Have aerosols caused the observed North Atlantic multidecadal variability?

By Jon Robson

Multidecadal changes in the North Atlantic sea surface temperature (NASST) have been linked to a range of important climate impacts in Europe, Africa (most notably Sahel rainfall) and North and South America. Indeed, in the mid 1990s an increase in hurricane numbers, and a shift in European climate (notably to wet and dull summers in the UK)  coincided with a shift back to anomalously warm temperatures last seen in the 1930s-1950s. In fact, the observed NASST has generally evolved somewhat differently to the global-mean  sea surface temperature for the past 150 years (which has become known as the Atlantic Multidecadal Oscillation or the AMO, see figure 1).

Figure 1. The Atlantic Multidecadal Oscillation (AMO). The red line shows the low-pass filtered area-averaged North Atlantic sea surface temperature (NASST). Blue shows the low-pass filtered sea surface temperature outside the North Atlantic. Taken from Sutton and Dong, 2012.

In general, the majority of research has attributed the multidecadal variability in NASST to natural variability. Such an attribution arises because climate models could not recreate the observed changes in NASST when simulating the response to changes in historical radiative forcing (e.g. greenhouse gases, or aerosols). However,  the same models generally simulate AMO like variations in NASST spontaneously in their control integrations (i.e. when there are no forcings). The details of exactly how NASST varies usually differs between models. However, generally, the model’s NASST variability is  associated with coherent multidecadal fluctuations of  ocean circulation and its associated heat transports.

In contrast to natural variability, it has also been suggested that radiative forcing, particularly from aerosols, could still have played a role in the evolution of NASST. Aerosols can adjust the radiation budget by interacting with the incoming solar shortwave radiation directly, by absorbing or scattering the photons, or indirectly, by modifying cloud properties. Interestingly, emissions of sulfur dioxide (an important anthropogenic aerosol that acts to cool the climate) increased substantially from the U.S.A. and Europe  in the 1950s-1980s, and then reduced following clean-air legislation. As aerosols are only resident in the troposphere for a short period (days-weeks), aerosol emissions from a particular source, (e.g. the U.S.A or Europe), could have large regional effects (e.g. on the Atlantic, over which much of the aerosol is transported).

Although previous climate models have been hindered in their simulation of aerosol effects due to a lack of fidelity, current state-of-the-art models are increasing the range of complex aerosol processes they simulate (especially the indirect aerosol effects). Interestingly, the latest Met Office Hadley Centre model, HadGEM2-ES,  captures the magnitude and the phase of the observed area-average NASST (see figure 2).  The authors argue that this particular model is able to capture the evolution due to the improved representation of the aerosol modulation of the cloud albedo, which in turn modulates the SST, corroborating the view that aerosols emission have impacted on multidecadal NASST.  On the basis of these results the authors concluded that aerosols were “the prime driver of 20th Century North Atlantic climate variability”.

Figure 2. Area-average North Atlantic SST (NASST) anomaly (75-7.5W, 0-60N). Black line: Observed (HadISST), Red Line: Ensemble mean of HadGEM2-ES historical simulations with all external forcing. Orange shading: 1 standard deviation of ensemble spread of HadGEM2-ES All forcings. All anomalies are relative to 1871-2000 mean.

Although aerosol forcing appears to control the mulitdecadal NASST variability in HadGEM2-ES, it isn’t so clear cut for the real world. In HadGEM2-ES multidecadal changes in NASST are apparently not linked to changes in the ocean circulation. However,  in the real world the 1990s shift to anomalously warm NASST was associated with substantial changes Atlantic circulation. In particular, a large weakening and shifting of the circulation in the subpolar north Atlantic (between 50N-65N), which has been linked to changes in the wider Atlantic circulation and heat transports. The HadGEM2-ES simulations also differ with observations in many other respects. In particular, HadGEM2-ES simulates zero net warming of the upper 700m of the Atlantic ocean since 1950, in contrast to the observed warming, and HadGEM2-ES also simulates the spatial pattern of NASST variability poorly. The latter point is especially clear for the 1960s (see figure 3), where the observed cooling is centred in the subpolar North Atlantic indicative of ocean heat transport changes, but the simulated cooling is almost global.  As these variables are influenced by the aerosol forcing in the model, such disagreement with observations casts some doubt on whether the aerosol forcing in HadGEM2-ES is realistic.

Figure 3. SST differences between the North Atlantic cold period (1961-1980) and the North Atlantic warm period (1941-1960). a) shows the observed difference and b) shows the difference simulated by HadGEM2-ES

So where are we? Because of the important impacts of  NASST on regional climates, especially regions sensitive rainfall (e.g. the Sahel), understanding and predicting the multidecadal changes in NASST remains a key challenge. Given the evidence it seems likely that anthropogenic aerosol have had an impact on past NASST variability, especially in the Tropics. However, given the disagreement of the simulated changes in HadGEM2-ES with those observed it is clear that further understanding is needed. For instance, what was the magnitude of the aerosol affects, and did the anthropogenic aerosol influence the phasing of the natural variability by forcing dynamical feedbacks, as has been suggested recently for Volcanic forcing.

Future model testing and improvement may shed light on what’s controlling  multidecadal NASST, but ultimately, we may have to wait for more, and improved, observations of future changes. Either way, it’s an interesting time to be thinking about Atlantic Multidecadal variability and its consequences.

Find out more

Booth, et al., 2012, Aerosols implicated as the prime driver of twentieth-century North Atlantic climate variability, Nature
Zhang, et al., 2013, Have aerosols caused the observed North Atlantic multidecadal variability? JAS, In Press

Leave a Reply

Your email address will not be published. Required fields are marked *