Climate science

Origins of Atlantic decadal swings

Temperature variability in the North Atlantic Ocean is the result of many competing physical processes, but the relative roles of these processes is a source of contention. Here, scientists present two perspectives on the debate.

The topic in brief

  • Atlantic multidecadal variability (AMV) describes alternating swings in sea surface temperature in the North Atlantic Ocean that occur with a characteristic timescale of several decades (Fig. 1).

    Figure 1: Atlantic multidecadal variability.

    The difference between the average sea surface temperature of the North Atlantic Ocean and the average global sea surface temperature swings between cool (blue) and warm (red) phases, with a characteristic timescale of several decades. Although this phenomenon has been studied for more than 20 years, there is still uncertainty about what processes drive the interdecadal variability. (Data from ref. 18.)

  • Such temperature changes have been linked to weather and climate phenomena ranging from the extreme African droughts of the 1980s to the current high levels of Atlantic hurricane activity1.

  • Conventional wisdom holds that AMV is driven mainly by internal processes in the Atlantic climate system — in particular, by naturally occurring changes in atmospheric and ocean circulation1.

  • However, some researchers have suggested that these cycles are instead driven by external factors, including anthropogenic emissions2.

  • Understanding the mechanisms that underlie AMV is crucial for developing models to predict future climate.

Integrate the whole system

Gabriel A. Vecchi & Thomas L. Delworth

To understand AMV, we must consider climate variability involving both oceanic and atmospheric processes, as well as the Atlantic's response to radiative forcing — the alteration in Earth's energy balance resulting from human-induced or natural changes. The structure, history and predictability of AMV provide insight into the mechanisms that control it.

AMV is thought to arise from internal variations in the climate system that are tied to a large-scale ocean current called the Atlantic meridional overturning circulation (AMOC)1. Variations in the AMOC modulate a northward movement of near-surface warm water and a compensating southward movement of deep, colder waters, driving changes in ocean temperature. However, the origin of AMV is not confined to the ocean. Shifts in the strength and position of North Atlantic winds, known as the North Atlantic Oscillation, can strengthen or weaken the AMOC and result in multidecadal temperature swings in the Atlantic3.

In 2015, it was argued that AMV could be a result of the inherent variability of the North Atlantic Oscillation and its rate of heat transfer to the ocean surface, with no role for ocean-circulation changes4. However, the North Atlantic Oscillation and the ocean temperature have a lagged relationship on decadal timescales, with Atlantic warming following atmospheric variability that extracts heat from the subpolar ocean3. This observation is inconsistent with the idea that AMV is a direct thermal response of the ocean to the North Atlantic Oscillation, without a role for ocean-circulation changes. Furthermore, AMOC variations have been successfully used to predict sea surface temperature in the North Atlantic5. The observed AMV therefore requires both atmospheric variations and a dynamic ocean adjustment3.

Another perspective is that the most-recent AMV cycle is a response to radiative forcing that is outside the North Atlantic climate system2. It has been argued that cooling in the 1960s to 1980s was partially due to increased volcanic activity and transport of dust from North Africa, and a peak in European and North American emissions of sulfate. Suspended particles (such as sulfate and dust) in the atmosphere, called aerosols, can reduce the amount of sunlight that reaches the ocean surface, and can alter cloud characteristics2. Evidence that aerosols contributed to the most recent AMV cycle is limited by large uncertainties in the climatic impact of aerosols — in particular, their interactions with clouds6.

To reconcile the seemingly divergent perspectives of internal climate variability and radiative forcing, we must recognize that they are not mutually exclusive. The North Atlantic can exhibit intrinsic decadal variability that is linked to ocean-circulation changes, while also responding to decadal changes in radiative forcing. The tropics and extratropics (the areas poleward of the tropics) have different AMV amplitudes, suggesting that distinct mechanisms might apply to each region, and AMV is seen in records extending for thousands of years7. The relevance of aerosols to AMV before the past half-century requires frameworks in which aerosols are an element of climate variability itself8 or in which aerosols trigger the swings in the AMOC9.

Although single explanations for climate phenomena are appealing, AMV involves both climate variability — resulting from atmospheric and oceanic processes — and the influence of radiative forcing. Progress in understanding and predicting AMV therefore requires the integration of a wide range of fields and consideration of the combined impact of a number of factors, rather than a reductionist focus on an individual process.

External modulation

Ben Booth

AMV is generally thought to be caused by internal processes in the Atlantic Ocean1. However, an emerging body of literature points to a different explanation. External forcing is capable of driving much (or even most) of the magnitude and timing of the most recent AMV cycles, in addition to multidecadal changes in associated tropical precipitation, wind shear and hurricane activity10. This forcing includes large volcanoes and changes in solar radiation9, and variations in industrial aerosol emissions2.

The idea that external forcing could have the dominant role in driving AMV has been slow to gain traction, for at least three reasons. First, the internal and external mechanisms share many patterns of response, which means that separating their respective fingerprints in the observed temperature record is difficult. Second, doubts have been raised6 about aerosol forcing because, in the study in which this was investigated2, the aerosol levels obtained were linked to biases in simulated oceanic heat content. Although other climate models reproduce similar external forcing of AMV without such biases11, this is not widely known.

Third, and perhaps most importantly, the majority of current models neglect aerosol–cloud interactions that have been shown to be a key factor in external forcing of AMV2. In the past decade, many estimates of external forcing have either used climate models that ignore these interactions12 or focused headline results on the average over many models, of which only a minority included these interactions13. However, in analyses that have incorporated aerosol–cloud interactions, there is good agreement that external forcing has driven AMV over the past 50 years13,14. Where such analyses disagree is on whether external forcing had the leading role in driving AMV before the past half-decade, and it is here that there is greater potential scope for a dominant internal-variability role.

The responses of AMV to external forcing and internal variability are similar, but they differ in the tropics. Climate models that include key processes (such as low cloud, wind–evaporation–temperature and dust feedbacks) simulate surface-temperature gradients better than those that omit them, but such models reproduce observed gradients only when external forcing is also included15. The tropical AMV response is important because it drove historical changes in tropical rainfall and hurricane activity. Current climate-model simulations of AMV driven by internal variability cannot reproduce the magnitude of the historical rainfall shifts — they can match15 or approach16 the observed magnitude only when external forcing is also taken into account.

The idea that AMV must be driven either by external forcing or by internal variability is probably a false dichotomy, but separating their relative roles remains the biggest challenge. For instance, modelling suggests that Atlantic Ocean circulation responds to external forcing, either through decadal variations in surface radiative forcing9,17 or through changes in the North Atlantic Oscillation9.

The presence of these credible and competing explanations should force us to critically re-evaluate both scenarios. What do models suggest that the magnitude, period and latitudinal spatial coherence of internal variability should actually be? Models that simulate externally forced AMV should not escape similar scrutiny. External forcing could have only a minor role in AMV, given the current modelling uncertainties in the key processes involving cloud feedbacks and aerosol–cloud interactions. However, if internal variability did indeed dominate observed changes, I would find it a remarkable coincidence that these changes match the timing and magnitude of AMV that we would expect from models driven by external forcing. Footnote 1


  1. 1.

    See all news & views


  1. 1

    Knight, J. R., Folland, C. K. & Scaife, A. A. Geophys. Res. Lett. 33, L17706 (2006).

    ADS  Article  Google Scholar 

  2. 2

    Booth, B. B. B., Dunstone, N. J., Halloran, P. R., Andrews, T. & Belloin, N. Nature 484, 228–232 (2012).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Delworth, T. L. et al. J. Clim. 30, 3789–3805 (2017).

    ADS  Article  Google Scholar 

  4. 4

    Clement, A. et al. Science 350, 320–324 (2015).

    ADS  CAS  Article  Google Scholar 

  5. 5

    Yeager, S., Karspeck, A., Danabasoglu, G., Tribbia, J. & Teng, H. J. Clim. 25, 5173–5189 (2012).

    ADS  Article  Google Scholar 

  6. 6

    Zhang, R. et al. J. Atmos. Sci. 70, 1135–1144 (2013).

    ADS  Article  Google Scholar 

  7. 7

    Knudsen, M. F., Seindenkrantz, M.-S., Jacobsen, B. H. & Kuijpers, A. Nature Commun. 2, 178 (2011).

    ADS  Article  Google Scholar 

  8. 8

    Wang, C., Dong, S., Evan, A. T., Foltz, G. R. & Lee, S. -K. J. Clim. 25, 5404–5415 (2012).

    ADS  Article  Google Scholar 

  9. 9

    Otterå, O. H., Bentsen, M., Drange, H. & Suo, L. Nature Geosci. 3, 688–694 (2010).

    ADS  Article  Google Scholar 

  10. 10

    Dunstone, N. J., Smith, D. M., Booth, B. B. B., Hermanson, L. & Eade, R. Nature Geosci. 6, 534–539 (2013).

    ADS  CAS  Article  Google Scholar 

  11. 11

    Takahashi, C. & Watanabe, M. Nature Clim. Change 6, 768–772 (2016).

    ADS  Article  Google Scholar 

  12. 12

    Ting, M., Kushnir, Y., Seager, R. & Li, C. J. Clim. 22, 1469–1481 (2009).

    ADS  Article  Google Scholar 

  13. 13

    Steinman, B. A., Mann, M. E. & Miller, S. K. Science 347, 988–991 (2015).

    ADS  MathSciNet  CAS  Article  Google Scholar 

  14. 14

    Bellucci, A., Mariotti, A. & Gualdi, S. J. Clim. (2017).

  15. 15

    Martin, E. R., Thorncroft, C. & Booth, B. B. B. J. Clim. 27, 784–806 (2014).

    ADS  Article  Google Scholar 

  16. 16

    Allen, R. J., Evan, A. T. & Booth, B. B. B. J. Clim. 28, 8219–8246 (2015).

    ADS  Article  Google Scholar 

  17. 17

    Cheng, W., Chiang, J. C. H. & Zhang, D. J. Clim. 26, 7187–7197 (2013).

    ADS  Article  Google Scholar 

  18. 18

    van Oldenborgh, G. J., te Raa, L. A., Dijkstra, H. A. & Philip, S. Y. Ocean Sci. 5, 293–301 (2009).

    ADS  Article  Google Scholar 

Download references

Author information



Corresponding authors

Correspondence to Gabriel A. Vecchi or Thomas L. Delworth or Ben Booth.

Related links

Related links

Related links in Nature Research

Climate science: Clouds unfazed by haze

Ocean science: The origins of a climate oscillation

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Vecchi, G., Delworth, T. & Booth, B. Origins of Atlantic decadal swings. Nature 548, 284–285 (2017).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.