There were several centennial-scale fluctuations in the climate and oceanography of the North Atlantic region over the past 1,000 years, including a period of relative cooling from about AD 1450 to 1850 known as the Little Ice Age1. These variations may be linked to changes in solar irradiance, amplified through feedbacks including the Atlantic meridional overturning circulation2. Changes in the return limb of the Atlantic meridional overturning circulation are reflected in water properties at the base of the mixed layer south of Iceland. Here we reconstruct thermocline temperature and salinity in this region from AD 818 to 1780 using paired δ18O and Mg/Ca ratio measurements of foraminifer shells from a subdecadally resolved marine sediment core. The reconstructed centennial-scale variations in hydrography correlate with variability in total solar irradiance. We find a similar correlation in a simulation of climate over the past 1,000 years. We infer that the hydrographic changes probably reflect variability in the strength of the subpolar gyre associated with changes in atmospheric circulation. Specifically, in the simulation, low solar irradiance promotes the development of frequent and persistent atmospheric blocking events, in which a quasi-stationary high-pressure system in the eastern North Atlantic modifies the flow of the westerly winds. We conclude that this process could have contributed to the consecutive cold winters documented in Europe during the Little Ice Age.
The import of salt to higher latitudes by the North Atlantic Current (NAC) is essential for maintaining the high density of surface waters in the Nordic and Labrador seas3,4, a prerequisite for deepwater formation. The formation of deep waters is critical for the Atlantic meridional overturning circulation (AMOC) and therefore of great importance to the climate system. Furthermore, the heat released from the NAC, aided by the westerly winds, contributes to ameliorating the climate of Europe5. Because of its large heat capacity, the ocean is expected to be among the most predictable components of the climate system on multidecadal timescales. It is therefore of paramount importance to study past variability in the properties of the NAC beyond the instrumental record to better constrain natural ocean variability and its potential impacts on future regional and global climate.
To investigate multidecadal hydrographic variability of the NAC during the past millennium, we use marine sediment core RAPiD-17-5P (61° 28.90′ N, 19° 32.16′ W, 2,303 m water depth; Fig. 1) recovered from the Iceland Basin. The upper 600 m of the water column at the core site are dominated by the northward-flowing NAC (ref. 3). Temperature and salinity reconstructions were produced by analysing paired Mg/Ca–δ18O signals in the shells of the planktonic foraminifera Globorotalia inflata (Supplementary Methods). The concentration of Mg in calcite foraminiferal tests is an established proxy for temperature6, which combined with the δ18O composition of the same calcite, allows the isolation of the δ18O of sea water (δ18Osw) and the estimation of salinity. G. inflatalives close to the base of the seasonal thermocline7 and, owing to the limited seasonal variation at this depth, it principally records mean annual temperatures8. The chronology for RAPiD-17-5P was obtained using 12 accelerator mass spectrometry radiocarbon dates, which yielded a linear sedimentation rate of 0.16 cm yr−1, providing an integrated sample resolution of ∼6 years between AD 818 and 1780 years (Supplementary Methods).
Our results reveal abrupt multidecadal to centennial shifts in the temperature and salinity of the NAC waters of ∼3.5 ± 1.1 °C and ∼1.2 ± 0.8, respectively, during the past millennium (Fig. 2b, c). The magnitude of the hydrographic variability is substantial and comparable to that recorded in a lower resolution record spanning the present interglacial from a nearby site9, which highlights the similarities in the ocean variability on a diverse range of timescales. The timing of the hydrographic shifts shows a strong correlation with total solar irradiance (TSI) variability10 (Fig. 2d). Periods of solar minima (maxima) generally correspond to cold and fresh (warm and salty) conditions in the NAC (Fig. 2). A Pearson’s correlation coefficient of 0.51 (n = 77) with 95% confidence interval (0.31; 0.67) was estimated when correlating temperature and TSI records, following Gaussian interpolation to a common time-step of ∼12 years (the minimum resolution of the temperature record; (Supplementary Fig. 3).
Wavelet transform analysis of the temperature record shows a clear 200-year cycle with enhanced power between AD 1200 and 1650 (Supplementary Fig. 5). Furthermore, cross-spectral analysis shows that temperature and TSI are coherent above the 90% confidence level in the frequency range 177–227 years (Supplementary Fig. 4). This variance is similar to the de Vries solar activity cycle (∼210 years) and supports the correlation found between the NAC temperature and TSI records over the past millennium.
To investigate the feedback processes linking TSI variability and the recorded abrupt ocean changes we analysed climate model simulations carried out using the Community Climate System Model version 4.0 (CCSM4), forced with TSI variability and volcanic aerosols for the past millennium (AD 850–1850)11. The modelling results also present a strong positive correlation between temperature and salinity south of Iceland and solar irradiance (Fig. 3a, b), although the hydrographic variability in the model is of smaller amplitude than in the proxy data. The highest correlations are found in the pathway of the NAC and particularly in the path of its western branch, the Irminger Current. Additional temperature and salinity proxy reconstructions of the Irminger Current, from a sediment core south of Greenland (RAPiD-35-25B, Fig. 1), show broad similarities with the results from RAPiD-17-5P (Supplementary Figs 6–8), which confirm the westward propagation of the anomalies in the warm Atlantic waters by means of the Irminger Current found in CCSM4 (Fig. 3a, b).
The similar timing of volcanic eruptions and solar minima during the past millennium (Fig. 2a) makes the separation of their relative climatic influence difficult and has been the subject of much debate in recent literature. For instance, the injection of aerosols into the stratosphere by volcanic activity may have additionally contributed towards the cold fresh events recorded south of Iceland (Fig. 2a–c; for example, ref. 12). Here, decomposition of the relative contribution of the solar and volcanic forcing to the ocean changes was explored by carrying out a series of sensitivity tests in CCSM4. In these experiments we find that changes in volcanic forcing yield a qualitatively different dynamic response of the atmosphere–ocean system in our region of study compared with solar forcing, which consistently explain the key changes described in the transient simulation (Supplementary Figs 11–13). We therefore conclude that solar irradiance was the dominant forcing mechanism of the centennial scale ocean changes.
The NAC and its northwestern branch, the Irminger Current, constitute the main boundary currents of the subpolar gyre (SPG; Fig. 1). Changes in the strength of the SPG therefore influence the properties, structure and volume transport of the surface circulation in the North Atlantic13. Previous modelling and palaeodata studies have interpreted changes in the hydrographic properties of the NAC, and particularly salinity south of Iceland, to be controlled by frontal mixing resulting from changes in the spatial extent of the SPG as a response to changes in its strength9,13. For example, during a weak and contracted SPG circulation a displacement of the subpolar front to the west would increase the contribution of subtropical versus subpolar waters to the NAC, making it warm and salty. Here, however, volume transport analysis of the SPG in CCSM4 over the past millennium indicates that warmer and saltier conditions found south of Iceland and in the pathway of the Irminger Current correspond to periods of stronger SPG circulation (Fig. 3c). This is in agreement with recent observations that show advection may play a dominant role in determining the properties of water masses along the Irminger Current14 (Supplementary Discussion). An increase in the heat and particularly salt transport by the Irminger Current into the Labrador Sea may have additionally promoted deep convection in this region4, potentially impacting the AMOC.
As ocean gyres are largely driven by wind-stress forcing, changes in the SPG strength and NAC properties found in the proxy and model results are likely to be linked to shifts in atmospheric circulation. The North Atlantic Oscillation (NAO) is the dominant mode of atmospheric variability in the North Atlantic15. During a positive NAO state the increase in the strength of the westerlies promotes surface heat loss and ultimately leads to deeper convection in the Labrador Sea, baroclinically driving a stronger SPG. However, an emergent view derived from both model and observational data is that small-scale atmospheric patterns in the Northeast Atlantic, such as atmospheric blocking events as part of the east Atlantic pattern or polar mesoscale storms, may contribute considerably to driving North Atlantic surface circulation16,17,18.
Atmospheric blocking events are mid-latitude weather systems where a quasi-stationary high-pressure system located in the Northeast Atlantic modifies the flow of the westerly winds by blocking or diverting their pathway. Blocking events derive from instabilities of the jet stream and predominantly develop in winter, typically in association with a negative NAO (ref. 19). The impacts of the frequency and magnitude of these small-scale atmospheric systems are not restricted to the ocean16,17,18 but also have important effects on European temperatures, as they block the meridional transport of warm maritime winds (which are replaced by the cold northeasterlies). For example, Atlantic blocking events are thought to have been responsible for several recent cold European winters (for example, 1963, 2009, 2010 and 2013).
The analysis of sea level pressure (SLP) patterns in our CCSM4 simulation reveals the presence of an anomalous high-pressure system off west Europe during periods of solar minima (Fig. 4), which corresponds to a weaker SPG (Fig. 3c) and a colder and fresher NAC (Figs 2 and 3a, b). This finding is in line with recent studies that suggest a decrease in SPG strength with more frequent and stronger atmospheric blocking events on decadal timescales16,18. The results agree with the early concept that the severe winters experienced in Europe during the Maunder Minimum were caused by periods of increased atmospheric blocking1 and are also consistent with SLP field reconstructions that show a high-pressure system over northwest Europe towards the end of the Spörer and during the Maunder Minimum20. Similarly, a number of studies suggest a negative NAO state during the Maunder Minimum or other periods of low TSI (ref. 21), in agreement with increased blocking arising from the weaker westerly winds.
Growing evidence for the linkage between solar variability and frequency of blocking in the Northeast Atlantic has also been provided by meteorological studies. Modern observations show strong solar modulation of the blocking frequency and positioning during the 11-year solar cycles for the past 50 years, impacting substantially on UK winter temperatures22,23. Periods of solar minima, such as the Maunder Minimum, have also been shown to correspond to cold temperatures in the Central England Temperature record, which is dominated by the frequency of winter blockings24. The regional atmospheric response to solar forcing has often been explained through variability in stratospheric temperatures as the response of ozone formation to changes in ultraviolet radiation21,22,25. Changes in stratospheric temperatures have a top-down effect on tropospheric dynamics and hence induce variability of the jet stream22,26. Nonetheless, modelling studies with a simplified representation of the upper atmosphere, such as CCSM4, find a similar response to solar forcing suggesting that other feedbacks related to internal climate dynamics such as ocean feedbacks on the atmosphere and Pacific teleconnections may also be influential21. On decadal timescales, modelling and observational studies have previously identified separate relationships between solar irradiance and Atlantic blocking events22,23,26, and blocking events and SPG strength16,18 individually. Our findings support a direct link between these three components of the Earth’s climate system, which probably shaped the North Atlantic climate over the past millennium.
Climate variability on decadal timescales is largely believed to be dominated by internal processes rather than external forcing, which presents large difficulties for much-needed climate projections of the coming decades. However, the proxy evidence presented here, supported by model results, suggests that external forcing by solar variability has a considerable impact on multidecadal to centennial ocean–atmospheric dynamics, with important effects on regional climate such as European winters. In this context, predictions of a forthcoming prolonged period of low solar activity27 imply direct climatic consequences.
Despite the hemispheric temperature changes expected from solar minima being much smaller than the warming from future CO2 emissions, regional climate variability associated with solar-induced ocean–atmosphere feedbacks could be substantial and should be taken into consideration when projecting future climate changes.
Paired δ18O and Mg/Ca analyses were carried out on 6–20 G. inflata (300–355 μm) tests. Samples were prepared using the method outlined by ref. 28 and analysed using a Finnigan Element XR high-resolution inductively coupled plasma mass spectrometer (Cardiff University). Calculation of average shell weights and investigation of the covariability of Mg/Ca record to metals such as Fe, Mn and Al shows that no secondary effects such as partial dissolution or trace metal contamination have altered the primary temperature signal in the Mg/Ca record. Mg/Ca values were converted to calcification temperatures using Mg/Ca = 0.675 exp(0.1 × T) after the core-top calibration by ref. 9. Stable isotope measurements were carried out on a Thermo Finnigan MAT 252 isotope ratio mass spectrometer coupled to a Kiel II carbonate preparation device at Cardiff University. For more details see Supplementary Methods.
We are grateful to J. Becker and A. Morte-Ródenas for laboratory assistance and thank N. McCave and the crew of RV Charles Darwin 159 for their assistance with sample collection. We also thank C. C. Raible for discussions and C-F. Schleussner for early discussions on the subject. This work was financially supported by the UK National Environmental Research Council (NERC). Computing resources were provided by NCAR’s Computational and Information Systems Laboratory (CISL) sponsored by the National Science Foundation and other agencies. A.B. is supported by the European Commission under the Marie Curie Intra-European Fellowship ECLIPS (PIEF-GA-2011-300544) and the National Centre for Excellence in Research: Climate of the Swiss National Science Foundation. P.M.S. and I.R.H. also gratefully acknowledge the support of the Climate Change Consortium of Wales (www.c3wales.org).
About this article
Nature Communications (2018)