Timing and climate forcing of volcanic eruptions for the past 2,500 years

Abstract

Volcanic eruptions contribute to climate variability, but quantifying these contributions has been limited by inconsistencies in the timing of atmospheric volcanic aerosol loading determined from ice cores and subsequent cooling from climate proxies such as tree rings. Here we resolve these inconsistencies and show that large eruptions in the tropics and high latitudes were primary drivers of interannual-to-decadal temperature variability in the Northern Hemisphere during the past 2,500 years. Our results are based on new records of atmospheric aerosol loading developed from high-resolution, multi-parameter measurements from an array of Greenland and Antarctic ice cores as well as distinctive age markers to constrain chronologies. Overall, cooling was proportional to the magnitude of volcanic forcing and persisted for up to ten years after some of the largest eruptive episodes. Our revised timescale more firmly implicates volcanic eruptions as catalysts in the major sixth-century pandemics, famines, and socioeconomic disruptions in Eurasia and Mesoamerica while allowing multi-millennium quantification of climate response to volcanic forcing.

Full text

ARTICLE
doi:10.1038/nature14565
Timing and c limate forcing of volcanic
eruptions for the past 2,500 years
M. Sigl
1
{, M. Winstrup
2
, J. R. McConnell
1
, K. C. Welten
3
, G. Plunkett
4
, F. Ludlow
5
,U.Bu
¨
ntgen
6,7,8
, M. Caffee
9,10
, N. Chellman
1
,
D. Dahl-Jensen
11
, H. Fischer
7,12
, S. Kipfstuhl
13
, C. Kostick
14
, O. J. Maselli
1
, F. Mekhaldi
15
, R. Mulvaney
16
, R. Muscheler
15
,
D. R. Pasteris
1
, J. R. Pilcher
4
, M. Salzer
17
,S.Schu
¨
pbach
7,12
, J. P. Steffensen
11
, B. M. Vinther
11
& T. E. Woodruff
9
Volcanic eruptions contribute to climate variability, but quantifying these contributions has been limited by inconsis-
tencies in the timing of atmospheric volcanic aerosol loading determined from ice cores and subsequent cooling from
climate proxies such as tree rings. Here we resolve these inconsistencies and show that large eruptions in the tropics and
high latitudes were primary drivers of interannual-to-decadal temperature variability in the Northern Hemisphere
during the past 2,500 years. Our results are based on new records of atmospheric aerosol loading developed from
high-resolution, multi-parameter measurements from an array of Greenland and Antarctic ice cores as well as
distinctive age markers to constrain chronologies. Overall, cooling was proportional to the magnitude of volcanic
forcing and persisted for up to ten years after some of the largest eruptive episodes. Our revised timescale more
firmly implicates volcanic eruptions as catalysts in the major sixth-century pandemics, famines, and socioeconomic
disruptions in Eurasia and Mesoamerica while allowing multi-millennium quantification of climate response to volcanic
forcing.
Volcanic eruptions are primary drivers of natural climate variabil-
ity—their sulfate aerosol injections into the stratosphere shield the
Earth’s surface from incoming solar radiation, leading to short-term
cooling at regional-to-global scales
1
. Temperatures during the past
2,000 years have been reconstructed at regional
2
, continental
3
, and
global scales
4
using proxy information from natural archives. Tree-
ring-based proxies provide the vast majority of climate records from
mid- to high-latitude regions of (predominantly) the Northern
Hemisphere, whereas ice-core records (for example, d
18
O) represent
both polar regions
3
.
Climate forcing reconstructions for the Common Era (
CE)—
including solar (for example,
10
Be)
5
and volcanic (for example, sulfate)
6,7
activity—derive mostly from ice-core proxies. Any attempt to attrib-
ute reconstructed climate variability to external volcanic forcing, and
to distinguish between response, feedback, and internal variability of
the climate system, requires ice-core chronologies that are synchron-
ous with those of other climate reconstructions. In addition, multi-
proxy climate reconstructions
2–4
derived from ice cores and other
proxies such as tree rings will have diminished high- to mid-frequency
amplitudes if the individual climate records are on different time-
scales.
The magnitudes and relative timing of simulated Northern
Hemisphere temperature responses to large volcanic eruptions are
in disagreement with reconstructed temperatures obtained from tree
rings
8,9
, but it is unclear to what extent this model/data mismatch
is caused by limitations in temperature reconstructions, volcanic
reconstructions, or implementation of aerosol forcing in climate
models
9–11
. The hypothesis of chronological errors in tree-ring-based
temperature reconstructions
8,9
offered to explain this model/data mis-
match has been tested and widely rejected
11–14
, while new ice-core
records have become available providing different eruption ages
15,16
and more precise estimates of atmospheric aerosol mass loading
17
than for previous volcanic reconstructions.
Using documented
18
and previous ice-core-based eruption ages
16
,
strong summer cooling following large volcanic eruptions has been
recorded in tree-ring-based temperature reconstructions during the
second millennium
CE with a one-to-two year lag similar to that
observed in instrumental records after the 1991 Pinatubo eruption
19
.
An apparent seven-year delayed cooling observed in individual tree-
ring series relative to Greenland ice-core acidity peaks during the first
millennium
CE, however, suggests a bias in existing ice-core chronolo-
gies
20,21
. Using published ice-core chronologies, we also observed a
seven-year offset between sulfate deposition in North Greenland and
the start of tree-ring growth reduction in a composite of five multi-
centennial tree-ring summer temperature reconstructions (‘N-Tree’)
from the Northern Hemisphere between 1 and 1000
CE (Methods),
whereas tree-ring response was effectively immediate for eruptions
occurring after 1250
CE (Fig. 1a).
Precise time marker across hemispheres
Independent age markers with which to test the accuracy of tree-ring
and ice-core chronologies have recently become available with the
detection of abrupt enrichment events in the
14
C content of tree rings.
Rapid increases of atmospheric
14
C were first identified in individual
growth increments of cedars from Japan between 774
CE and 775 CE
22
and between 993 CE and 994 CE
23
. The presence and timing of
1
Desert Research Institute, Nevada System of Higher Education, Reno, Nevada 89512, USA.
2
Department of Earth and Space Sciences, University of Washington, Seattle, Washington 98195, USA.
3
Space
Sciences Laboratory, University of California, Berkeley, California 94720, USA.
4
School of Geography, Archaeology and Palaeoecology, Queen’s University Belfast, Belfast BT7 1NN, UK.
5
Yale Climate and
Energy Institute, and Department of History, Yale University, New Haven, Connecticut 06511, USA.
6
Swiss Federal Research Institute WSL, 8903 Birmensdorf, Switzerland.
7
Oeschger Centre for
Climate Change Research, University of Bern, 3012 Bern, Switzerland.
8
Global Change Research Centre AS CR, 60300 Brno, Czech Republic.
9
Department of Physics, Purdue University, West Lafayette,
Indiana 47907, USA.
10
Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana 47907, USA.
11
Centre for Ice and Climate, Niels Bohr Institute, University
of Copenhagen, 2100 Copenhagen, Denmark.
12
Climate and Environmental Physics, University of Bern, 3012 Bern, Switzerland.
13
Alfred-Wegener-Institut Helmholtz-Zentrum fu
¨
r Polar- und
Meeresforschung, 27570 Bremerhaven, Germany.
14
Department of History, The University of Nottingham, Nottingham NG7 2RD, UK.
15
Department of Geology, Quaternary Sciences, Lund University,
22362 Lund, Sweden.
16
British Antarctic Survey, Natural Environment Research Council, Cambridge CB3 0ET, UK.
17
The Laboratory of Tree-Ring Research, University of Arizona, Tucson, Arizona 85721,
USA. {Present address: Laboratory of Radiochemistry and Environmental Chemistry, Paul Scherrer Institut, 5232 Villigen, Switzerland.
00 MONTH 2015 | VOL 000 | NATURE | 1
G
2015 Macmillan Publishers Limited. All rights reserved

the event in 775 CE (henceforth, the 775 event) has been reproduced
by all radiocarbon measurements performed on tree rings at annual
(or higher) resolution—including tree cores from Germany
24
, the
Alps
12
, the Great Basin
25
(USA), and Siberia
25
. Recent identification
of the same 775
CE event in kauri wood samples from New Zealand in
the Southern Hemisphere demonstrates the global extent of the rapid
14
C increase and provides further constraints on the event’s exact
timing (March 775 6 6 months) owing to the asynchronous
Southern Hemisphere growing season
26
. While the cause of the 775
and 994 events is still debated
22,24,27
, we expect that they might also
have produced an excess of cosmogenic
10
Be through the interaction
of high-energy particles with atmospheric constituents
28,29
. Since both
of these radionuclides are incorporated rapidly into proxy archives via
aerosol deposition (
10
Be in ice cores) and photosynthesis (
14
CO
2
in
tree rings), isotopic anomalies caused by these extraterrestrial events
provide a global age marker with which to link ice-core records to
tree-ring chronologies directly
27
. The latter serve as an absolute and
precise age marker, verified (at least since 775
CE) by the coherence
of the rapid increase in
14
C in all tree-ring records for which high-
resolution radiocarbon analyses were performed, including those
speculated to be at risk of missing rings
8
.
We measured
10
Be concentrations at approximately annual reso-
lution in four ice cores—NEEM-2011-S1, TUNU2013, and NGRIP
in Greenland, and the West Antarctic Ice Sheet Divide Core
(WDC)—over depth ranges encompassing the year 775
CE as dated
in existi ng ice- core chronologies in order to provide a direc t, phys-
ically based test of any dating bias in these chronologies (Fig. 1,
Extended Data Fig. 1, Methods, Supplementary Data 1). Both polar
ice sheets contain
10
Be concentrations exceeding the natural back-
ground concentration (.150%; 6s) for one-to-two consecutive
years, compatible with the 775
CE event observed in tree rings.
Using the original ice-core age models
16,30
, the ages of the
10
Be max-
ima in NEEM-2011-S1, NGRIP, and WDC are 768
CE, offset by 7
years from the tree-ring event. A further
10
Be anomaly measured in
NEEM-2011-S1 at 987
CE, com patible with the 994 CE event in tree
rings, suggests that a chronological offset was present by the end of
the first millennium
CE (Fig. 1). Several different causes may have
contributed to the offset (see a summary in the Methods section),
among which is the use of a previous dating constraint
30
for
Greenland, where volcanic fallout in the ice was believed to indicate
the his toric (79
CE) eruption of Vesuvius.
Revised ice-core chronologies
Given the detection of a bias in existing ice-core chronologies, we
developed new timescales before the 1257 Samalas eruption in
Indonesia
31
using highly resolved, multi-parameter aerosol concen-
tration records from three ice cores: NEEM-2011-S1, NEEM, and
WDC. We used the StratiCounter program, an automated, objective,
annual-layer detection method based on Hidden Markov Model
(HMM) algorithms
32
(Methods). For NEEM-2011-S1, the confidence
intervals obtained for the layer counts allowed us to improve the
dating further by constraining the timescale using the 775
CE
10
Be
anomaly and three precisely dated observations of post-volcanic aero-
sol loading of the atmosphere (Fig. 2, Extended Data Tables 1–3,
Methods, Supplementary Data S2).
We evaluated the accuracy of our new chronologies (‘WD2014’
for WDC and ‘NS1-2011’ for NEEM) by comparison to (1) an
extensive database of historical volcanic dust veil observations
(Extended Data Fig. 2, Methods, Supplementary Data 2), (2) ice-core
tephra evidence (Methods), and (3) the 994
CE event (Methods, Fig. 2).
Using the new timescales, we found large sulfate signals in Greenland
(for example, in 682
CE,574CE, and 540 CE)between500CE and 2000 CE
that frequently occurred within one year of comparable—and indepen-
dently dated—signals in Antarctica. These bipolar signals can now be
confidently attributed to large tropical eruptions (Fig. 2). Back to 400
BCE, other large sulfate peaks (for example, 44 BCE)weresynchronousto
within three years (Fig. 2). We conclude that the revised ice-core time-
scales are accurate to within less than five years during the past 2,500
years, on the basis of combined evidence from radionuclides, tree rings,
tephra analyses, and historical accounts. Compared to the previous
chronologies, age models differ before 1250
CE by up to 11 years
(GICC05, Greenland) and 14 years (WDC06A-7, Antarctica)
(Extended Data Fig. 3).
1010
980
990 1000
990 1000980
+7 yr
Cedar (Japan)
c
994 CE event
–30
–20
–10
0
5
15
25
35
Year of original ice core dating (CE)
Year
(
CE
)
–5 0 5 10
Years relative to sulfate deposition
–3
–2
–1
0
1
Tree-ring growth anomaly (z-scores)
78–1000 CE
7 largest eruptions
1250–2000
CE
10 largest eruptions
+7 yr
a
NEEM
2011-S1
(Greenland)
760 770 780 790
Year
(
CE
)
WDC
(Antarctica)
TUNU2013
(Greenland)
NEEM
2011-S1
(Greenland)
Cedar (Japan)
Oak (Germany)
+7 yr
760 770 780 790
b
775 CE event
+7 yr
+7 yr
–30
–20
–10
0
5
15
25
35
5
15
25
35
5
15
25
35
Year of original ice core dating (
CE)
NGRIP
(Greenland)
10
Be (10
3
atoms g
–1
)
10
Be (10
3
atoms g
–1
)
10
Be (10
3
atoms g
–1
)
10
Be (10
3
atoms g
–1
)
Δ
14
C (‰)
Δ
14
C (‰)
Figure 1
|
Annual
10
Be ice-core records and post-volcanic cooling from
tree rings under existing ice-core chronologies. a, Superposed epoch analysis
for the largest volcanic signals in NEEM-2011-S1 between 78 and 1000
CE
(n 5 7; orange trace) and for the largest eruptions between 1250 and 2000 CE
(n 5 10; grey trace)
16
. Shown are standardized growth anomalies (z scores
relative to 1000–1099
CE) from a multi-centennial, temperature-sensitive
tree-ring composite (N-Tree
42,43,76–78
, Methods) ten years after the year of
volcanic sulfate deposition at the NEEM ice core site in Greenland (GICC05
timescale), relative to the level five years before sulfate deposition. b, Annually
resolved
10
Be concentration records from the WDC, TUNU2013, NGRIP, and
NEEM-2011-S1 ice cores on their original timescales and annually resolved
D
14
C series from tree-ring records between 755 CE and 795 CE
22,24
, with green
arrows representing the suggested time shifts for synchronization; error bars
are 1s measurement uncertainties; the estimated relative age uncertainty
for TUNU2013 at this depth interval from volcanic synchronization with
NEEM-2011-S1 is 61 year. c, Annually resolved
10
Be concentration record
from NEEM-2011-S1 ice core on its original timescale and annually resolved
D
14
C series from tree rings in 980 CE and 1010 CE
23
; error bars are 1s
measurement uncertainties.
2 | NATURE | VOL 000 | 00 MONTH 2015
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

–500 –300 –100 100 300 500 700 900 1100 1300
Year (BCE or CE)
0
40
80
nssS (p.p.b.)
–4
–2
0
2
Tree growth anomaly
(z-scores with respect to 1000–1099)
0
40
80
120
160
nssS (p.p.b.)
0
40
80
0
80
160
nssS (p.p.b.)
0
40
–426
NEEM
NEEM-2011-S1
N-Tree
Tianchi tephra (946/47)
–246
–168
–44
263
88
168
432
536
540
574
626
682
939
775
993
*
*
*
*
–47
169
434
266
540
575
682
775
–248
–172
–430
–425
–43
627
574
536
540–547
940
a
1108
1258
*
1171
1109
1231
*
1258
1230
1171
1109
–146
–150
–5
0
5
10
Years relative to ice core si
g
nal
–1.5
–1.0
–0.5
0
Tree growth anomaly
(z-scores)
–500–1250 CE
1250–2000 CE
n =18
n =10
b
c
d
Greenland
Antarctica
WDC
B40
–6
*
10
Be (10
3
atoms g
–1
)
10
Be (10
3
atoms g
–1
)
Figure 2
|
Re-dated ice-core, non-sea-salt sulfur
records from Greenland and Antarctica in
relation to growth anomalies in the N-Tree
composite. a, Ice-core, non-sea-salt sulfur (nssS
in parts per billion, p.p.b.) records from Greenland
(NEEM, NEEM-2011-S1) on the NS1-2011
timescale between 500
BCE and 1300 CE, with the
identified layer of Tianchi tephra
67
highlighted
(orange star). Calendar years are given for the start
of volcanic sulfate deposition. Events used as
fixed age markers to constrain the dating (536
CE,
626
CE, 775 CE, 939 CE and 1258 CE) are
indicated (purple stars). Annually resolved
10
Be
concentration record (green) from NEEM-2011-S1
encompassing the two D
14
C excursion events in
trees from 775
CE and 994 CE. b, Tree-ring
growth anomalies (relative to 1000–1099
CE) for
the N-Tree composite
42,43,76–78
. c, nssS records
from Antarctica (red, WDC; pink, B40) on the
WD2014 timescale and annually resolved
10
Be
concentrations from WDC. d, Superposed epoch
analysis for 28 large volcanic signals during the past
2,500 years. Tree-ring growth anomalies relative
to the timing of reconstructed sulfate deposition
in Greenland (NS1-2011) are shown for 1250–
2000
CE (black trace) and 500BCE to 1250 CE
(green trace).
a
1.0
0.5
00
–0.5
–1.0
–1.5
Temperature anomaly
(°C relative to 1961–1990)
–7.5 W m
2
1815
1257
1458
–426
–44
540
19001700150013001100900700500300100–100–300–500
Year (
BCE or CE)
Tropical eruption (n = 81)
Northern Hemisphere eruption (n = 140)
Southern Hemisphere eruption (n = 62)
40 largest eruptions
Global volcanic aerosol forcing
(W m
–2
)
–40
–30
–20
–10
–40
–3
–2
–1
0
1
2
Tree growth anomaly
(z-scores relative to 1000–1099)
N-Tree composite
40 minimum (coldest) years
12 minimum (coldest) decades
PAGES-2k (Europe+Arctic)
b
Figure 3
|
Global volcanic aerosol forcing and Northern Hemisphere
temperature variations for the past 2,500 years. a, 2,500-year record of tree-
growth anomalies (N-Tree
42,43,76–78
; relative to 1000–1099 CE) and reconstructed
summer temperature anomalies for Europe and the Arctic
3
with the 40
coldest single years and the 12 coldest decades based on N-Tree indicated.
b, Reconstructed global volcanic aerosol forcing from bipolar sulfate composite
records from tropical (bipolar), Northern Hemisphere, and Southern
Hemisphere eruptions. Total (that is, time-integrated) forcing values are
calculated by summing the annual values for the duration of volcanic sulfur
deposition. The 40 largest volcanic signals are indicated, and ages are given
for events representing atmospheric sulfate loading exceeding that of the
Tambora 1815 eruption.
00 MONTH 2015 | VOL 000 | NATURE | 3
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

History of volcanic forcing
Employing our revised timescales and new high-resolution, ice-core
sulfur measurements, we developed an extended reconstruction of
volcanic aerosol deposition since early Roman times for both polar
ice sheets, from which we then estimated radiative forcing using
established transfer functions
15
(Fig. 3, Methods, Supplementary
Data 3–5). This forcing series is characterized by improved dating
accuracy, annual resolution, and a larger number of ice-core records
in the Antarctic ice-core sulfate composite
17
than in previous recon-
structions
6,7
. It spans 2,500 years, allowing investigation of climate–
volcano linkages more accurately and earlier than with previous
reconstructions. It also provides a perspective on volcanic influences
during major historical epochs, such as the growth of Roman imperial
power and subsequent decline during the ‘Migration Period’ (the early
part of the first millennium
CE) in Europe—times of (1) demographic
and economic expansion as well as relative societal stability and
(2) political turmoil and population instability, respectively
33
. With
improved dating and lower volcanic-sulfate detection limits from
the Antarctic array
17
, we were able to detect, estimate, and attribute
volcanic aerosol loading and forcing from 283 individual eruptive
events during this period (Fig. 3).
We attributed about half of these to mid- to high-latitude Northern
Hemisphere sources, while 81 were attributed to tropical eruptions
(having synchronous sulfate deposition on both polar ice sheets).
These tropical volcanic eruptions contributed 64% of total volcanic
forcing throughout the period, with five events exceeding the sulfate
loading from the 1815 Tambora eruption in Indonesia (Fig. 3,
Extended Data Table 4). Events in 426
BCE and 44 BCE rival the great
1257
CE Samalas eruption as the largest sulfate-producing eruptions
during this time. These two earlier events have not been widely
regarded as large tropical eruptions with potential for strong climate
impact
20
, owing to the lack of complete and synchronized sulfate
records from Greenland and Antarctica. We base the claim that
these two eruptions were tropical in origin and caused large radiative
perturbations on the observation that ice cores from Greenland and
Antarctica record coeval (within their respective age uncertainties)
and exceptionally high volcanic sulfate concentrations. Both of these
events were followed by strong and persistent growth reduction in
tree-ring records
34
(Fig. 2) as is typically observed after large tropical
eruptions during the Common Era (Fig. 3).
Post-volcanic summer cooling
Superposed epoch analyses (Methods) performed on the ‘N-Tree’
composite record centred on the largest volcanic signals between
500
BCE and 1250 CE as well as between 1250 CE and 2000 CE, show a
clear post-volcanic cooling signal. For both periods, maximum tree-
ring response lagged the date of initial increase of sulfate deposition by
one year (Fig. 2), consistent with the response observed if using only
Large
eruption
–5 0 5 10 15
–1.5
–1.0
–0.5
0.0
Europe
–0.42 °C
a
Year from peak forcing
JJA temperature anomaly ( °C)
–5 0 5 10 15
–1.5
–1.0
–0.5
0.0
Year from peak forcing
JJA temperature anomaly ( °C)
–5 0 5 10 15
–1.5
–1.0
–0.5
0.0
Year from peak forcing
JJA temperature anomaly ( °C)
N = 24
Large
eruption
N = 24
Large
eruption
N = 24
Northern Europe
–0.58 °C
b
Central Europe
–0.46 °C
c
–5 0 5 10 15
–1.5
–1.0
–0.5
0.0
Year from peak forcing
JJA temperature anomaly ( °C)
–5 0 5 10 15
–1.5
–1.0
–0.5
0.0
Year from peak forcing
JJA temperature anomaly ( °C)
–5 0 5 10 15
–1.5
–1.0
–0.5
0.0
Year from peak forcing
JJA temperature anomaly ( °C)
Tropical
eruption
–0.43 °C
d
N = 19
Tropical
eruption
N = 19
Tropical
eruption
N = 19
–0.62 °C
e
–0.53 °C
f
JJA temperature anomaly ( °C)
–5 0 5 10 15
Year from peak forcin
g
–5 0 5 10 15
Year from peak forcin
g
–5 0 5 10 15
Year from peak forcin
g
Northern
Hemisphere
eruption
–2.0
–1.0
0.0
JJA temperature anomaly ( °C)
–2.0
–1.0
0.0
JJA temperature anomaly ( °C)
–2.0
–1.0
0.0
–0.37 °C
g
N = 5
Eruption
≥Tambora
1815
Eruption
≥Tambora
1815
–0.89 °C
h
N = 4
N = 4
i
–1.03 °C
Figure 4
|
Post-volcanic cooling. Superposed
composites (time segments from selected periods
in the Common Era positioned so that the years
with peak negative forcing are aligned) of the JJA
(June, July and August) temperature response to
the 24 largest eruptions (exceeding the Pinatubo
1991 eruption). a–c, For three regional
reconstructions in Europe
3,35,42
. d–f, For the 19
largest tropical eruptions. g, For the five largest
Northern Hemisphere eruptions. h, i, For the
eruptions with negative forcing larger than that of
the Tambora 1815 eruption for Northern Europe
(h) and for Central Europe (i). Note the different
scale for g–i. JJA temperature anomalies (in uC)
for 15 years after reconstructed volcanic peak
forcing, relative to the five years before the volcanic
eruption, are shown. Dashed lines present twice
the standard error of the mean (2 s.e.m.) of the
temperature anomalies associated with the
multiple eruptions. Five-year average post-volcanic
temperatures are shown for each reconstruction
(lag 0 to lag 14 years, grey shading).
4 | NATURE | VOL 000 | 00 MONTH 2015
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

historically documented eruptions with secure dating for the past 800
years
18
. The sharp and immediate (that is, less than one year lag time)
response of tree growth to the ice-core volcanic signal throughout the
past 2,500 years further corroborates the accuracy of our new ice-core
timescales (Extended Data Fig. 4).
Of the 16 most negative tree-growth anomalies (that is, the coldest
summers) between 500
BCE and 1000 CE, 15 followed large volcanic
signals—with the four coldest (43
BCE, 536 CE, 543 CE, and 627 CE)
occurring shortly after several of the largest events (Extended Data
Tables 4 and 5). Similarly, the coldest summers in Europe during
the Common Era
3
were associated with large volcanic eruptions
(Extended Data Table 5). Reduced tree growth after volcanic
eruptions was also prominent in decadal averages of the ‘N-Tree’
composite. All 16 decades with the most reduced tree growth for
our 2,500-year period followed large eruptions (Fig. 3, Extended
Data Table 5). In many cases, such as the coldest decade, from
536
CE to 545 CE
3
, sustained cooling was associated with the combined
effect of several successive volcanic eruptions.
Strong post-volcanic cooling was not restricted to tropical erup-
tions; it also followed Northern Hemisphere eruptions (Fig. 4), with
maximum cooling in the year of volcanic-sulfate deposition. In con-
trast to the average of the 19 largest
CE tropical eruptions, however, the
Northern-Hemisphere-only eruptions did not give rise to any notice-
able long-term cooling effect (Fig. 4). The persistence of implied post-
volcanic cooling following the largest tropical eruptions is strongly
expressed in temperature reconstructions based on tree-ring width
measurements (for example, those from the Alps), with recovery
times of more than ten years. Persistent cooling, with temperature
reduction notably below the pre-eruption baseline for six consecutive
years, is also observed in temperature reconstructions based on
maximum latewood density (for example, those from Northern
Scandinavia), which is the preferred proxy with which to quantify
volcanic cooling contributions on climate owing to its less marked
biological memory effects
35
(Fig. 4). These findings indicate that erup-
tion-induced climate anomalies following large tropical eruptions
may last longer than is indicated in many climate simulations
(,3–5 years)
9,36,37
and that potential positive feedbacks initiated after
large tropical eruptions (for example, sea-ice feedbacks) may not be
adequately represented in climate simulations
38,39
.
The five-year averaged (lag 0 to lag 4 years) cooling response
over three Northern Hemisphere regions (Methods) following the
19 largest Common Era tropical eruptions was 20.6 6 0.2 uC (two
standard errors of the mean, 2 s.e.m.), and that of large Northern
Hemisphere eruptions was 20.4 6 0.4 uC, with the strongest cooling
induced in the high latitudes. Overall, cooling was proportional to the
magnitude of volcanic forcing, with stratospheric sulfate loading
exceeding that of the Tambora eruption inducing the strongest res-
ponse of 21.1 6 0.6 uC (Figs 3 and 4).
Global climate anomalies in 536–550 CE
Our new dating allowed us to clarify long-standing debates concern-
ing the origin and consequences of the severe and apparently global
climate anomalies observed in the period 536–550
CE, which began
with the recognition of the ‘‘mystery cloud’’ of 536
CE
40
observed in the
Mediterranean basin. Under previous ice-core dating, it has been
argued that this dust veil corresponded to an unknown tropical erup-
tion dated 533–534
CE (62 years)
41
. Using our revised timescales, we
found at least two large volcanic eruptions around this period (Fig. 5).
The first eruptive episode in 535
CE or early 536 CE injected large
amounts of sulfate and ash into the atmosphere, apparently in the
Northern Hemisphere. Geochemistry of tephra filtered from the
NEEM-2011-S1 ice core at a depth corresponding to 536
CE indicated
multiple North American volcanoes as likely candidates for a com-
bined volcanic signal (Extended Data Fig. 5, Methods, Supplementary
Data 5). Historical observations (Extended Data Table 3) identified
atmospheric dimming as early as 24 March 536
CE, and lasting up to
18 months. The summer of 536
CE appeared exceptionally cold in all
tree-ring reconstructions in the extra-tropical Northern Hemisphere
from North America
34
, over Europe
35,42,43
to Asia
44
. Depending on the
reconstruction method used, European summer temperatures in
536
CE dropped 1.6–2.5 uC relative to the previous 30-year average
3
.
The second eruptive episode in 539
CE or 540 CE, identified in both
Greenland and Antarctica ice-core records and hence probably trop-
ical in origin, resulted in up to 10% higher global aerosol loading than
nssS (p.p.b.)
500 520 540 560 580 600 620 640 660 680 700
Year (
CE)
0
30
60
nssS (p.p.b.)
WDC
-6
-4
-2
0
2
Tree growth anomaly
(z-scores with respect to 1000–1099)
0
40
80
120
160
nssS (p.p.b.)
0
60
120
B40
N-Tree
NEEM
201-S1
TUNU
536
540
574
626
682
540
575
682
536
540-47
574
627
Bipolar event
Northern Hemisphere event
Southern Hemisphere event
Bristlecone pine late wood frost rings
–3
–2
–1
0
1
Temperature anomaly
(°C with respect to 1961–1990)
Europe
a
b
c
Figure 5
|
Volcanism and temperature variabil-
ity during the migration period (500–705
CE).
a, Ice-core non-sea-salt sulphur (nssS) records
from Greenland (black trace, NEEM-2011-S1; blue
trace, TUNU2013). Calendar years for five large
eruptions are given for the start of volcanic sulfate
deposition. b, Summer temperature anomalies
(orange trace) for Europe
3
, and reconstructed
N-Tree growth anomalies (green trace) and
occurrence of frost rings in North American
bristlecone pine tree-ring records. c, nssS records
from Antarctica (red trace, WDC; pink trace, B40)
on the WD2014 timescale; attribution of the
sulfur signals to bipolar, Northern Hemisphere,
and Southern Hemisphere events based on the
timing of deposition on the two independent
timescales is indicated by shading.
00 MONTH 2015 | VOL 000 | NATURE | 5
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

the Tambora 1815 eruption reconstructed from our bipolar sulfate
records. Summer temperatures consequently dropped again, by
1.4–2.7 uC in Europe in 541
CE
3
, and cold temperatures persisted in
the Northern Hemisphere until almost 550
CE
3,33,34,42
(Figs 2, 3, 5).
This provides a notable environmental context to widespread fam-
ine and the great Justinian Plague of 541–543
CE that was responsible
for decimating populations in the Mediterranean and potentially
China
45,46
. Although certain climatic conditions (for example, wet
summers) have been linked to plague outbreaks in the past
47
, a direct
causal connection of these two large volcanic episodes and subsequent
cooling to crop failures and outbreaks of famines and plagues is dif-
ficult to prove
33
. However, the exact delineation of two of the largest
volcanic signals—with exceptionally strong and prolonged Northern
Hemisphere cooling, written evidence of famines and pandemics,
as well as the socio-economic decline observed in Mesoamerica
(the ‘‘Maya Hiatus’’
48
), Europe, and Asia—supports the idea that
the latter may be causally associated with volcanically induced
climatic extremes.
Detailed study of major volcanic events during the sixth century
(Fig. 5) and an assessment of post-volcanic cooling throughout the
past 2,500 years using stacked tree-ring records and regional temper-
ature reconstructions (Fig. 4, Extended Data Fig. 4) demonstrated that
large eruptions in the tropics and high latitudes were primary drivers
of interannual-to-decadal Northern Hemisphere temperature vari-
ability. The new ice-core chronologies imply that previous multi-
proxy reconstructions of temperature that include ice-core records
2–4
have diminished high- to mid-frequency amplitudes and must be
updated to accurately capture the timing and full amplitude of palaeo-
climatic variability.
By creating a volcanic forcing index independent of but consistent
with tree-ring-indicated cooling, we provide an essential step towards
understanding of external forcing on natural climate variability dur-
ing the past 2,500 years. With the expected detection of additional
rapid D
14
C enrichment events from ongoing efforts in annual-reso-
lution
14
C tree-ring analyses
49
, there will be opportunities to further
constrain ice-core dating throughout the Holocene and develop a
framework of precisely dated, globally synchronized proxies of past
climate variability and external climate forcing.
Online Content Methods, along with any additional Extended Data display items
and Source Data, are available in the online version of the paper; references unique
to these sections appear only in the online paper.
Received 21 November 2014; accepted 6 May 2015.
Published online 8 July 2015.
1. Robock, A. Volcanic eruptions and climate. Rev. Geophys. 38, 191–219 (2000).
2. Hanhija
¨
rvi, S., Tingley, M. P. & Korhola, A. Pairwise comparisons to reconstruc t
mea n temperature in the Arctic At lantic Region over the last 2,000 years. Clim.
Dyn. 41, 2039–2060 (2013).
3. PAGES 2k Consortium. Continental-scale temperature variability during the past
two millennia. Nature Geosci. 6, 503 (2013).
4. Mann, M. E. et al. Proxy-based reconstructions of hemispheric and global surface
temperature variations over the past two millennia. Proc. Natl Acad. Sci. USA 105,
13252–13257 (2008).
5. Usoskin, I. G. A history of solar activity over millennia. Living Rev. Sol. Phys 10, 1
(2013).
6. Gao, C. C., Robock, A. & Ammann, C. Volcanic forcing of climate over the past 1500
years: an improved ice core-based index for climate models. J. Geophys. Res. 113,
http://dx.doi.org/10.1029/2008JD010239 (2008).
7. Crowley, T. J. & Unterman, M. B. Technical details concerning development of a
1200-yr proxy index of global volcanism. Earth System Sci. Data 5, 187–197
(2013).
8. Mann, M. E., Fuentes, J. D. & Rutherford, S. Underestimation of volcanic cooling in
tree-ring-based reconstructions of hemispheric temperatures. Nature Geosci. 5,
202–205 (2012).
9. Mann, M. E., Rutherford, S., Schurer, A., Tett, S. F. B. & Fuentes, J. D. Discrepancies
between the modeled and proxy-reconstructed response to volcanic forcing over
the past millennium: implicationsandpossible mechanisms. J. Geophys. Res. 118,
7617–7627 (2013).
10. Schurer, A. P., Hegerl, G. C., Mann, M. E., Tett, S. F. B. & Phipps, S. J. Separating
forced from chaotic climate variability over the past millennium. J. Clim. 26,
6954–6973 (2013).
11. Anchukaitis, K. J. et al. Tree rings and volcanic cooling. Nature Geosci. 5, 836–837
(2012).
12. Bu
¨
ntgen, U. et al. Extraterrestrial confirmation of tree-ring dating. Nature Clim.
Change 4, 404–405 (2014).
13. Esper, J., Bu
¨
ntgen, U., Luterbacher, J. & Krusic, P. J. Testing the hypothesis of post-
volcanic missing rings in temperature sensitive dendrochronological data.
Dendrochronologia 31, 216–222 (2013).
14. D’Arrigo, R., Wilson, R. & Anchukaitis, K. J. Volcanic cooling signal in tree ring
temperature records for the past millennium. J. Geophys. Res. 118, 9000–9010
(2013).
15. Plummer, C. T. et al. An independently dated 2000-yr volcanic record from Law
Dome, East Antarctica, including a new perspective on the dating of the 1450s CE
eruption of Kuwae, Vanuatu. Clim. Past 8, 1929–1940 (2012).
16. Sigl, M. et al. A new bipolar ice core record of volcanism from WAIS Divide and
NEEM and implications for climate forcing of the last 2000 years. J. Geophys. Res.
118, 1151–1169 (2013).
17. Sigl, M. et al. Insights from Antarctica on volcanic forcing during the Common Era.
Nature Clim. Change 4, 693–697 (2014).
18. Esper, J. et al. European summer temperature response to annually dated volcanic
eruptions over the past nine centuries. Bull. Volcanol. 75, 736 (2013).
19. Douglass, D. H. & Knox, R. S. Climate forcing by the volcanic eruption of Mount
Pinatubo. Geophys. Res. Lett. 32, L05710 (2005).
20. Baillie, M. G. L. Proposed re-dating of the European ice core chronology by seven
years prior to the 7th century AD. Geophys. Res. Lett. 35, L15813 (2008).
21. Baillie, M. G. L. & McAneney, J. Tree ring effects and ice core acidities clarify the
volcanic record of the 1st millennium. Clim. Past 11, 105–114 (2015).
22. Miyake, F., Nagaya, K., Masuda, K. & Nakamura, T. A signature of cosmic-ray
increase in AD 774–775 from tree rings in Japan. Nature 486, 240–242 (2012).
23. Miyake, F., Masuda, K. & Nakamura, T. Another rapid event in the carbon-14
content of tree rings. Nature Commun. 4, http://dx.doi.org/10.1038/
Ncomms2783 (2013).
24. Usoskin, I. G. et al. The AD775 cosmic event revisited: the Sun is to blame. Astron.
Astrophys. 552, http://dx.doi.org/10.1051/0004-6361/201321080 (2013).
25. Jull, A. J. T. et al. Excursions in the
14
C record at A. D. 774–775 in tree rings from
Russia and America. Geophys. Res. Lett. 41, 3004–3010 (2014).
26. Gu
¨
ttler, D. et al. Rapid increase in cosmogenic
14
C in AD 775 measured in New
Zealand kauri trees indicates short-lived increase in
14
C production spanning both
hemispheres. Earth Planet. Sci. Lett. 411, 290–297 (2015).
27. Miyake, F. et al. Cosmic ray event of AD 774–775 shown in quasi-annual
10
Be data
from the Antarctic Dome Fuji ice core. Geophys. Res. Lett. 42, 84–89 (2015).
28. Webber, W. R., Higbie, P. R. & McCracken, K. G. Production of the cosmogenic
isotopes H-3, Be-7, Be-10, and Cl-36 in the Earth’s atmosphere by solar and
galactic cosmic rays. J. Geophys. Res. 112, A10106 (2007).
29. 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).
30. Vinther, B. M. et al. A synchronized dating of three Greenland ice cores throughout
the Holocene. J. Geophys. Res. 111, D13102 (2006).
31. Lavigne, F. et al. Source of the great A.D. 1257 mystery eruption unveiled, Samalas
volcano, Rinjani Volcanic Complex, Indonesia. Proc. Natl Acad. Sci. USA 110,
16742–16747 (2013).
32. Winstrup, M. et al. An automated approach for annual layer counting in ice cores.
Clim. Past 8, 1881–1895 (2012).
33. McCormick, M. et al. Climate change during and after the Roman Empire:
reconstructing the past from scientific and historical evidence. J. Interdisc. Hist. 43,
169–220 (2012).
34. Salzer, M. W. & Hughes, M. K. Bristlecone pine tree rings and volcanic eruptions
over the last 5000 yr. Quat. Res. 67, 57–68 (2007).
35. Esper, J., Duthorn, E., Krusic, P. J., Timonen, M. & Buntgen, U. Northern European
summer temperature variations over the Common Era from integrated tree-ring
density records. J. Quat. Sci. 29, 487–494 (2014).
36. Crowley, T. J. Causes of climate change over the past 1000 years. Science 289,
270–277 (2000).
37. Driscoll, S., Bozzo, A., Gray, L. J., Robock, A. & Stenchikov, G. Coupled Model
Intercomparison Project 5 (CMIP5) simulations of climate following volcanic
eruptions. J. Geophys. Res. 117, D17105 (2012).
38. Schneider, D. P., Ammann, C. M., Otto-Bliesner, B. L. & Kaufman, D. S. Climate
response to large, high-latitude and low-latitude volcanic eruptions in the
Community Climate System Model. J. Geophys. Res. 114, D15101 (2009).
39. Zanchettin, D. et al. Inter-hemispheric asymmetry in the sea-ice response to
volcanic forcing simulated by MPI-ESM (COSMOS-Mill). Earth Syst. Dyn. 5,
223–242 (2014).
40. Stothers, R. B. Mystery cloud of Ad-536. Nature 307, 344–345 (1984).
41. Larsen, L. B. et al. New ice coreevidence for a volcanic cause of the AD 536 dust veil.
Geophys. Res. Lett. 35, L04708 (2008).
42. Bu
¨
ntgen, U. et al. 2500 years of European climate variability and human
susceptibility. Science 331, 578–582 (2011).
43. Esper, J. et al. Orbital forcing of tree-ring data. Nature Clim. Change
2, 862–866
(2012).
44. D’Arrigo, R. et al. 1738 years of Mongolian temperature variability inferred from a
tree-ring width chronology of Siberian pine. Geophys. Res. Lett. 28, 543–546
(2001).
45. Zhang, Z. B. et al. Periodic climate cooling enhanced natural disasters and wars in
China during AD 10–1900. Proc. R. Soc. B 277, 3745–3753 (2010).
46. Stothers, R. B. Volcanic dry fogs, climate cooling, and plague pandemics in Europe
and the Middle East. Clim. Change 42, 713–723 (1999).
6 | NATURE | VOL 000 | 00 MONTH 2015
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

47. Stenseth, N. C. et al. Plague dynamics are driven by climate variation. Proc. Natl
Acad. Sci. USA 103, 13110–13115 (2006).
48. Dull,R. A. Evidencefor forestclearance, agriculture, and human-inducederosion in
Precolumbian El Salvador. Ann. Assoc. Am. Geogr. 97, 127–141 (2007).
49. Taylor, R. E. & Southon, J. Reviewing the Mid-First Millennium BC C-14
‘‘warp’’ using C-14/bristlecone pine data. Nucl. Instrum. Meth. B 294, 440–443
(2013).
Supplementary Information is available in the online version of the paper.
Acknowledgements We thank the many people involved in logistics, drill development
and drilling, and ice-core processing and analysis in the field and our laboratories. This
work was supported by the US National Science Foundation (NSF). We appreciate the
support of the WAIS Divide Science Coordination Office (M. Twickler and J. Souney) for
collection and distribution of the WAIS Divide ice core; Ice Drilling and Design and
Operations (K. Dahnert) for drilling; the National Ice Core Laboratory (B. Bencivengo)
for curating the core; Raytheon Polar Services (M. Kippenhan) for logistics support in
Antarctica; and the 109th New York Air National Guard for airlift in Antarctica. NEEM is
directed and organized by the Center of Ice and Climate at the Niels Bohr Institute and
the US NSF, Office of Polar Programs. It is supported by funding agencies and
institutions in Belgium (FNRS-CFB and FWO), Canada (NRCan/GSC), China (CAS),
Denmark (FIST), France (IPEV, CNRS/INSU, CEA and ANR), Germany (AWI), Iceland
(RannIs), Japan (NIPR), Korea (KOPRI), The Netherlands (NWO/ALW), Sweden (VR),
Switzerland (SNF), the UK (NERC), and the USA (the US NSF, Office of Polar Programs).
We thank B. Nolan, O. Amir, K. D. Pang, M. McCormick, A. Matthews, and B. Rossignol for
assistance in surveying and/or interpreting the historical evidence. We thank S. Kuehn
for commenting on possible correlations for the tephra. We thank A. Aldahan and
G. Possnert for their support in the NGRIP
10
Be preparations and measurements at the
Department of Earth Sciences and the Tandem laboratory at Uppsala University. We
gratefully acknowledge R. Kreidberg for his editorial advice. The following individual
grants supported this work: NSF/OPP grants 0839093, 0968391, and 1142166 to
J.R.M. for development of the Antarctic ice core recordsand NSF/OPP grants 0909541,
1023672, and 1204176 to J.R.M. for development of the Arctic ice core records.
M.W. was funded by the Villum Foundation. K.C.W. was funded by NSF/OPP grants
0636964 and 0839137. M.C. and T.E.W. were funded by NSF/OPP grants 0839042
and 0636815.F.L. was funded by the Yale Climate and Energy Institute, Initiative for the
Science of the Human Past at Harvard, and the Rachel Carson Center for Environment
and Society of the Ludwig-Maximilians-Universita
¨
t (LMU Munich). C.K. was fundedby a
Marie Curie FP7 Integration Grant within the 7th European Union Framework
Programme.M. Salzer was funded by NSF grant ATM 1203749. R.M. was fundedby the
Swedish Research Council (DNR2013-8421). The division of Climate and
Environmental Physics, Physics Institute, University of Bern, acknowledges financial
support by the SNF and the Oeschger Centre.
Author Contributions M. Sigl designed the study with input from J.R.M., M.W., G.P., and
F.L. The manuscript was written by M. Sigl, M.W., F.L., and J.R.M.,withcontributions from
K.C.W., G.P., U.B., and B.M.V. in interpretation of the measurements. Ice-core chemistry
measurements were performed by J.R.M., M. Sigl, O.J.M., N.C., D.R.P. (NEEM, B40,
TUNU2013), and by S.S., H.F., R. Mulvaney (NEEM). K.C.W., T.E.W., and M.C. completed
ice core
10
Be measurements. F.M. and R. Muscheler were responsible for the NGRIP ice
core
10
Be measurements. M. Sigl, M.W., B.M.V., and J.R.M. analysed ice-core data and
developed age models. F.L. and C.K. analysed historical documentary data. G.P. and
J.R.P. performed ice-core tephra analysis and data interpretation. U.B. and M. Salzer
contributed tree-ring data. D.D.-J., B.M.V., J.P.S., S.K., and O.J.M. were involved in drilling
of the NEEM ice core. TUNU2013 was drilled by M. Sigl, N.C. and O.J.M., and the B40 ice
core was drilled by S.K. and made available for chemistry measurements. D.D.-J. and
J.P.S. were responsible for NEEM project management, sample distribution, logistics
support, and management. All authors contributed towards improving the final
manuscript.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Readers are welcome to comment on the online version of the paper. Correspondence
and requests for materials should be addressed to J.R.M. (joe.mcconnell@dri.edu).
00 MONTH 2015 | VOL 000 | NATURE | 7
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

METHODS
Ice cores. This study included new and previously described ice-core records
from five drilling sites (Extended Data Fig. 1, Supplementary Data 1). The
upper 577 m of the 3,405-m WAIS Divide (WDC) core from central West
Antarctica and a 410-m intermediate-length core (NEEM-2011-S1) drilled in
2011 close to the 2,540-m North Greenland Eemian Ice Drilling (NEEM)
50
ice
core have previously been used to reconstruct sulfate deposition in both polar
ice sheets
16
. These coring sites are characterized by relatively high snowfall
(,200 kg m
22
yr
21
) and have comparable elevation, latitude, and deposition
regimes. WDC and NEEM-2011-S1 provided high-resolution records that
allowed annual-layer dating based on seasonally varying impurity content
16
.
New ice-core analyses included the upper 514 m of the main NEEM core used
to extend the record of NEEM-2011-S1 to cover the past 2,500 years, as well as
B40 drilled in 2012 in Dronning Maud Land in East Antarctica and TUNU2013
drilled in 2013 in Northeast Greenland—both characterized by lower snowfall
rates (,70–100 kg m
22
yr
21
). Volcanic sulfate concentration from B40 had been
reported previously for the past 2,000 years
17
, but we extended measurements to
200 m depth to cover the past 2,500 years.
High-resolution, ice-core aerosol analyses. Ice-core analyses were performed at
the Desert Research Institute (DRI) using 55–100-cm-long, longitudinal ice-core
sections (33 mm 3 33 mm wide). The analytical system for continuous analysis
included two Element2 (Thermo Scientific) high-resolution inductively coupled
plasma mass spectrometers (HR-ICP-MS) operating in parallel for measurement
of a broad range of ,35 elements; an SP2 (Droplet Measurement Technologies)
instrument for black carbon measurements; and a host of fluorimeters and spec-
trophotometers for ammonium (NH
4
1
), nitrate (NO
3
2
), hydrogen peroxide
(H
2
O
2
), and other chemical species. All measurements were exactly co-registered
in depth, with depth resolution typically less than 10–15 mm
51–53
. We corrected
total sulfur (S) concentrations for the sea-salt S contribution using sea-salt Na
concentrations
16
. Measurements included TUNU2013 and NEEM (400–515 m)
in Greenland, and B40 in Antarctica (Extended Data Fig. 1). Gaps (that is, ice not
allocated to DRI) in the high-resolution sulfur data of the NEEM core were filled
with ,4-cm-resolution discrete sulfate measurements using fast ion-chromato-
graphy techniques
54
performed in the field between 428 m and 506 m depth.
Independent analyses of the upper part of the NEEM main core were per-
formed in the field using a continuous flow analysis (CFA) system
55
recently
modified to include a new melter head design
56
.Ca
21
,NH
4
1
, and H
2
O
2
were
analysed by fluorescence spectroscopy; Na
1
and NO
3
2
by absorption spectro-
scopy; conductivity of the meltwater by a micro flow cell (Amber Science); and a
particle detector (Abakus, Klotz) was used for measuring insoluble dust particle
concentrations and size distribution
57
. Effective depth resolution was typically
better than 20 mm. Measurements were exactly synchronized in depth using a
multicomponent standard solution; the accuracy of the depth assignment for all
measurements was typically better than 5 mm.
High-resolution measurements of
10
Be in ice cores using AMS. Accelerator
mass spectrometry (AMS) was used to analyse samples from the NEEM-2011-S1,
WDC, NGRIP, and TUNU2013 ice cores encompassing the time period of
the D
14
C anomalies from tree-ring records
12,22–25
were used for
10
Be analysis
(Supplementary Data 1). NEEM-2011-S1 and WDC were sampled in exact
annual resolution, using the maxima (minima in WDC) of the annual cycles of
Na concentrations to define the beginning of the calendar year
16
. NGRIP was
sampled at a constant resolution of 18.3 cm, providing an age resolution of about
one year. Similarly, TUNU2013 was sampled in quasi-annual resolution accord-
ing to the average annual-layer thickness expected at this depth based on prior
volcanic synchronization to NEEM-2011-S1. The relative age uncertainty for
TUNU2013 with respect to the dependent NEEM-2011-S1 chronology at this
depth is assumed to be 61 year at most, given a distinctive match for selected
volcanic trace elements in both ice-core records (752–764
CE, NS1-2011 time-
scale). Sample masses ranged between 100 g and 450 g, resulting in median overall
quantification uncertainties of less than 4%–7%. The
10
Be/
9
Be ratios of samples
and blanks were measured relative to well documented
10
Be standards
13
by AMS
at Purdue’s PRIME laboratory (WDC, NEEM-2011-S1, Tunu2013) and Uppsala
University (NGRIP)
58,59
. Results were corrected for an average blank
10
Be/
9
Be
ratio, corresponding to corrections of 2%–10% of the measured
10
Be/
9
Be ratios.
Annual-layer dating using the StratiCounter algorithm. For annual-layer
interpretation, we used DRI’s broad-spectrum aerosol concentration data from
WDC (188–577 m), NEEM-2011-S1 (183–411 m), and NEEM (410–515 m), as
well as NEEM aerosol concentration data (183–514 m) from the field-based
CFA system. The original timescale for NEEM-2011-S1 was based on volcanic
synchronization to the NGRIP sulfate record on the GICC05 timescale and
annual-layer interpretation between the volcanic age markers, whereas WDC
was previously dated by annual-layer counting
16
.
Parameters with strong intra-annual variability included tracers of sea salt (for
example, Na, Cl, Sr), dust (for example, Ce, Mg, insoluble particle concentration),
and marine biogenic emissions such as non-sea-salt sulfur (nssS). Tracers of
biomass-burning emissions, such as BC, NH
4
1
, and NO
3
2
, also showed strong
seasonal variations in deposition during pre-industrial times
16,60,61
. The data sets
used for annual-layer interpretation are provided in Extended Data Table 1. For
NEEM-2011-S1, the final database used for annual-layer dating included 13
parameters and the ratio of nssS/Na. For WDC, the final database included five
parameters and the ratio of nssS/Na. For NEEM (410–515 m depth), the final
database included eight parameters (Na
1
,Ca
21
,NH
4
1
,H
2
O
2
,NO
3
2
,
conduc-
tivity, insoluble particle concentrations, and electrical conductivity
62
) from the
field-based measurements and eleven parameters (Na, Cl, Mg, Mn, Sr, nssS, nssS/
Na, nssCa, black carbon, NO
3
2
,NH
4
1
) from the DRI system.
We focused here on the time period before the large volcanic eruption of
Samalas in 1257 CE
31
, clearly detectable as an acidic peak in both ice-core records,
and consequently started annual-layer counting of NEEM-2011-S1, NEEM, and
WDC at the depth of the corresponding sulfur signal. For the time period 1257
CE
to present, ice-core chronologies were constrained by numerous historic erup-
tions and large sulfate peaks, showing a strong association to Northern
Hemisphere cooling events as indicated by tree-ring records
16
.
We applied the StratiCounter layer-detection algorithm
32
to the multi-para-
meter aerosol concentration records (n 5 14 for NEEM-2011-S1; n 5 6 for WDC;
n 5 8 for NEEM , 410 m; n 5 19 for NEEM . 410 m) to objectively determine
the most likely number of annual layers in the ice cores along with corresponding
uncertainties. The StratiCounter algorithm is based on statistical inference in
Hidden Markov Models (HMMs), and it determines the maximum-likelihood
solution based on the annual signal in all aerosol records in parallel. Some of these
displayed a high degree of similarity, so we weighted these records correspond-
ingly lower. The algorithm was run step-wise down the core, each batch covering
approximately 50 years, with a slight overlap. All parameters for the statistical
description of a mean layer and its inter-annual variability in the various aerosol
records were determined independently for each batch as the maximum-like-
lihood solution. The algorithm simultaneously computes confidence intervals for
the number of layers within given sections, allowing us to provide uncertainty
bounds on the number of layers between selected age-marker horizons (Extended
Data Table 2).
Annual-layer detection in the NEEM main core below 410 m was made more
difficult by frequent occurrence of small gaps in the two independent high-reso-
lution aerosol data sets. Depending on the parameter, data gaps from the CFA
field measurements accounted for up to 20% of the depth range between 410 m
and 515 m, but the combined aerosol records from both analyses provided an
almost complete aerosol record with 96% data coverage. As this was the first time
that the StratiCounter algorithm was used simultaneously on data records from
two different melt systems, with different characteristics and lack of exact co-
registration, we also manually determined annual layers below 410 m using the
following approaches: one investigator used Na and nssCa concentrations and the
ratio of nssS/Na (from DRI analysis) as well as Na
1
and insoluble particle con-
centrations (from CFA analysis) as primary dating parameters. Black carbon,
NH
4
1
, nssS, and conductivity were used as secondary dating parameters where
annual-layer interpretation was ambiguous. A second investigator used DRI’s Na,
Ca, BC, NH
4
1
and CFA Na
1
,Ca
21
, and NH
4
1
measurements as parameters. The
annual-layer interpretation of the NEEM core between 410 m and 514 m from
investigator 1 was within the interpretation uncertainties of the StratiCounter
output, from which it differed by less than a single year over the majority of this
section, and it differed from independently counted timescales (for example,
GICC05)
62
by on average less than three years (Extended Data Fig. 2). This set
of layer counts was used for the resulting timescale.
New ice-core chronologies for NS1-2011 and WD2014. We defined the depth of
NEEM-2011-S1 containing the maximum
10
Be concentration as the year 775 CE.
Relative to this constraint, the maximum-likelihood ages for three large volcanic
sulfate peaks were within one year of documented historical reports from early
written sources of prominent and sustained atmospheric dimming observed in
Europe and/or the Near East (Extended Data Table 3, Supplementary Data 2).
Automated-layer identification for NEEM-2011-S1 was therefore constrained by
tying the respective ice-core volcanic signals to the corresponding absolute his-
torically dated ages of 536
CE, 626 CE, and 939 CE (Extended Data Table 2)—
thereby creating a new ice-core timescale (NS1-2011). The volcanic sulfur signal
corresponding to the eruption of Samalas believed to have occurred in late 1257
31
was constrained to 1258 CE to account for several months’ delay in sulfate depos-
ition in the high latitudes. Before 86
CE (the bottom depth of NEEM-2011-S1), the
NS1-2011 timescale was extended using the manually derived annual-layer inter-
pretation of the combined NEEM aerosol data sets back to 500
BCE (Fig. 2).
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

In NS1-2011 we did not attribute acid layers to the historical eruptions Vesuvius
79 and Hekla 1104, due to a lack of corroborative tephra at these depths in this and a
previous study
63
. Possible Vesuvian tephra was reported from the Greenland Ice
Sheet Project (GRIP) ice core at 429.3 m depth
64
, but in view of the new annual-
layer dating results (Extended Data Fig. 3), we concluded that this layer dates to
87/88
CE. Furthermore, volcanic sulfate deposition values for the corresponding
event show a strong spatial gradient over Greenland with highest values in north-
west Greenland
16
and lowest in central and south Greenland
65
, favouring the
attribution of a volcanic source from the high latitudes. Documentary sources
(Supplementary Data 2) also suggest that the main vector of ash transport following
the Vesuvius 79
CE eruption was towards the eastern Mediterranean
66
.
For WDC, we do not have other sufficiently well determined age constraints
besides the rapid
10
Be increase in 775 CE and the sulfur signal of the Samalas 1257
eruption. Therefore, no additional constraints were used when creating the new
ice-core timescale (‘‘WD2014’’) from the StratiCounter annual-layer interpreta-
tion back to 396
BCE.
Depth-age information for six distinctive marker horizons in Greenland is
given, and five of these horizons were used to constrain NS1-2011 (Extended
Data Table 3). Similarly, depth information, the number of annual layers, and
95% confidence intervals between distinctive volcanic marker horizons are given
for NEEM, NEEM-2011-S1, and WDC, supporting attribution of these ice-core
signals to eruptions in the low latitudes with bipolar sulfate deposition.
Evaluation of NS1-2011 using independent age information. We evaluated
timescale accuracy using additional distinctive age markers not used during
chronology development:
(1) Tephra from the eruption of Changbaishan/Tianchi (China)
67
was detected
in NEEM-2011-S1 in 946–947
CE, in agreement with widespread documentary
evidence of an eruption in that region in winter 946/47
CE
68
also supported by a
high-precision
14
C wiggle-match age of 946 6 3 CE obtained from a tree killed
during this eruption
68
.
(2) The rapid increase of
10
Be from the 994 CE event occurred in NEEM-2011-
S1 in 993
CE, consistent with D
14
C from Japanese tree rings showing that the
rapid increase in radionuclide production took place between the Northern
Hemisphere growing seasons of 993
CE and 994 CE
23
.
(3) To assess the accuracy of the NS1-2011 timescale before the earliest age
marker at 536
CE, we compiled an independent time series of validation points,
featuring years with well dated historical reports of atmospheric phenomena
associated with high-altitude volcanic dust and/or aerosols (Supplementary
Data 2) as known from modern observations to occur after major eruptions
(for example, the Krakatau eruption of 1883). These phenomena include dimin-
ished sunlight, discoloration of the solar disk, solar coronae (that is, Bishop’s
Rings), and deeply red twilights (that is, volcanic sunsets)
69,70
. Thirty-two events
met our criteria as validation points for the pre-536
CE NS1-2011 timescale. For
the earliest in 255
BCE, it was reported in Babylon that ‘‘the disk of the sun looked
like that of the moon’’
71
. For the latest in 501 CE, it was reported in North China
that ‘‘the Sun was red and without brilliance’’
72
. We found that NEEM volcanic
event years (including both NEEM and NEEM-2011-S1 data) occurred closely in
time (that is, within a conservative 63-year margin) to 24 (75.0%) of our valid-
ation points (Extended Data Fig. 2). To assess whether this association arose
solely by chance, we conducted a Monte Carlo equal means test with 1,000,000
iterations (Supplementary Data 2) and found that the number of volcanic event
years within three years of our validation points was significantly greater than
expected randomly (P , 0.001). A significant association was also observed
(P , 0.001) when using less conservative error margins (61 and 62 years) and
when excluding any historical observations with less certainty of a volcanic origin
(Supplementary Data 2). When placing volcanic event years on the original
GICC05 timescale, we did not observe any statistically significant association
with our independent validation points.
Potential causes of a previous ice-core dating bias. Interpretation of annual
layers in ice cores is subject to accumulating age uncertainty due to ambiguities in
the underlying ice-core profiles
30,73
. Bias in existing chronologies may arise from
several factors, including: (1) low effective resolution of some ice-core measure-
ments (NGRIP, GRIP); (2) use of only single (or few) parameters for annual-layer
interpretation (GRIP, Dye-3 ice cores); (3) intra-annual variations in various ice-
core parameters falsely interpreted as layer boundaries (for example, caused by
summer melt in Dye-3)
74
; (4) use of tephra believed to originate from the 79 CE
Vesuvian eruption
64
as a fixed reference horizon to constrain the Greenland ice-
core dating
30
; (5) use of manual-layer interpretation techniques that may favour
interpretations consistent with a priori knowledge or existing chronologies
(WDC)
16,21
.
Volcanic synchronization of B40, TUNU2013, and NGRIP. Two high-resolution
sulfur ice-core records (TUNU2013, Greenland and B40, Antarctica) were syn-
chronized to NEEM-2011-S1 and WDC, respectively, using volcanic stratigraphic
age markers
17
with relative age uncertainty between the tie-points estimated
to not exceed 62 years. The NGRIP sulfate record measured at 5 cm depth
resolution
15
similarly was synchronized to NS1-2011 using 124 volcanic tie-
points between 226 and 1999 CE. During the time period with no sulfur record
yet available for WDC (before 396
BCE), a tentative chronology for B40 was
derived by linearly extrapolating mean annual-layer thickness for B40 as derived
from the synchronization to WDC between the earliest volcanic match points.
2,500 year global volcanic forcing ice-core index. We constructed an index of
global volcanic aerosol forcing by (1) re-dating and extending to 500
BCE an
existing reconstruction of sulfate flux from an Antarctic ice-core array
17
by apply-
ing an area weighting of 80/20 between East Antarctica and West Antarctica
to B40 and WDC volcanic sulfate flux values, respectively; (2) compositing
NGRIP and the NEEM-2011-S1/NEEM sulfate flux records to a similar
Greenland sulfate deposition composite back to 500
BCE; (3) using established
scaling functions
6,75
to estimate hemispheric sulfate aerosol loading from both
polar ice-core composites; and (4) scaling global aerosol loading to the total (that
is, time-integrated) radiative volcanic aerosol forcing following the Tambora
1815 eruption
7
. Since the NS1-2011 and WD2014 timescales are independent
of each other, the timing of bipolar events had to be adjusted to follow a single
timescale to derive a unified global volcanic forcing series. We chose NS1-2011 as
the reference chronology for most of the volcanic time series because this age
model was constrained and validated by more stratigraphic age markers than
WD2014. WD2014 was used as the reference chronology only between 150
CE
and 450 CE, because of better data quality during that time period. TUNU2013
was not included in the Greenland ice-core composite because annual-layer
thickness variability at this site is influenced strongly by glaciological processes,
leading to relatively large uncertainties in atmospheric sulfur-deposition deter-
minations.
Northern Hemisphere tree-ring composite. Tree-ring records from certain
locations reflect summer cooling (as is widely observed after volcanic eruptions)
with no age uncertainty in annual ring-width dating, thus allowing independent
validation of ice-core timescales and the derived volcanic forcing indices.
However, no tree-ring-based temperature reconstructions of large spatial scales
span the full 2,500 years represented by our new ice-core chronologies. To thus
evaluate our new ice-core chronologies and assess the consistency of response
throughout the past 2,500 years, we compiled a composite (entitled ‘N-Tree’) of
multi-centennial tree growth records at locations where temperature is the lim-
iting growth factor. We selected available Northern Hemisphere tree-ring records
that provided a continuous record of .1,500 years and showed a significant
positive relationship with JJA temperatures during the instrumental period
(1901–2000
CE) with P , 0.005 (adjusted for a reduced sample size owing to
autocorrelation of the data sets). In total, five tree-ring chronologies (three based
on ring-width measurements, two based on measurements of maximum late-
wood density) met these criteria
42,43,76–78
of which three are located in the high
latitudes of Eurasia (Extended Data Fig. 1).
As various climatic and non-climatic parameters may influence sensitivity of
tree growth to temperatures during the twentieth century
79–81
, we used the time
period 1000-1099
CE as a common baseline for standardizing tree growth anom-
alies among the five chronologies and built a tree growth composite record called
N-Tree (z-scores) by averaging the individual records. Correlations between
N-Tree (N 5 5) and the average of three regional reconstructions for the
Arctic, Europe, and Asia (N . 275)
3
between 1800 CE and 2000 CE are very high
(r 5 0.86, N 5 201, P , 0.0001), suggesting that much of the large-scale variation
in temperature is explained by these selected tree-ring records. Three records in
N-Tree cover the period from 138
BCE to the present, thus allowing at least a
qualitative assessment of the coherence of growth reduction following large vol-
canic eruptions before the Common Era (Fig. 2, Extended Data Fig. 4).
Temperature reconstructions. To quantify the Common Era climate impact and
investigate regional differences, we used tree-ring-based JJA temperature recon-
structions covering the past 2,000 years with a demonstrated strong relationship
(r $ 0.45; P , 0.0001; Extended Data Fig. 1) to instrumental JJA temperature
data
82
between 1901 and 2000. For regions where this criterion was met by several
reconstructions (for example, Scandinavia), we limited the analysis to the most
recently updated reconstruction
35
. Three regional reconstructions from Central
Europe
42
, Northern Europe
35
, and Northern Siberia (Yamal, not shown)
76
as well
as a continental-scale reconstruction for Europe
3
met this criterion and were used
to quantify the average response of summer temperature to volcanic forcing
during the Common Era (Figs 3 and 4).
Superposed epoch analyses. To assess tree-ring growth reduction and summer
cooling following large eruptions, we used superposed epoch analyses
83,84
.
We selected all volcanic eruptions (28 events in total, 24
CE events) with time-
integrated volcanic forcing greater than 27.5 W m
22
(that is, eruptions larger
than Pinatubo 1991) and aligned the individual segments of N-Tree and regional
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

JJA temperature reconstructions relative to ice-core-indicated peak forcing. The
composite response was calculated for the average of the individual series (lag 0 to
lag 10 or 15 years) relative to the average values five years before individual
volcanic events (lag 25 to lag 21 year). 95% confidence intervals represent 2
s.e.m. of the tree-growth (Extended Data Fig. 4) and temperature anomalies
(Fig. 4) associated with the multiple eruptions.
Cryptotephra analyses of the 536
CE sample from NEEM-2011-S1. We analysed
samples from NEEM-2011-S1 for tephra between 326.73 m and 328.06 m depth,
corresponding to 531–539
CE (NS1-2011 timescale). Samples (200 g to 500 g)
were filtered, and elemental composition of recovered volcanic glass shards
determined by electron microprobe analysis at Queen’s University Belfast
using established protocols
63,67,85
and secondary glass standards
86,87
. Between
326.73 m and 327.25 m, large volume samples were cut at 8 cm depth resolution
(#0.5 years) and with an average cross-section of 26 cm
2
. Between 327.25 m and
328.06 m, the average cross-section was 7 cm
2
and depth resolution 20 cm (,1yr
resolution). Tephra particles (n $ 17) were isolated from a sample of ice (327.17–
327.25 m depth, 251 g) corresponding to the sulfate spike at 536
CE. The glass
shards were heterogeneous in size (20–80 mm), morphology (platey, blocky, vesi-
cular, microlitic), and geochemistry (andesitic, trachytic, rhyolitic). Individual
shards had geochemical compositions that share affinities with volcanic systems
in the Aleutian arc (Alaska)
88
, Northern Cordilleran volcanic province (British
Columbia)
89
, and Mono-Inyo Craters area (California)
90,91
—indicating at least
three synchronous eruptive events, all situated in western North America
between 38 uN and 58 uN (Extended Data Fig. 5; Supplementary Data 5).
Data and code availability. Ice-core data (chemistry, including sulphur and
10
Be), the resulting timescales, and the volcanic forcing reconstruction are
provided as Supplementary Data 1, and 3–5). Historical documentary data are
provided as Supplementary Data 2. The code for the StratiCounter program
is accessible at the github repository (http://www.github.com/maiwinstrup/
StratiCounter). NGRIP SO
4
data can be obtained at http://www.iceandclimate.
nbi.ku.dk/data/2012-12-03_NGRI P_SO4_5cm_Plummet_et_al_CP_2012.txt.
Tree-ring records and temperature reconstructions are from the Supplementary
Database S1 and S2 of the Pages-2k Consortium (ref. 3; http://www.nature.com/
ngeo/journal/v6/n5/full/ngeo1797.html#supplementary-information).
50. Dahl-Jensen, D. et al. Eemian interglacial reconstructed from a Greenland folded
ice core. Nature 493, 489–494 (2013).
51. McConnell, J. R. Continuous ice-core chemical analyses using inductively coupled
plasma mass spectrometry. Environ. Sci. Technol. 36, 7–11 (2002).
52. McConnell, J. R. & Edwards, R. Coal burning leaves toxic heavy metal legacy in the
Arctic. Proc. Natl Acad. Sci. USA 105, 12140–12144 (2008).
53. Pasteris, D. R. et al. Seasonally resolved ice core records from West Antarctica
indicate a sea ice source of sea-salt aerosol and a biomass burning source of
ammonium. J. Geophys. Res. 119, 9168–9182 (2014).
54. Abram, N. J., Mulvaney, R. & Arrowsmith, C. Environmental signals in a highly
resolved ice core from James Ross Island, Antarctica. J. Geophys. Res. 116,
D20116 (2011).
55. Kaufmann, P. R. et al. An improved continuous flow analysis system for high-
resolution field measurements on ice cores. Environ. Sci. Technol. 42, 8044–8050
(2008).
56. Bigler, M. et al. Optimization of High-Resolution Continuous Flow Analysis for
Transient Climate Signals in Ice Cores. Environ. Sci. Technol. 45, 4483–4489
(2011).
57. Ruth, U., Wagenbach, D., Steffensen, J. P. & Bigler, M. Continuous record of
microparticle concentration and size distribution in the central Greenland NGRIP
ice core during the last glacial period. J. Geophys. Res. 108 (2003).
58. Woodruff, T. E., Welten, K. C., Caffee, M. W. & Nishiizumi, K. Interlaboratory
comparisonof Be-10 concentrations in two ice cores from Central West Antarctica.
Nucl. Instrum. Meth. B 294, 77–80 (2013).
59. Berggren, A. M. et al. Variability of Be-10 and delta O-18 in snow pits from
Greenland and a surface traverse from Antarctica. Nucl. Instrum. Meth. B 294,
568–572 (2013).
60. Bisiaux, M. M. et al. Changes in black carbon deposition to Antarctica from two
high-resolution ice core records, 1850-2000 AD. Atmos. Chem. Phys. 12,
4107–4115 (2012).
61. Pasteris, D., McConnell, J. R., Edwards, R., Isaksson, E. & Albert, M. R. Acidity decline
in Antarctic ice cores during the Little Ice Age linked to changes in atmospheric
nitrate and sea salt concentrations. J. Geophys. Res. 119, 5640–5652 (2014).
62. Rasmussen, S. O. et al. A first chronology for the North Greenland Eemian Ice
Drilling (NEEM) ice core. Clim. Past 9, 2713–2730 (2013).
63. Coulter, S. E. et al. Holocene tephras highlight complexity of volcanic signals in
Greenland ice cores. J. Geophys. Res. 117, D21303 (2012).
64. Barbante, C. et al. Greenland ice core evidence of the 79 AD Vesuvius eruption.
Clim. Past 9, 1221–1232 (2013).
65. Clausen, H. B. et al. A comparison of the volcanic records over the past 4000 years
from the Greenland Ice Core Project and Dye 3 Greenland Ice Cores. J. Geophys.
Res. 102, 26707–26723 (1997).
66. Rolandi, G., Paone, A., Di Lascio, M. & Stefani, G. The 79 AD eruption of Somma: the
relationship between the date of the eruption and the southeast tephra dispersion.
J. Volcanol. Geotherm. Res. 169, 87–98 (2008).
67. Sun, C. Q. et al. Ash from Changbaishan millennium eruption recorded in
Greenland ice: implications for determining the eruption’s timing and impact.
Geophys. Res. Lett. 41, 694–701 (2014).
68. Xu, J. D. et al. Climatic impact of the millennium eruption of Changbaishan volcano
in China: new insights from high-precision radiocarbon wiggle-match dating.
Geophys. Res. Lett. 40, 54–59 (2013).
69. Deirmendjian, D. On volcanic and other particulate turbidity anomalies. Adv.
Geophys. 16, 267–296 (1973).
70. Vollmer, M. Effects of absorbing particles on coronas and glories. Appl. Opt. 44,
5658–5666 (2005).
71. Sachs,A. J. & Hunger, H. Astronomical Diaries and Related Texts from Babylonia Vol.3
Diaries from 164 B.C. to 61 B.C. (Verlag der O
¨
sterreichischen Akademie der
Wissenschaften, 1996).
72. Wittmann, A. D. & Xu, Z. T. A catalog of sunspot observations from 165 BC to AD
1684. Astron. Astrophys. (Suppl.) 70, 83–94 (1987).
73. Rasmussen, S. O. et al. A new Greenland ice core chronology for the last glacial
termination. J. Geophys. Res. 111, D06102 (2006).
74. Herron, M. M., Herron, S. L. & Langway, C. C. Climatic signal of ice melt features in
southern Greenland. Nature 293, 389–391 (1981).
75. Gao, C. H., Oman, L., Robock, A. & Stenchikov, G. L. Atmospheric volcanic loading
derived from bipolar ice cores: accounting for the spatial distribution of volcanic
deposition. J. Geophys. Res. 112, D09109 (2007).
76. Briffa, K. R. et al. Reassessing the evidence for tree-growth and inferred
temperature change during the Common Era in Yamalia, northwest Siberia. Quat.
Sci. Rev. 72, 83–107 (2013).
77. Grudd, H. Tornetrask tree-ring width and density AD 500-2004: a test of climatic
sensitivity and a new 1500-year reconstruction of north Fennoscandian summers.
Clim. Dyn. 31, 843–857 (2008).
78. Salzer, M. W., Bunn, A. G., Graham, N. E. & Hughes, M. K. Five millennia of
paleotemperature from tree-rings in the Great Basin, USA. Clim. Dyn. 42,
1517–1526 (2014).
79. McMahon, S. M., Parker, G. G. & Miller, D. R. Evidence for a recent increase in forest
growth. Proc. Natl Acad. Sci. USA 107, 3611–3615 (2010).
80. Salzer, M. W., Hughes, M. K., Bunn, A. G. & Kipfmueller, K. F. Recent unprecedented
tree-ring growth in bristlecone pine at the highest elevations and possible causes.
Proc. Natl Acad. Sci. USA 106, 20348–20353 (2009).
81. Briffa, K. R. et al. Reduced sensitivity of recent tree-growth to temperature at high
northern latitudes. Nature 391, 678–682 (1998).
82. Rohde, R. et al. A new estimate of the average land surface temperature spanning
1753 to 2011. Geoinform. Geostat. Overview
1, http://dx.doi.org/10.4172/2327-
4581.1000101 (2013).
83. Mass, C. F. & Portman, D. A. Major volcanic eruptions and climate: a critical
evaluation. J. Clim. 2, 566–593 (1989).
84. Fritts, H. C., Lofgren, G. R. & Gordon, G. A. Variations in climate since 1602 as
reconstructed from tree rings. Quat. Res. 12, 18–46 (1979).
85. Jensen, B. J. L. et al. Transatlantic distribution of the Alaskan White River Ash.
Geology 42, 875–878 (2014).
86. Oskarsson, N., Sigvaldason, G. E. & Steinthorsson, S. A dynamic-model of rift-zone
petrogenesis and the regional petrology of Iceland. J. Petrol. 23, 28–74 (1982).
87. Kuehn, S. C., Froese, D. G., Shane, P. A. R. & Participants, I. I. The INTAV
intercomparison of electron-beam microanalysis of glass by tephrochronology
laboratories: results and recommendations. Quat. Int. 246, 19–47 (2011).
88. Kaufman, D. S. et al. Late Quaternary tephrostratigraphy, Ahklun mountains, SW
Alaska. J. Quat. Sci. 27, 344–359 (2012).
89. Lakeman, T. R. et al. Holocene tephras in lake cores from northern British
Columbia, Canada. Can. J. Earth Sci. 45, 935–947 (2008).
90. Bursik, M., Sieh, K. & Meltzner, A. Deposits of the most recent eruption in the
Southern Mono Craters, California: description, interpretation and implications for
regional marker tephras. J. Volcanol. Geotherm. Res. 275, 114–131 (2014).
91. Sampson, D. E. & Cameron, K. L. The geochemistry of the Inyo volcanic chain—
multiple magma systems in the Long Valley region, eastern California. J. Geophys.
Res. 92, 10403–10421 (1987).
92. Veres, D. et al. The Antarctic ice core chronology (AICC2012): an optimized multi-
parameter and multi-site dating approach for the last 120 thousand years. Clim.
Past 9, 1733–1748 (2013).
93. Siebert, L., Simkin, T. & Kimberly, P. Volcanoes of the World 3rd edn, (University of
California Press, 2010).
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Figure 1
|
Location of study sites. a, Map showing locations
(blue circles) of the five ice cores (WDC, B40, NEEM, NGRIP and TUNU)
used in this study. Sites of temperature-limited tree-ring chronologies
(green)
42,43,76–78
and sites with annual D
14
C measurements from tree-rings in
the eighth century
CE (red outline) are marked. b, Metadata for the ice cores,
tree-ring width (RW), maximum latewood density (MXD) chronologies and
temperature reconstructions used
3,12,16,17,25,35,42,43,76,77,78,82
. m water equ. a
21
,
metres of water equivalent per year.
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Figure 2
|
Volcanic dust veils from historical documentary
sources in relation to NEEM. Time series of 32 independently selected
chronological validation points from well dated historical observations of
atmospheric phenomena with known association to explosive volcanism (for
example, diminished sunlight, discoloured solar disk, solar corona or Bishop’s
Ring, red volcanic sunset) as reported in the Near East, Mediterranean
region, and China, before our earliest chronological age marker at 536
CE. Black
lines represent the magnitude (scale on y axes) of annual sulfate deposition
measured in NEEM (NEEM and NEEM-2011-S1 ice cores) from explosive
volcanic events on the new NS1-2011 timescale. Red crosses depict the 24 (75%)
historical validation points for which NEEM volcanic events occur within a
conservative 63-year uncertainty margin. Blue crosses represent the eight
points for which volcanic events are not observed. The association between
validation points and volcanic events is statistically significantly non-random
at.99.9% confidence (P , 0.001). ppb, parts per billion.
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Figure 3
|
Timescale comparison. Age differences of the
timescales NS1-2011 and GICC05 for the NEEM-2011-S1/NEEM ice cores
(a) and WD2014 and WDC06A-7 for WDC (b). Differences before 86
CE (the
age of the ice that is now at the bottom of the ice core NEEM-2011-S1) deriving
from the annual-layer counting of the NEEM core are shown for major volcanic
eruptions relative to the respective signals in NGRIP on the annual-layer
counted GICC05 timescale. Marker events used for constraining the annual-
layer dating (solid line) and for chronology evaluation (dashed lines) are
indicated. Triangles mark volcanic signals. Also indicated is the difference
between WD2014 and the Antarctic ice-core chronology (AICC2012)
92
, based
on volcanic synchronization between the WDC and EDC96 ice cores.
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Figure 4
|
Post-volcanic suppression of tree growth.
Superposed epoch analysis for large volcanic eruptions using the 28 largest
volcanic eruptions (a); the 23 largest tropical eruptions (b); the five largest
Northern Hemisphere eruptions (c); and eruptions larger than Tambora 1815
with respect to sulfate aerosol loading (d). Shown are growth anomalies of a
multi-centennial tree-ring composite record (N-Tree) 15 years after the year
of volcanic sulfate deposition, relative to the average of five years before the
events. Dashed lines indicate 95% confidence intervals (2 s.e.m.) of the tree-ring
growth anomalies associated with the multiple eruptions.
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Figure 5
|
Major-element composition for ice core tephra
QUB-1859 and reference material. Shown are selected geochemistry data:
SiO
2
versus total alkali (K
2
O 1 Na
2
O) (a); FeO (total iron oxides) versus TiO
2
(b); SiO
2
versus Al
2
O
3
(c); and CaO versus MgO (d) from 11 shards extracted
from the NEEM-2011-S1 ice core at 327.17–327.25 m depth, representing
the age range 536.0–536.4
CE on the new, NS1-2011 timescale. Data for Late
Holocene tephra from Mono Craters (California) are from the compilation by
ref. 90; data for Aniakchak (Alaska) are from reference material published
by ref. 88; and data for the early Holocene upper Finlay tephra, believed to be
from the Edziza complex in the Upper Cordilleran Volcanic province (British
Columbia), are from ref. 89. (See Supplementary Information for the Upper
Finlay tephra.)
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Table 1
|
Ice-core dating
Parameters used for annual-layer interpretation. Parameters measured by the CFA system in the field are underlined. NH, Northern Hemisphere.
*Stratigraphic age marker used to constrain annual-layer counting.
{Horizons used to evaluate the timescale.
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Table 2
|
Annual-layer results using the StratiCounter program
Maximum-likelihood number of annual layers and confidence intervals derived from annual-layer counting between distinctive marker horizons and corresponding ages relative to the 775 CE
10
Be event. wrt, with
respect to.
*Unattributed events (UE) give volcanic signal and year of sulfate deposition based on final age models.
{Year calculated from the number of annual layers relative to the fixed age marker in 775
CE (negative numbers are years BCE).
{Depth has been estimated from the average depth offset between NEEM-2011-S1 and NEEM.
1Fixed age marker based on the
10
Be maximum annual value.
jjSection with 6-m gap in the NEEM 2011-S1 core DRI data (this section is not used for calculating average age).
"This section is based on the NEEM field CFA data, since the DRI data does not cover the entire interval.
#Section is based on combined data set of DRI and field-measured CFA data. The number of annual layers in this section from manual interpretation by investigator 1 was 383 (67), and that of investigator 2 was
393 (68) layers. Most of the difference between the three layer counts occurred below 480 m (before 300
BCE), where data gaps were more frequent.
qIndependent age markers used to constrain annual-layer dating in a second iteration to derive the final ice-core age model NS1-2011.
**Tephra particles were extracted from the depth range 327.17–327.25 m (see Supplementary Data).
{{Unattributed volcanic signal that was previously attributed to the historic 79
CE eruption of Vesuvius
64
.
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Table 3
|
Historical documentary evidence for key volcanic eruption age markers 536-939 CE
A comprehensive list of sources, including translations and assessment of the confidence placed in each source and its chronological information is given in Supplementary Data. NW, northwest.
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Table 4
|
Large volcanic eruptions during the past 2,500 years
Years with negative numbers are BCE. Tentative attribution of ice-core signals to historic volcanic eruptions is based on the Global Volcanism Program volcanic eruption database
93
. Average (summer) temperature
for the associated cold year is given for the average of Europe and the Arctic
3
. Volc., volcanic.
*Total global aerosol forcing was estimated by scaling the total sulfate flux from both polar ice sheets to the reconstructed total (that is, time integrated) aerosol forcing for Tambora 1815
7
(Methods); for high-
latitude Northern Hemisphere eruptions, Greenland fluxes were scaled by a factor of 0.57
6
.
{Unattributed volcanic events (UE) and tentative attributions for non-documented historic eruptions (?) are marked.
ARTICLE RESEARCH
G
2015 Macmillan Publishers Limited. All rights reserved

Extended Data Table 5
|
Post-volcanic cooling
Coldest years and decades (1–2000 CE, JJA temperature with respect to 1901–2000) for Europe
3
and years (500BCE–1250 CE) and decades (500BCE–2000 CE) with strong growth reduction in the N-Tree composite
(with respect to 1000–1099). Ages of the volcanic events from the ice cores reflect the start of volcanic sulfate deposition in Greenland (NS1-2011 timescale) with the largest 40 events indicated in bold letters and
tropical eruptions underlined. Years with negative numbers are before the Common Era (
BCE).
*Latewood frost ring in bristlecone pines within one year
34
.
RESEARCH ARTICLE
G
2015 Macmillan Publishers Limited. All rights reserved