An index of water-circulation strength in the North Atlantic Ocean has been derived from sea-level measurements. This provides fresh evidence of the ocean's leading role in multidecadal climate variability. See Letter p.508
The Earth has warmed considerably during the twentieth and twenty-first centuries, most probably because of the effects of greenhouse gases emitted as a result of human activities. But strong natural variability on a wide range of timescales, from monthly to multidecadal, is superimposed on the long-term global-warming trend and also causes considerable spatial variation of that trend. We therefore need to understand this variability to discriminate anthropogenic effects from natural climate forcing. On page 508 of this issue, McCarthy et al.1 report a method that enables the effects of ocean circulation on one of the most prominent examples of long-term climate variability — the Atlantic Multidecadal Oscillation (AMO) — to be identified from long-term sea-level data.
Natural climate variability can be generated internally by interactions within or between climate-system components such as the atmosphere, ocean and sea ice, and externally by factors such as volcanic eruptions. The AMO is an example of long-term climate variability associated with the ocean2. It represents quasi-periodic oscillations of sea-surface temperature (SST) in the North Atlantic Ocean that have a period of about 70 years. The AMO has been described from instrumental SST records2 going back to the mid-nineteenth century and reconstructed from proxy data3 for the past few centuries.
Climate models2,4 and analysis of surface heat fluxes5 have suggested that the AMO is an internal mode of climate variability originating from changes in the circulation of the Atlantic Ocean, but its origin is still debated. Signals seen in SST may derive from changes in the ocean's interior that drive circulation changes6, or from the influence of factors outside the ocean7. Lack of data from the ocean's subsurface hinders attempts to quantify the relative contributions of internal and external processes to the AMO.
McCarthy and colleagues' method for identifying how ocean circulation affects the AMO is based on the hypothesis that ocean currents on relatively large scales of space and time (several tens of kilometres and several weeks) are broadly geostrophic — that is, the direction and strength of the currents depends on the balance between the Coriolis force associated with Earth's rotation and the horizontal pressure-gradient force in the ocean (Fig. 1). Their approach builds on the proposal8 that the difference in dynamic height (the sea level caused by variations in the ocean's depth-integrated density) between the centres of large, primarily wind-driven, ocean-circulation systems called gyres in the subtropical and subpolar North Atlantic provides a proxy for long-term changes in mass transport by the Gulf Stream and its extension, the North Atlantic Current. The difference in heights is a measure of the pressure gradient between the gyres' centres and has previously been estimated8 from offshore sea-level records in the subtropics at Bermuda, and from long-term hydrographic data at subpolar latitudes.
But the Bermuda sea-level data are noisy; coastal tide gauges would provide more-robust signals1,8. By taking sea-level differences between coastal tide gauges to the south and north of Cape Hatteras — the boundary between the subtropical and subpolar gyres — as an estimate of the pressure gradient between the gyres, McCarthy and co-workers derived a simple index of the North Atlantic circulation strength (see Fig. 1 of the paper1) that closely correlates with heat transport in ocean models. Positive values of this index reflect an enhanced pressure gradient compared with the long-term mean and imply strong northward heat transport and increasing upper-ocean heat content in mid-latitude and subpolar regions of the North Atlantic. Negative values imply lowering of the pressure gradient, weakening of northward heat transport and diminishing heat content in the upper ocean. Although this index is influenced by many other factors in the Gulf Stream region, it seems to be an effective proxy for the AMO.
McCarthy et al. also find that fluctuations in the sea-level index (and therefore changes in ocean circulation) precede anomalies in upper-ocean heat content in the subpolar-gyre region by approximately two years. This time lag is important, because it potentially enables prediction of heat-content changes in the subpolar-gyre region from sea-level data, and thus prediction of AMO evolution and related climate anomalies.
The authors also used a numerical model of the ocean (an eddy-permitting ocean general circulation model, for those in the know), forced by observed atmospheric conditions at the surface, to investigate the interconnections between dynamic sea-level gradients, circulation strength and upper-ocean heat content, including time lags. Despite the limitations of such models, especially in the Gulf Stream region and at high latitudes, the simulations largely support the results from the data (Fig. 2 of ref. 1), which is reassuring.
McCarthy and colleagues' study supports previous conjectures6 that the ocean integrates chaotic atmospheric variability primarily associated with a climatic phenomenon known as the North Atlantic Oscillation9, and responds by generating SST variability on multidecadal time scales through dynamical ocean processes and with a time lag of about a decade — specifically by causing changes in the large-scale ocean circulation called the Atlantic Meridional Overturning Circulation (AMOC). The findings therefore further reinforce the idea that the AMO is an internal mode of climate variability.
The work also has crucial implications for our understanding of the ocean's role in climate variability and change. First, if the AMO originates from ocean dynamics, then attempts to detect early signs of the effects of anthropogenic climate change in the North Atlantic will be hampered. For example, the climate effects of AMOC slowdown projected by many climate models in response to global warming could be masked for decades by a positive phase of the AMO. Second, the AMO has been linked to climate anomalies in many regions of the globe10,11. The study therefore suggests that decadal predictions of climate over the North Atlantic and adjacent continents are possible if enough ocean data are available for forecast initialization. This would be of enormous societal benefit to many countries.
Further efforts are needed to conclusively prove that the ocean steers the AMO. Relevant instrumental records (including sea levels from tide gauges) are available only for the past several decades, with the longest dating back to the 1920s and 1930s. But a thorough assessment of the ocean's role in the multidecadal climate variability of the North Atlantic sector will require proxy data dating back for many centuries or even millennia. Multi-year experiments to measure ocean currents and cross-sections — using moored measuring devices, satellite data, and buoys that sample the upper ocean — have enabled studies over the past two decades and have captured long-term data. These programmes must continue, given the long timescales involved in the AMO. Climate models initialized with such data will provide more-reliable predictions of future climate than are currently available.
The synthesis of models and data will potentially allow the prediction of multidecadal variations. However, climate models still suffer from large biases, especially in the North Atlantic. Improvements to climate models are therefore also urgently needed. But it remains to be seen whether atmosphere–ocean interactions such as those described by McCarthy et al. will remain the same under anthropogenic climate change.Footnote 1
McCarthy, G. D., Haigh, I. D., Hirschi, J. J.-M., Grist, J. P. & Smeed, D. A. Nature 521, 508–510 (2015).
Knight, J. R., Allan, R. J., Folland, C. K., Vellinga, M. & Mann, M. E. et al. Geophys. Res. Lett. 32, L20708 (2005).
Delworth, T. L. & Mann, M. E. Clim. Dynamics 16, 661–676 (2000).
Latif, M. et al. J. Climate 17, 1605–1614 (2004).
Gulev, S.K., Latif, M., Keenlyside, N., Park, W. & Koltermann, K. P. Nature 499, 464–467 (2013).
Latif, M. in Ocean Circulation and Climate 2nd edn (eds Siedler, G., Griffies, S., Gould, J. & Church, J.) (Academic, 2013).
Booth, B. B. B., Dunstone, N. J., Halloran, P. R., Andrews, T. & Bellouin, N. Nature 484, 228–232 (2012).
Curry, R. G. & McCartney, M. S. J. Phys. Oceanogr. 31, 3374–3400 (2001).
Hurrell, J. W. Science 269, 676–679 (1995).
Sutton, R. T. & Hodson, D. L. R. Science 309, 115–118 (2005).
Zhang, R. & Delworth, T. L. Geophys. Res. Lett. 33, L17712 (2006).
About this article
Investigating the contour dynamics of the vortex eye on the laboratory model of a differentially rotating fluid
IOP Conference Series: Earth and Environmental Science (2019)
Contrasting spatial structures of Atlantic Multidecadal Oscillation between observations and slab ocean model simulations
Climate Dynamics (2019)
How Predictable Are the Arctic and North Atlantic Oscillations? Exploring the Variability and Predictability of the Northern Hemisphere
Journal of Climate (2018)
Oceanic forcing of the interhemispheric SST dipole associated with the Atlantic Multidecadal Oscillation
Environmental Research Letters (2018)
Predicting the Temporal Structure of the Atlantic Multidecadal Oscillation (AMO) for Agriculture Management in Mexico's Coastal Zone
Journal of Coastal Research (2018)