Adolphi et al. Solar Influence on Greenland Climate

1

Persistent link between solar activity and Greenland climate during the Last Glacial Maximum

Florian Adolphi1*, Raimund Muscheler1, Anders Svensson2, Ala Aldahan3,4, Göran Possnert5, Jürg Beer6, Jesper Sjolte1, Svante Björck1, Katja Matthes7 and Rémi Thiéblemont7

1Department of Geology—Quaternary Sciences, Lund University, 22362 Lund, Sweden, 2Centre for Ice and Climate, Niels Bohr Institute, University of Copenhagen, 2100 Copenhagen, Denmark, 3Department of Earth Sciences, Uppsala University, 75236 Uppsala, Sweden, 4Department of Geology, United Arab Emirates University, 17551 Al Ain, UAE, 5Tandem Laboratory, Uppsala University, 75120 Uppsala, Sweden, 6Swiss Federal Institute of Aquatic Science and Technology, Eawag, 8600 Dübendorf, Switzerland, 7Division of Ocean Circulation and Climate, GEOMAR Helmholtz Centre for Ocean Research Kiel, 24105 Kiel, Germany. *e-mail: Florian.Adolphi@geol.lu.se

Changes in solar activity have previously been proposed to cause decadal- to millennial-scale fluctuations in both the modern and Holocene climates1. Direct observational records of solar activity, such as sunspot numbers, exist for only the past few hundred years, so solar variability for earlier periods is typically reconstructed from measurements of cosmogenic radionuclides such as 10Be and 14C from ice cores and tree rings2,3. Here we present a high-resolution 10Be record from the ice core collected from central Greenland by the Greenland Ice Core Project (GRIP). The record spans from 22,500 to 10,000 years ago, and is based on new and compiled data4–6. Using 14C records7,8 to control for climate-related influences on 10Be deposition, we reconstruct centennial changes in solar activity. We find that during the Last Glacial Maximum, solar minima correlate with more negative δ18O values of ice and are accompanied by increased snow accumulation and sea-salt input over central Greenland. We suggest that solar minima could have induced changes in the stratosphere that favour the development of high-pressure blocking systems located to the south of Greenland, as has been found in observations and model simulations for recent climate9,10. We conclude that the mechanism behind solar forcing of regional climate change may have been similar under both modern and Last Glacial Maximum climate conditions.

The Sun is the main energy source for the Earth’s climate system. Satellite observations indicate variations in total solar irradiance (TSI) of about 1 W/m2 associated with the solar 11 yr cycle1.Despite these small changes in forcing there is compelling evidence for a solar influence on climate arising from palaeoclimate studies (see ref. 1 and references therein). One proposed mechanism to amplify the Sun’s influence on climate involves the relatively large modulation of the solar ultraviolet output, which alters the radiative balance in the stratosphere through ozone feedback processes and eventually propagates downwards causing changes in the tropospheric circulation1. Palaeoclimate studies allow an assessment of solar forcing of climate under various past orbital configurations and mean climate states, and thus may provide valuable insight into climate sensitivity to and mechanisms of solar forcing.

Before the satellite era and observations of sunspots, cosmogenic radionuclides, such as 10Be and 14C, provide the most reliable information about solar variability. Their atmospheric production rates depend on the flux of galactic cosmic rays impinging on the Earth’s atmosphere, which is in turn modulated by the variable shielding through the Earth’s and solar magnetic fields2, the latter

being correlated to TSI variations during the satellite era3. In addition to this production component, palaeo-records of 10Be (from, for example, ice cores) and 14C (from, for example, tree rings and speleothems) are affected by ‘system effects’ such as changes in transport and deposition (ref. 11 and references therein), and the carbon cycle4, respectively. As the expected system effects are fundamentally different for the two radionuclides, a combined analysis of 10Be and 14C records can help to isolate production rate variations more reliably. In summary, a reconstruction of past solar variability from cosmogenic radionuclides requires an assessment of system effects in 14C and/or 10Be records, and the elimination of production rate variations due to geomagnetic modulation. Further support for a solar origin of production rate variations can be drawn from identification of well-known long-term solar cycles, and comparison of the inferred amplitudes to expectations deduced from physically based models12. In the absence of suitable data this approach has so far been limited to the Holocene (for example, ref. 3). Nevertheless, the presence of the solar de Vries cycle (~207 yr) during parts of the last glacial has been demonstrated from 10Be alone13. Here we present the first reconstruction of solar activity variations for the end of the last glaciation from 22.5 to 10 kyr BP

Adolphi et al. Solar Influence on Greenland Climate

2

rate changes on ice-core 10Be concentrations6,16, especially visible at transitions between stadials and interstadials (Fig. 1). However, both records show similar high-frequency variations after normalization (Methods and Fig. 2). This shows that accumulation rate changes do not dominate the records on sub-millennial timescales when accumulation rate changes are small. Moreover, the normalized 10Be variations are largely independent of other atmospheric aerosol species measured in the GISP2 ice core17 indicating minor climate-related depositional influences16 on 10Be (Supplementary Fig. 1). The normalization also removes unresolved differences in the millennial variations of the GISP2 and GRIP 10Be series (Supplementary Fig. 2). Most importantly, the resulting 10Be record is consistent with the tree-ring and speleothem 14C production rates even over stadial –interstadial transitions where system effects are expected to be largest (Fig. 2 and Supplementary Figs 3 and 4). The 14C production rates were derived from the 14C records (Fig. 1) using a carbon-cycle box-diffusion model18 that corrects for known carbon cycle effects on the atmospheric 14C content (Supplementary Methods and Supplementary Fig. 6).

(thousand years before present, AD 1950) based on new and published 10Be data from the GRIP and GISP2 ice cores4-6 supported by independent estimates of atmospheric 14C concentrations7,8. In addition, we provide the first evidence for a solar forcing of Greenland climate during Greenland Stadial 2 (GS-2, 22.9 – 14.7 kyr BP; ref. 14) that seems coherent with increased frequencies of high-pressure blocking patterns south of Greenland during low-solar-activity winters – a relationship that has been reported previously from modern observations and climate model experiments9,15.

The new high-resolution GRIP 10Be record (10.8 – 18.6 kyr BP, see Supplementary Methods) is shown in Fig. 1. In combination with previously published GRIP/GISP2 10Be data4-6 the resulting record covers the investigated period with an average resolution of about 20 years. In the following we will address the above-mentioned points to evaluate the GRIP/GISP 10Be record as a proxy record for solar variability.

The difference of 10Be concentrations and fluxes reflects the known effects of snow accumulation

0

20

40

60

80

1001 1.5 2

x 104

Holocene GS−1 GI−1 GS−2

a

δ 18O (‰

)

−44

−42

−40

−38

−36

−34

b

10Be

con

cent

ratio

n

(104 a

tom

s * g

ice−1

)

1

2

3

4

5

6

c

10Be flux(10

6 atoms * cm

−2 * yr −1)0

0.25

0.5

0.75

1

d

calendar years BP

∆14

C (‰

)

10,000 12,000 14,000 16,000 18,000 20,000 22,000100

200

300

400

500

600

Figure 1. Key data used in this study. a, δ18O variations as recorded in the GRIP ice core21. b, 10Be concentrations from the GRIP (red: this study, black: refs 4,5) and GISP2 (ref. 6; blue) ice cores. c, 10Be fluxes using accumulation rates inferred from the GICC05 age scale (ref. 28 and references therein) and ice-flow modelling29 (line colouring as in b). d, 14C (that is, 14C concentration after correction for fractionation and decay, relative to a standard) from the tree rings7 (pink) and Hulu Cave speleothem H82 (ref. 8; black). Black dots indicate single measurements and grey shading shows the ±1 σ envelope (Supplementary Methods). Top bar, INTIMATE event stratigraphy14. GS, Greenland Stadial; GI, Greenland Interstadial.

Adolphi et al. Solar Influence on Greenland Climate

3

The agreement of 10Be and 14C records strongly supports our interpretation of the 10Be record being production dominated. The coherence of 10Be and H82 14C slightly decreases back in time, which is probably due to the variable sampling resolution of the H82 speleothem and timescale differences. Nevertheless, for most of the time the normalized 10Be and 14C records are consistent within errors and indicate similar spectral properties (Supplementary Fig. 4) and amplitudes (Fig. 2 c,d). It should be noted that the timescales have not been adjusted, which would increase consistency between the records but prevent an independent comparison.

The quantification of geomagnetic modulation is an additional uncertainty of solar activity reconstructions from cosmogenic radionuclides19. However, it has been shown that detectable geomagnetic influences on Holocene cosmogenic radionuclide production rates are limited to timescales of several centuries to millennia19. In addition, the relative variations of solar-induced production rate changes are independent of the geomagnetic field intensity except for very low field

strengths12,20. Hence, the applied normalization of the 10Be and 14C production rates minimizes the influence of the geomagnetic field on our solar activity reconstruction focused on centennial variations. We note that applying this normalization to a stack of Holocene cosmogenic radionuclide records3 leads to a linear scaling to the correspondingly band-pass-filtered TSI reconstruction3, where geomagnetic field reconstructions have been considered explicitly (Supplementary Fig. 7). This does not preclude a remaining geomagnetic field influence in the 10Be data, but the absence of high-quality, high-resolution global geomagnetic intensity data inhibits a more detailed assessment.

In further support for the reliability of our solar activity reconstruction we see a coherent amplitude modulation of the well-known solar de Vries cycle (~207 yr) in both 10Be and 14C production rates (Fig. 2d), closely resembling the Holocene modulation pattern (Supplementary Fig. 8). Moreover, the relative amplitude of the 10Be variations is within the expected ranges induced by solar activity variations as inferred from physics based production rate

a 10Be

(nor

mal

ized

)

0.80.91.01.11.2

b

10Be

| 14

Cpr

od(n

orm

aliz

ed)

0.8

0.9

1.0

1.1

1.2

c

10Be

| 14

Cpr

od(n

orm

aliz

ed)

0.70.80.91.01.11.21.3

d

10Be

| 14

Cpr

od(n

orm

aliz

ed)

0.9

1.0

1.1

e

λ N

yqui

st(y

ears

)

calendar years BP10,000 12,000 14,000 16,000 18,000 20,000 22,000

150

500

Holocene GS−1 GI−1 GS−2

Figure 2. Comparison of normalized 10Be and 14C production rate changes. a, 10Be concentration and flux (thin green and bold orange lines, respectively, and expected amplitude for solar modulation variations corresponding to differences between the Maunder Minimum and the Modern Solar Maximum3,20 (grey band). b, 10Be flux (orange line) and 14C production rate changes modelled from tree rings (blue; ref. 7) ±1 σ uncertainty (blue shading) without timescale adjustments30 (Supplementary Fig. 3). c, 10Be flux (orange) and 14C production rate modelled from the H82 speleothem (black; ref. 8) ±1 σ errors (grey shading, Supplementary Methods). d, The solar de Vries cycle (180–230 yr) of 14C production rate (H82 speleothem, black) and 10Be flux (orange). e, Shortest resolvable wavelength by the 10Be records (orange) and H82 speleothem (black, ±1 σ uncertainty in grey). Horizontal dashed–dotted lines indicate the bandwidth of the normalization applied in a–c (Methods).

Adolphi et al. Solar Influence on Greenland Climate

4

δ18O from the GRIP (ref. 21) and GISP2 (ref. 22) ice cores reveals a significant positive correlation (r2 = 0.3 and 0.2, p < 0.01, for 10Be concentrations and flux, respectively) during GS-2 (Fig. 3 and Supplementary Fig. 9). Significant (95%) spectral coherence of δ18O and the solar activity proxy 10Be at known solar cycle wavelengths (Supplementary Fig. 5) strengthens the hypothesis of a solar influence on climate. This sun – climate relationship is accompanied by increased inputs of sea salt, higher snow accumulation, and a decrease in terrestrial aerosols (Supplementary Fig. 10). This pattern is interpreted as episodes of a more meridional atmospheric circulation during solar minima advecting relatively moist North Atlantic air masses to Greenland. Modern observations indicate that this type of flow pattern is enhanced during winters with high-pressure blocking situations south of Greenland, which in turn have been found to occur more often during solar minimum periods9. Recently, this mechanism was also shown to be present on centennial timescales10. Supporting this, we find increased meridional wind speeds south of Greenland accompanied by increased precipitation over the ice sheet during solar minima winters in the twentieth-century reanalysis23 and high-top chemistry–climate model experiments (Fig. 4 and Supplementary Methods). On synoptic scales these high-pressure blocking situations can be described as cyclonic Rossby wave breaking events over the North Atlantic, often accompanied by a southward

calculations12,20 (Fig. 2a, grey band). In conclusion, all evidence suggests that relative centennial variations in the 10Be record are largely free of detectable system effects and dominated by variations of solar activity. Therefore, this record allows for an investigation of sub-millennial solar forcing of climate.

Comparing the solar activity reconstruction to

δ18O

(‰

)

−44

−42

−40

−38

−36

δ18O

ano

mal

y (‰

)

−1−0.5

00.5

11.5

calendar years BP

10Be

(nor

mal

ized

)

15,000 16,000 17,000 18,000 19,000 20,000 21,000 22,000

0.60.70.80.911.11.21.3

Figure 3. Sun–climate linkages during GS-2. Top panel: GRIP (ref. 21; purple) and GISP2 (ref. 22; green) δ18O during GS-2. Dark lines are low-pass-filtered (<1/150 yr). Bottom panel: Sub-millennial (150–500 yr) GRIP (purple) and GISP2 (green) δ18O anomalies, their mean (dark blue, thick), and 10Be-based solar activity variations (orange). The 10Be axis is reversed. Note the high coherence of solar activity and Greenland climate during GS-2, which is robust also for the individual δ18O records (Supplementary Fig. 9). For GS-1 and GI-1 the results are less robust.

Figure 4. Solar forcing response in twentieth-century reanalysis23 and a coupled chemistry–climate model. Sea-level pressure (left), 850 hPa wind speed and direction (centre) and precipitation (right) anomalies for solar minimum–maximum winters (December–February) as seen in twentieth-century reanalysis23 (1948–2010, top) and a 145 yr coupled chemistry–climate model simulation (bottom, Supplementary Methods). The yellow dots indicate the GRIP ice-core location. The data have been divided into solar min/max periods following refs 9,15. Significance levels are indicated by black (90%) and white (95%) stippling (Supplementary Methods).

Adolphi et al. Solar Influence on Greenland Climate

5

MethodsNormalization of production rates. Following ref. 20 we normalize the 14C production rates and 10Be concentrations and fluxes by dividing each record by its low-pass-filtered copy (PLP500, cutoff 1/500yr-1). Before this, each record is low-pass-filtered (PLP150, cutoff 1/150 yr-1) to reduce noise and increase comparability between the 14C and 10Be records arising from their different and irregular sampling resolution. This normalization is summarized in equation (1):

Pnormalized = PLP150/PLP500 (1)

where P is the production rate (that is, 10Be concentrations or fluxes, or 14C production rates).

References1. Gray, L. J. et al. Solar influences on climate. Rev. Geophys.

48, RG4001 (2010).2. Lal, D. & Peters, B. in Handbuch der Physik. Band XLVI/2

(eds Fluegge, S. & Sitte, K.) 551_612 (Springer, 1968).3. Steinhilber, F. et al. 9,400 years of cosmic radiation and solar

activity from ice cores and tree rings. Proc. Natl Acad. Sci. USA 109, 5967_5971 (2012).

4. Muscheler, R. et al. Changes in the carbon cycle during the last deglaciation as indicated by the comparison of Be-10 and C-14 records. Earth Planet. Sci. Lett. 219, 325_340 (2004).

5. Yiou, F. et al. Beryllium 10 in the Greenland Ice Core Project ice core at Summit, Greenland. J. Geophys. Res. 102, 26783_26794 (1997).

6. Finkel, R. C. & Nishiizumi, K. Beryllium 10 concentrations in the Greenland Ice Sheet Project 2 ice core from 3_40 ka. J. Geophys. Res. 102, 26699_26706 (1997).

7. Reimer, P. J. et al. IntCal 13 and Marine 13 radiocarbon age calibration curves 0_50,000 years cal BP. Radiocarbon 55, 1869_1887 (2013).

8. Southon, J., Noronha, A. L., Cheng, H., Edwards, R. L. &Wang, Y. A high-resolution record of atmospheric 14C based on Hulu Cave speleothem H82. Quat. Sci. Rev. 33, 32_41 (2012).

9. Woollings, T., Lockwood, M., Masato, G., Bell, C. & Gray, L. Enhanced signature of solar variability in Eurasian winter climate. Geophys. Res. Lett. 37, L20805 (2010).

10. Mo_a-Sanchez, P., Born, A., Hall, I. R., Thornalley, D. J. R. & Barker, S. Solar forcing of North Atlantic surface temperature and salinity over the past millennium. Nature Geosci. 7, 275_278 (2014).

11. Heikkilä, U. & Smith, A. M. Production rate and climate influences on the variability of 10Be deposition simulated by ECHAM5-HAM: Globally, in Greenland, and in Antarctica. J. Geophys. Res. 118, 2506_2520 (2013).

12. Masarik, J. & Beer, J. An updated simulation of particle fluxes and cosmogenic nuclide production in the Earth’s atmosphere. J. Geophys. Res. 114, D11103 (2009).

13. Wagner, G. et al. Presence of the solar de Vries cycle (_205 years) during the last ice age. Geophys. Res. Lett. 28, 303_306 (2001).

14. Blockley, S. P. E. et al. Synchronisation of palaeoenvironmental records over the last 60,000 years, and an extended INTIMATE event stratigraphy to 48,000 b2k. Quat. Sci. Rev. 36, 2_10 (2012).

15. Ineson, S. et al. Solar forcing of winter climate variability in the Northern Hemisphere. Nature Geosci. 4, 753_757 (2011).

16. Alley, R. B. et al. Changes in continental and sea-salt atmospheric loadings in central Greenland during the most

displacement of the eddy-driven jet and negative North Atlantic Oscillation anomalies24,25. Both are connected to solar variability in reanalysis and model experiments (Fig. 4). At present, there are no high-top chemistry–climate model experiments under glacial boundary conditions to test whether this mechanism applies during the glacial. However, a multi-model study indicates that the presence of the Laurentide ice sheet leads to favourable conditions for cyclonic wave breaking during the Last Glacial Maximum (LGM) compared with today24, and hence may be indicative of more frequent high-pressure blocking. In addition, despite an altered atmospheric circulation during the LGM the weather patterns that led to precipitation over the ice sheet were probably comparable to present-day conditions26. Hence, increased winter precipitation over the Greenland ice sheet through enhanced meridional moisture transport would result in a net depletion of the ice-core δ18O signal, which is otherwise dominated by summer precipitation during the LGM (ref. 27). An increased winter–summer temperature difference during the LGM (ref. 27) would amplify this effect. Hence, we reason that the increased winter precipitation during periods of low solar activity could explain the positive correlation between our solar activity reconstruction and GRIP/GISP2 δ18O. Oceanic feedback to changed wind patterns may have acted as an additional amplification mechanism10. This would suggest that a top-down solar influence on high-pressure blocking frequency and thus, Greenland climate, as seen today9, may have been active 20,000 years ago under a very different climate regime. In addition, this provides a testable hypothesis for an orbital alteration of sun–climate linkages because the mean latitudinal position and strength of the eddy driven jet is, among other factors, related to orbital forcing. This may alter the baseline for the likelihood of high-latitude blocking24 and its potential alteration through solar activity changes.

Adolphi et al. Solar Influence on Greenland Climate

6

been performed within the Helmholtz-University Young Investigators Group NATHAN of K.M. financially supported by the Helmholtz-Association through the President’s Initiative and Networking Funds, the GFZ Potsdam, Freie Universität Berlin and since 2012 by GEOMAR Helmholtz Centre for Ocean Research Kiel. We thank C. Petrick for conducting the model calculations at the Deutsche Klimarechenzentrum (DKRZ) Hamburg. The climate model analysis is part of the WCRP SPARC-SOLARIS/HEPPA project (http://solarisheppa.geomar.de/solarisheppa/) and the EU-Cost Action ES1005 `TOSCA’ (www.cost-tosca.eu).

recent deglaciation: Model-based estimates. J. Glaciol. 41, 503_514 (1995).

17. Mayewski, P. A. et al. Major features and forcing of high-latitude northern hemisphere atmospheric circulation using a 110,000-year-long glaciochemical series. J. Geophys. Res. 102, 26345_26366 (1997).

18. Siegenthaler, U. Uptake of excess CO2 by an outcrop-di_usion model of the ocean. J. Geophys. Res. 88, 3599_3608 (1983).

19. Snowball, I. & Muscheler, R. Palaeomagnetic intensity data: An Achilles heel of solar activity reconstructions. The Holocene 17, 851_859 (2007).

20. Muscheler, R. & Heikkilä, U. Constraints on long-term changes in solar activity from the range of variability of cosmogenic radionuclide records. Astrophys. Space Sci. Trans. 7, 355_364 (2011).

21. Johnsen, S. J. et al. The _18O record along the Greenland Ice Core Project deep ice core and the problem of possible Eemian climatic instability. J. Geophys. Res. 102, 26397_26410 (1997).

22. Stuiver, M., Braziunas, T. F., Grootes, P. M. & Zielinski, G. A. Is there evidence for solar forcing of climate in the GISP2 oxygen isotope record? Quat. Res. 48, 259_266 (1997).

23. Compo, G. P. et al. The twentieth century reanalysis project. Q. J. R. Meteorol. Soc. 137, 1_28 (2011).

24. Rivière, G., Laîné, A., Lapeyre, G., Salas-Mélia, D. & Kageyama, M. Links between Rossby wave breaking and the North Atlantic oscillation_Arctic oscillation in present-day and last glacial maximum climate simulations. J. Clim. 23, 2987_3008 (2010).

25. Woollings, T., Hoskins, B., Blackburn, M. & Berrisford, P. A new Rossby wave-breaking interpretation of the north Atlantic oscillation. J. Atmos. Sci. 65, 609_626 (2008).

26. Merz, N. et al. Greenland accumulation and its connection to the large-scale atmospheric circulation in ERA-Interim and paleoclimate simulations. Clim. Past 9, 2433_2450 (2013).

27. Werner, M., Mikolajewicz, U., Heimann, M. & Ho_mann, G. Borehole versus isotope temperatures on Greenland: Seasonality does matter. Geophys. Res. Lett. 27, 723_726 (2000).

28. Svensson, A. et al. A 60000 year Greenland stratigraphic ice core chronology. Clim. Past 4, 47_57 (2008).

29. Johnsen, S. J. et al. The Eem Stable Isotope Record along the GRIP Ice Core and its interpretation. Quat. Res. 43, 117_124 (1995).

30. Muscheler, R. et al. Tree rings and ice cores reveal 14C calibration uncertainties during the Younger Dryas. Nature Geosci. 1, 263_267 (2008).

AcknowledgementsWe dedicate this paper to S. Johnsen, who unfortunately left us last year. He was a great colleague and scientist who was always supportive of 10Be measurements in ice cores. A-M. Berggren and A. Sturevik-Storm are thanked for their help in the Uppsala laboratory. We appreciate comments by one of the original PIs of the GRIP 10Be project, G. Raisbeck. We acknowledge P. Kubik and M. Christl for performing 10Be measurements at ETH Zurich. The study was supported by the Swedish Research Council (VR) through a Linnaeus grant to Lund University (LUCCI) and the Crafoord Foundation. R.M. was supported by the Royal Swedish Academy of Sciences through a grant financed by the Knut and Alice Wallenberg Foundation and VR (Dnr: 2013-8421). The chemistry-climate model experiments have