2006). However, to date, much less is known about the impact aerosols have on the ocean circulation. It has been demonstrated that aerosols from episodic volcanic events have a strong multidecadal-long imprint on global oceanic heat content and hence sea level [Church et al., 2005; Gleckler et al., 2006], and anthropogenic aerosols, like volcanic aerosols, have a strong cooling signature in the globally averaged oceanic heat content [Delworth et al., 2005]. These studies focus on the global mean response and, as such, the underlying oceanic adjustment process is yet to be fully examined. Given that aerosols induce a stronger surface cooling in the NH than in the SH, one might expect aerosols to produce a stronger impact on oceanic heat content in the NH. However, sea level rise over the past 50 years from reconstructed data shows little difference between both hemispheres [Church et al., 2004], suggesting that an oceanic heat redistribution process from the SH to the NH operates to reduce the hemispheric contrast. We investigate this hypothesis using a 20th century simulation from a coupled ocean-atmosphere global climate model (GCM). We find that the primary mechanism for northward hemispheric heat transport is the global Conveyor [Gordon, 1986]. This is consistent with the finding of Delworth and Dixon  in another climate model, who show that aerosols delay the CO2-induced slowdown of the Atlantic overturning circulation. This is also consistent with the finding that multidecadal variations of interhemispheric temperature contrast are linked with global ocean variability [e.g. Schlesinger and Ramankutty, 1994].
2. Model and Experiments
 The atmospheric model used in this study is a low-resolution (spectral R21) version of the CSIRO atmospheric GCM. The R21 model has 18 hybrid vertical levels and a horizontal resolution of approximately 5.6° in longitude and 3.2° in latitude. The oceanic component is based on the Cox-Bryan code [Cox, 1984], and has the same horizontal resolution as the atmospheric model, with 21 levels in the vertical. The model uses flux adjustments to mitigate spurious climate drift. Further details can be found in Gordon and O'Farrell  and Cai et al. . Note that the model's climatological Atlantic overturning and the associated heat transport are too low compared with those inferred from observations, as will be discussed further.
 The model has a comprehensive aerosol scheme to perform climate simulations for the period 1871 to 2000, with and without the effects of anthropogenic aerosols (Rotstayn et al., submitted manuscript, 2006). The aerosol species treated interactively are sulfate, particulate organic matter (POM), black carbon (BC), mineral dust and sea salt. As with the direct effects of these aerosols on shortwave radiation, the indirect effects of sulfate, POM and sea salt on liquid-water clouds are included in the model. Historical emission inventories are used for sulfur, POM and BC derived from the burning of fossil fuels and biomass. Other forcings included are those due to changes in long-lived greenhouse gases, ozone, volcanic aerosol and solar variations. We performed an eight-run ensemble with all of these forcings (ALL ensemble) and a further eight runs with all forcings except those related to anthropogenic aerosols (AXA ensemble). The AXA ensemble only differs from that of the ALL ensemble in that the anthropogenic emissions of sulfur, POM and BC were fixed at their 1870 levels. The same set of experiments has been used to assess the impact of aerosols on Australian rainfall and cloudiness (Rotstayn et al., submitted manuscript, 2006). The differences between the two ensembles averages, ALL minus AXA (referred to as “aerosol-induced”), were used to deduce the anthropogenic aerosol effect. In addition, we use a multi-century control experiment where the model is run with fixed constant forcing at pre-industrial levels to determine the climatological state of the model solution and the uncertainty range.
 As in other studies, the inclusion of aerosols in GCMs improves the agreement between the models and observations as documented by Rotstayn et al. (submitted manuscript, 2006). The ALL ensemble clearly gives a better simulation of the observed global-mean surface temperature changes; without the cooling effect of aerosols, the AXA ensemble overestimates the warming, particularly after 1950 (Figure 1a).
3.1. Hemispheric Average of Surface Heat Flux and Oceanic Heat Content
Figure 1b shows a time series of aerosol-induced changes in the oceanic surface heat flux averaged over both hemispheres. A reduction in the heat going into the NH oceans, i.e., a surface cooling is induced by aerosol forcing, however there is no trend in the heat into the SH oceans. Thus if aerosols induce a SH cooling, it is not via the surface. Figure 1c shows a time series of aerosol-induced change in the NH and SH oceanic heat content in terms of full-depth averaged temperature. In sharp contrast to the surface heat flux, there is a substantial significant decrease in the SH oceanic heat content starting from the 1920s. Indeed, by the late 1990s, the reduction in SH oceans is comparable (but lagging slightly) to the reduction in the NH oceans (Figure 1c).
 To assess the statistical significance of aerosol-induced trend during 1921–2000, we divide the control experiment into two periods, determine the difference between the periods, and assess the amplitude of an 80-year trend due to natural variability. We obtain a time series of the 80-year trends using a sliding window, and calculate the standard deviation. Taking into account the ensemble members, the uncertainty is estimated against a 99% significant level. The NH heat flux trend is 0.6 ± 0.16 W m−2 century−1, the full-depth temperature trend for the NH is 0.0675 ± 0.004 °C century−1, and similar for the SH. Our result confirms that the trends in the NH heat flux and the oceanic full-depth temperature are statistically significant, as the uncertainty range is far smaller than the trends.
3.2. Trends in Meridional Heat Transport
 The lack of a SH oceanic surface flux trend means that the reduction in the SH oceanic heat content must be achieved through a cross-equatorial oceanic heat transport from the SH oceans to the NH oceans. This is confirmed by the aerosol-induced trends in the oceanic northward heat transport (Figure 2a) in both the Atlantic Ocean and Indian and Pacific Oceans combined (Indo-Pacific system). The northward oceanic heat transport is calculated from heat advected by meridional currents integrated from the east to the west boundary and from the bottom to the surface. The Indo-Pacific system is treated as one basin because of the presence of the Indonesian Throughflow passage.
Figure 2a shows that at all latitudes in the Atlantic, northward heat transport intensifies. Further, approximately two thirds of the total cross-equatorial northward heat transport takes place in the Atlantic and one third is through the Indo-Pacific system, meaning that two thirds of the cooling seen in the SH ocean heat content in Figure 1c is conducted through a cross-equatorial heat transport from the South to the North Atlantic. By the late 20th century, aerosols induce an increase in the cross-equatorial Atlantic heat transport of 0.045 PW, approximately 10% of the climatological value in the control experiment. Figure 2b shows that the trend starts to accelerate in the early 1960s. Again these trends are statistically significant; the standard deviation of 80-year trends in the control experiment due to variability is more than an order of magnitude smaller than the trends presented in Figure 2.
 The trend of the total surface heat loss through the NH ocean surface is 0.175 PW century−1. The contribution by the SH oceans to the NH heat loss is conducted through the global cross-equatorial heat transport change, about 0.11 PW century−1 (Figure 2a, blue curve); the NH oceans provide the difference, 0.065 PW century−1. Thus the contribution by the NH ocean is approximately 60% of the contribution from the SH oceans. Given that the volume of the NH oceans is about 60% of that of the SH oceans, this hemispheric partition of the contribution to the heat loss is consistent with the comparable reduction in the full-depth oceanic heat content between both hemispheres shown in Figure 1c.
 Out of the NH heat loss of 0.175 PW century−1, a total of 0.06 PW century−1 takes place in the northern Indian-Pacific system and the rest of 0.115 PW century−1 occurs in the North Atlantic, consistent with a greater cross-equatorial heat transport in the Atlantic. Associated with the increase in the Atlantic cross-equatorial heat transport is an aerosol-induced intensification in the Atlantic overturning circulation (Figure 3), similar to that simulated in the GFDL CM2.0 model [Delworth et al., 2005], highlighting the robustness of the identified process. As with that model, without aerosol forcing (but with increasing greenhouse gases and other forcing) the Atlantic overturning shows a significant decrease, however exhibits little change in the presence of aerosol forcing. In our model, aerosol induces an increase of approximately 2.5 Sv century−1, or 1.3 Sv per 50 years, i.e., some 10% of the climatological value of 12.5 Sv, commensurate with the increased northward heat transport in the equatorial Atlantic. The structure shows an increased northward flow from the South to the North Atlantic in the upper 1500 m, and increased southward flow below this depth.
 Recent observational evidence [Bryden et al., 2005] suggests that part of the deep southward flow of the Atlantic overturning has been decreasing over the past 50 years. Our results could mean that aerosols have been trying to “protect it”. Without this protection, the slowing would be more pronounced. In our model the overturning shows little change over the 20th century in the presence of the aerosol forcing, however without aerosol forcing the overturning weakens by approximately 10%. We note however that the observations imply an estimate of the Atlantic overturning of 15 Sv [Schmittner et al., 2005] and the Atlantic cross-equatorial heat transport of 0.7–0.9 PW [Trenberth and Caron, 2001]. Thus the modeled values are too weak, and it is not clear how they affect the model sensitivity to aerosol forcing.
 In the Indo-Pacific system (Figure 3b), between 20°S and 15°N a cross-equatorial overturning cell is generated, featuring a northward flow in the upper 250 m, consistent with the intensified cross-equatorial northward heat transport. The near surface northward flows are balanced by increased southward flows down to approximate 800 m; below this depth, the flows veer northwards and upwards from the southern Indo-Pacific to 40°N (North Pacific) and 15°S of the North Indian Ocean, and upwell to the surface. An examination reveals that most of the cross-equatorial heat transport increases in the Indo-Pacific system are carried out in the upper 250 m of the Pacific.
3.3. Trends in the Interconnected Conveyer Ocean Currents
 There are two mutually consistent factors that determine the feature of a greater aerosol-induced cross-equatorial transport in the Atlantic than the Indo-Pacific system. First, aerosols result in a greater heat loss in the North Atlantic than in the North Pacific. Second, the Atlantic is the only ocean with a climatological northward heat transport at all latitudes. As such it is the most effective in transporting heat from the SH oceans to the NH; this feature is in turn linked to the globally interconnected ocean currents of the Conveyor. The driving force of the Conveyor is the cold, salty water of the North Atlantic Ocean, which sinks to the deep ocean, a process known as the North Atlantic Deep Water Formation [Gordon, 1986]. This deep water flows southward to the southern tip of Africa. It then joins the Antarctic Circumpolar Current, veers northward into the Indo-Pacific system, eventually returning to the South Atlantic, propelling warm surface water back into the North Atlantic. The associated Atlantic meridional overturning circulation carries warm upper water from the South Atlantic into the North Atlantic. In this way, the heat that is transported to the North Atlantic is partially derived from the Indo-Pacific system.
 These processes are in place in the model control climate, and strengthen in response to changes in surface heat flux induced by aerosols. Figure 4 shows aerosol-induced trends in the upper-750m oceanic currents and full-depth averaged oceanic temperature for 1951–2000. The flow depicts a strengthening of the Conveyor, with a stronger compensating northward flow from the South Atlantic to the formation region in the North Atlantic as discussed above. This northward flow is fed by a stronger Agulhas outflow and a stronger Indonesian Throughflow, supported by the increased upwelling in the Indo-Pacific system (Figure 3b). In the Pacific a northward flow along the western boundary, as a part of a cyclonic circulation trend, extends from the south into the north, looping into the North Pacific and eventually to the Indian Ocean via the Indonesian Throughflow passage. This is consistent with the pathway identified by previous studies [Hirst and Godfrey, 1993].
 The heat transported to the NH oceans contributes to the heat loss to the atmosphere seen in Figure 1b. In this way, the aerosol-induced NH oceanic cooling draws heat from the SH oceans (approximately 60%). Thus our model suggests that aerosols induce a vigorous pan-oceanic adjustment process redistributing heat from the SH oceans to the NH and homogenizing the changes in ocean heat content and sea level. This explains why the observed sea level rise shows no hemispheric differential.
 The aerosol-induced changes in oceanic heat content (Figure 4, contour) are not spatially uniform. In general, larger reductions take place along the pathway of the Conveyor, with a maximum reduction in the South Atlantic, some 30°S, where the northward flows heading to the North Atlantic commence. In the North Atlantic Deep Water formation region, heat content increases as aerosol-induced northward flows converge and contribute to an enhanced local heat loss.
 Other features of the flow trends include convergences/divergences of heat, which may imply a spin up/down of the gyres. In the North Pacific, there is a convergence between 10°N and 35°N - implying a spin up of the gyre. In the SH, there is a divergence from the equator to 20°S in both the Pacific and Indian, and a convergence south of 25°S, particularly in the Indian Ocean. These changes are consistent with stronger westward flows along the equatorial Pacific.
 The importance of oceanic processes in the climate system's response to aerosol forcing and the mechanism whereby the increasing level of aerosols over the 20th century affects the SH oceanic heat content have not previously been explored. Using a set of GCM experiments forced with and without increasing anthropogenic aerosols, we show that the increasing level of aerosols induces a pan-oceanic heat redistribution, leading to a reduction in the SH oceanic heat content comparable to that in the NH oceans. The process includes a strengthening of the northward cross-equatorial oceanic heat transport in the Atlantic and Pacific, with the majority taking place in the Atlantic Ocean via the most effective pathway: the globally interconnected ocean current system associated with the Atlantic overturning. The heat associated with the increasing northward transport is derived from the SH oceans through the Conveyor, reducing an increase in oceanic heat content, hence sea level rise, of the SH to the same degree as the NH. The strengthening of the Atlantic heat transport mitigates the slow-down of the Atlantic overturning. As aerosol levels are expected to decrease during the 21st century, the “projective” effect on the Conveyor is likely to decrease, and the Atlantic overturning slow-down, sea level rise including that of the SH, will accelerate.
 This work is supported by the CSIRO Wealth from Ocean Flagship and the Australian Greenhouse Office. We thank members of the CSIRO Climate Model and Application Team for developing the model.