Antarctic Climate Variability On Regional and Cont

Clim.
Past, 13, 1609–1634, 2017

https://doi.org/10.5194/cp-13-1609-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.
Antarctic climate variability on regional and continental scales

over the last 2000 years
Barbara Stenni1,2 , Mark A. J. Curran3,4 , Nerilie J. Abram5,6 , Anais Orsi7 , Sentia Goursaud7,8 ,
Valerie Masson-Delmotte7 , Raphael Neukom9 , Hugues Goosse10 , Dmitry Divine11,12 , Tas van Ommen3,4 ,
Eric J. Steig13 , Daniel A. Dixon14 , Elizabeth R. Thomas15 , Nancy A. N. Bertler16,17 , Elisabeth Isaksson11 ,
Alexey Ekaykin18,19 , Martin Werner20 , and Massimo Frezzotti21
1 Department of Environmental Sciences, Informatics and Statistics, Ca’ Foscari University of Venice, Venice, Italy
2 Institute for the Dynamics of Environmental Processes, CNR, Venice, Italy
3 Australian Antarctic Division, 203 Channel Highway, Kingston, Tasmania 7050, Australia
4 Antarctic Climate & Ecosystems Cooperative Research Centre, University of Tasmania, Hobart 7001, Australia
5 Research School of Earth Sciences, Australian National University, Canberra ACT 2601, Australia
6 ARC Centre of Excellence for Climate System Science, Australian National University, Canberra ACT 2601, Australia
7 Laboratoire des Sciences du Climat et de l’Environnement (IPSL/CEA-CNRS-UVSQ UMR 8212), CEA Saclay,
91191 Gif-sur-Yvette CEDEX, France

8 Université Grenoble Alpes, Laboratoire de Glaciologie et Géophysique de l’Environnement (LGGE),
38041 Grenoble, France

9 University of Bern, Oeschger Centre for Climate Change Research & Institute of Geography, 3012 Bern, Switzerland
10 Université catholique de Louvain, Earth and Life Institute, Centre de recherches sur la terre et le climat Georges Lemaître,
1348 Louvain-la-Neuve, Belgium

11 Norwegian Polar Institute, Fram Centre, 9296 Tromsø, Norway
12 Department of Mathematics and Statistics, Faculty of Science, University of Tromsø – The Arctic University of Norway,
9037, Tromsø, Norway

13 Department of Earth and Space Sciences, University of Washington, Seattle, WA 98195, USA
14 Climate Change Institute, University of Maine, Orono, ME 04469, USA
15 British Antarctic Survey, Cambridge, CB3 0ET, UK
16 Antarctic Research Centre, Victoria University of Wellington, Wellington 6012, New Zealand
17 National Ice Core Research Facility, GNS Science, Gracefield 5040, New Zealand
18 Arctic and Antarctic Research Institute, St. Petersburg, Russia
19 Institute of Earth Sciences, Saint Petersburg State University, St. Petersburg, Russia
20 Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, 27570 Bremerhaven, Germany
21 ENEA Casaccia, Rome, Italy
Correspondence to: Barbara Stenni ([email protected])
Received: 28 February 2017 – Discussion started: 22 March 2017

Revised: 15 September 2017 – Accepted: 20 September 2017 – Published: 17 November 2017
Published by Copernicus Publications on behalf of the European Geosciences Union.

1610 B. Stenni et al.: Antarctic climate variability on regional and continental scales over the last 2000 years
Abstract. Climate trends in the Antarctic region remain 1 Introduction

poorly characterized, owing to the brevity and scarcity of
direct climate observations and the large magnitude of in- Antarctica is the region of the world where instrumental cli-
terannual to decadal-scale climate variability. Here, within mate records are shortest and sparsest. Estimates of tempera-
the framework of the PAGES Antarctica2k working group, ture change with reasonable coverage across the full Antarc-
we build an enlarged database of ice core water stable iso- tic continent are only available since 1958 CE (Nicolas and
tope records from Antarctica, consisting of 112 records. We Bromwich, 2014), and the large magnitude of year-to-year
produce both unweighted and weighted isotopic (δ 18 O) com- climate variability that characterizes Antarctica makes the
posites and temperature reconstructions since 0 CE, binned interpretation of trends in this data-sparse region problem-
at 5- and 10-year resolution, for seven climatically distinct atic (Jones et al., 2016). As a result, the knowledge of past
regions covering the Antarctic continent. Following earlier Antarctic temperature and climate variability is predomi-
work of the Antarctica2k working group, we also produce nantly dependent on proxy records from natural archives.
composites and reconstructions for the broader regions of While coastal proxy records are being developed from terres-
East Antarctica, West Antarctica and the whole continent. trial and marine archives, information on Antarctic climate
We use three methods for our temperature reconstructions: above the ice sheet exclusively relies on the climatic inter-
(i) a temperature scaling based on the δ 18 O–temperature re- pretation of ice core records.
lationship output from an ECHAM5-wiso model simulation Within the variety of measurements performed in bore-
nudged to ERA-Interim atmospheric reanalyses from 1979 holes and ice cores, only water stable isotopes can provide
to 2013, and adjusted for the West Antarctic Ice Sheet re- subdecadal-resolution records of past temperature changes
gion to borehole temperature data, (ii) a temperature scaling (Küttel et al., 2012). In high-accumulation areas of coastal
of the isotopic normalized anomalies to the variance of the zones and West Antarctica, annual layer counting is feasible
regional reanalysis temperature and (iii) a composite-plus- during the last centuries to millennia (Plummer et al., 2012;
scaling approach used in a previous continent-scale recon- Abram et al., 2013; Thomas et al., 2013; Sigl et al., 2016;
struction of Antarctic temperature since 1 CE but applied to Winstrup et al., 2017, under review), and annual water stable
the new Antarctic ice core database. Our new reconstruc- isotope signals can be delivered. However, in the dry regions
tions confirm a significant cooling trend from 0 to 1900 CE of the central Antarctic plateau, where the longest ice core
across all Antarctic regions where records extend back into records are available, chronologies are less accurate and rely
the 1st millennium, with the exception of the Wilkes Land on the identification of volcanic deposits that can be used to
coast and Weddell Sea coast regions. Within this long-term tie ice cores from different sites to a common Antarctic ice
cooling trend from 0 to 1900 CE, we find that the warmest core age scale (Sigl et al., 2014, 2015).
period occurs between 300 and 1000 CE, and the coldest in- The chemical and physical signals measured in an individ-
terval occurs from 1200 to 1900 CE. Since 1900 CE, signif- ual ice core reflect a local climatic signal archived through
icant warming trends are identified for the West Antarctic the deposition and reworking of snow layers. The intermit-
Ice Sheet, the Dronning Maud Land coast and the Antarc- tency of Antarctic precipitation (Masson-Delmotte et al.,
tic Peninsula regions, and these trends are robust across the 2011; Sime et al., 2009), variability in precipitation source
distribution of records that contribute to the unweighted iso- regions (Sodemann and Stohl, 2009), and post-depositional
topic composites and also significant in the weighted tem- effects of snow layers including wind drift and scouring, sub-
perature reconstructions. Only for the Antarctic Peninsula limation, and snow metamorphism (Frezzotti et al., 2007;
is this most recent century-scale trend unusual in the con- Ekaykin et al., 2014; Touzeau et al., 2016; Casado et al.,
text of natural variability over the last 2000 years. How- 2016; Hoshina et al., 2014; Steen-Larsen et al., 2014) can
ever, projected warming of the Antarctic continent during distort the climate signal preserved within ice cores and pro-
the 21st century may soon see significant and unusual warm- duces non-climatic noise. As a result, obtaining a robust cli-
ing develop across other parts of the Antarctic continent. The mate signal can only be achieved through the combination
extended Antarctica2k ice core isotope database developed of multiple ice core records from a given site and/or region,
by this working group opens up many avenues for devel- and through the site-specific calibration of the relationships
oping a deeper understanding of the response of Antarctic between water stable isotopes and temperature.
climate to natural and anthropogenic climate forcings. The Water can be characterized by the stable isotope ratios of
first long-term quantification of regional climate in Antarc- oxygen (δ 18 O: the deviation of the ratio of 18 O/16 O in a sam-
tica presented herein is a basis for data–model comparison ple, relative to that of the standard, Vienna Standard Mean
and assessments of past, present and future driving factors of Ocean Water) and of hydrogen (δD: the deviation of the ra-
Antarctic climate. tio of 2 H/1 H). Both of these parameters within ice cores pro-
vide information on past temperatures. There is solid theo-
retical understanding of distillation processes relating mois-
ture transport towards the polar regions with air mass cool-
ing and the progressive loss of heavy water molecules along
Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

B. Stenni et al.: Antarctic climate variability on regional and continental scales over the last 2000 years 1611
the condensation pathway (Jouzel and Merlivat, 1984). This and produced reconstructions of the continent-wide temper-
theoretical understanding is further supported by numerical ature history as well as specific West Antarctica and East
modelling performed using atmospheric general circulation Antarctica reconstructions. The skill of the reconstructions
models equipped with water stable isotopes (Jouzel, 2014). was limited by the number of available records through time
The effects of these processes are observed in the spatial (for instance, only one predictor in each region prior to
relationships between the isotopic composition of Antarctic 166 CE). This analysis identified significant (p < 0.01) cool-
precipitation and surface snow and surface air temperature ing trends from 166 to 1900 CE, 2.5 times larger in West
across the continent. However, relationships between water Antarctica than in East Antarctica. A robust cooling trend
stable isotopes in snow and surface temperature may vary over this time period has also been identified from terrestrial
through time as a result of changes between condensation and marine reconstructions from other regions (PAGES 2k
and surface temperature (in relation to changes in boundary Consortium, 2013; McGregor et al., 2015).
layer stability), changes in moisture origin and initial evap- The comparison of these first Antarctic 2k time series
oration conditions, changes in atmospheric transport path- with those from other regions obtained within the PAGES
ways, and changes in precipitation seasonality or intermit- 2k working groups identified three specificities: (i) recon-
tency (Masson-Delmotte et al., 2008). Investigations based structed Antarctic centennial variations did not correlate with
on the sampling of Antarctic precipitation have demonstrated those from other regions, (ii) the Antarctic region was the
that seasonal and inter-annual isotope vs. temperature slopes only one where a protracted cold period did not start around
are generally smaller than spatially derived relationships (van 1580 CE (iii) the Antarctic region was the only one where
Ommen and Morgan, 1997; Schneider et al., 2005; Stenni the 20th century was not the warmest century of the last
et al., 2016; Schlosser et al., 2004; Ekaykin et al., 2004; Fer- 2000 years. A recent effort to characterize Antarctic and sub-
nandoy et al., 2010). Moreover, emerging studies combin- Antarctic climate variability during the last 200 years also
ing the monitoring of surface water vapour isotopic compo- concluded that most of the trends observed since satellite
sition with the isotopic composition retained in surface snow climate monitoring began in 1979 CE cannot yet be distin-
and precipitation have revealed that snow–air isotopic ex- guished from natural (unforced) climate variability (Jones
changes during snow metamorphism affect surface snow iso- et al., 2016), and observed instrumental climate trends are
topic composition (Ritter et al., 2016; Casado et al., 2016; of the opposite sign to those produced by most forced cli-
Touzeau et al., 2016). It is not yet possible to assess the im- mate model simulations over the same post-1979 CE inter-
portance of such post-deposition processes for the interpre- val. The only exception to this conclusion was for changes
tation of ice core water stable isotope records, but they may in the Southern Annular Mode (SAM), the leading mode of
enhance the relationship between snow isotopic composition atmospheric circulation variability in the high latitudes of the
and surface temperature more than expected from the inter- Southern Hemisphere (SH), which has showed a significant
mittency of snowfall (Touzeau et al., 2016). Changes in ice and unusual positive trend since 1979 CE.
sheet height due to ice dynamics may also affect the surface While changes in the SAM have been related to the hu-
climate trends inferred from water stable isotope records; man influence on stratospheric ozone and greenhouse gases
however, this influence should be of second order over the (Thompson et al., 2011), major gaps remain in identifying
last 2000 year interval that is the focus of this study (Fe- the drivers of multi-centennial Antarctic climate variability.
gyveresi et al., 2011). For instance, the influence of solar and volcanic forcing on
As a result, the two key challenges to reconstruct past Antarctic climate variability remains unclear. This is due
changes in Antarctic temperature from ice core isotope to both the lack of observations and to the lack of confi-
records are (1) to develop methodologies to combine differ- dence in climate model skill for the Antarctic region (Flato
ent individual or stacked ice core records in order to deliver et al., 2013). Goosse et al. (2012) have used simulations from
regional-scale climate signals and (2) to quantify the temper- an intermediate complexity model to attribute the Antarc-
ature changes represented by water stable isotope variations. tic annual mean cooling trend from 850 to 1850 CE to vol-
Goosse et al. (2012) first calculated a composite of Antarc- canic forcing. Recent comparisons of climate model simula-
tic temperature simply by averaging seven standardized tem- tions with the PAGES 2k regional reconstructions have high-
perature records inferred from water stable isotopes using lighted greater model–data disagreement in the SH than in
a spatial isotope–temperature relationship for the last mil- the Northern Hemisphere (PAGES 2k–PMIP3 group, 2015;
lennium. The first coordinated effort to reconstruct Antarc- Abram et al., 2016); such disagreement could be due ei-
tic temperature during the last 2000 years (PAGES 2k Con- ther to model deficiencies or to large uncertainties in the re-
sortium, 2013) screened published ice core records for an- constructions, which were built on relatively small number
nual layer counting or alignment of volcanic sulfate records of records. Changes in ocean heat content and ocean heat
and overlap with instrumental temperature data (Steig et al., transport have likely contributed to the different temperature
2009), leading to the selection of 11 records. The recon- evolution at high southern latitudes compared to other re-
struction procedure used a composite-plus-scaling (CPS) ap- gions of the Earth (Goosse, 2017), and model-based stud-
proach similar to the methodology of Schneider et al. (2006) ies have suggested that circulation in the Southern Ocean
www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

may act to delay, by centuries, the development of sustained graphic control using volcanic markers (Sigl et al., 2014)
warming trends in high southern latitudes (Armour et al., and, whenever possible, a dating by annual layer counted
2016). Antarctic temperature reconstructions spanning the chronology. This last requirement is only possible in the
last 2000 years may help to better constrain the processes high-accumulation regions of West Antarctica, the Antarctic
and timescales by which natural and anthropogenic forcing Peninsula and coastal areas of East Antarctica. The inclusion
act to affect climate changes in the Antarctic region. of shorter records is designed to improve data coverage for
This motivates our efforts to produce updated Antarc- assessments of climatic trends in Antarctica during the past
tic temperature reconstructions. The previous continent-scale century. The 11 records included in the previous continent-
reconstruction (PAGES 2k Consortium, 2013), in which only scale reconstruction (PAGES 2k Consortium, 2013) relied
a limited number of records have been used, may mask im- on a highly precise chronological framework consisting of
portant regional-scale features of Antarctica’s climate evo- a common chronology, which used 42 volcanic events to
lution. Here we use an expanded paleoclimate database synchronize the records. Here, we use both high- and low-
of Antarctic ice core isotope records and new reconstruc- resolution records. Most of the records have a data resolution
tion methodologies to reconstruct the climate of the past ranging from 0.025 to 5 years (only three records have a reso-
2000 years, on a decadal scale and regional basis. Seven lution of > 10 years). Previous studies (Frezzotti et al., 2007;
distinct climatic regions have been selected: the Antarctic Ekaykin et al., 2014) have shown that post-depositional and
Peninsula, the West Antarctic Ice Sheet (WAIS), the East wind scouring effects, acting more effectively when the ac-
Antarctic Plateau and four coastal domains of East Antarc- cumulation rate is very low, limit our ability to obtain tem-
tica. This regional selection, which is supported by regional perature reconstructions at annual resolution in most of the
atmospheric RACMO2.3p2 model results (Thomas et al., interior of Antarctica. Because of this, in our regional recon-
2017; Van Wessem et al., 2014), is applied to both Antarc- structions we use 5-year-averaged data for reconstructing the
tic ice-core-derived isotopic (temperature proxy) and snow last 200 years, and 10-year averages for reconstructing the
accumulation rate reconstructions (see companion paper in last 2000 years. Using 5- or 10-year averages also decreases
the same issue by Thomas et al.). Section 2 describes the our dependence on an annually precise chronological con-
ice core and the temperature data sets used in this study, as straint between the ice core records, allowing us to more con-
well as the modelling framework used to support the anal- fidently use the expanded database. The data have been also
ysis. The climate region definition, the preprocessing of the screened for glaciological problems, with those records that
data and the different reconstruction methods are presented are very likely to be affected by ice flow dynamics excluded.
in Sect. 3. Section 4 discusses our new regional isotopic and This enlarged database consists of 112 isotopic records.
temperature reconstructions for Antarctica, including the ap- A list of the records used are reported in Table S1 (in the
plication of the previous methodology to the new database. Supplement) and their spatial distribution is shown in Fig. 1.
Finally, Sect. 5 presents the summary of our results and their Figure S1 shows the location of the ice core sites along with
implications. a visualization of the record lengths. Most of the records of
this new database cover the last 200 years and this is particu-
larly true for the more coastal areas. Within the database, 36
2 Data sets
records cover just the last 50 years or less, while 50 records
2.1 Ice core records
cover the whole length of the past 200 years. There are 15
records that cover the last 1000 years, while only nine records
Here we present and use a new expanded database that has reach as far back as 0 CE.
been compiled in the framework of the PAGES Antarc-
tica2k working group. The initial selection criteria are 2.2 Temperature product
those requested by the PAGES 2k network (http://www.
pages-igbp.org/ini/wg/2k-network/data) for the building of The instrumental record is very short in Antarctica, and most
the community-sourced database of temperature-sensitive ice core sites do not have weather station measurements asso-
proxy records (PAGES 2k Consortium, 2017). Briefly, (i) the ciated with the cores. In addition, the retrieval of the first me-
records must be publicly available and published, (ii) a re- tre of firn can be difficult, due to poor cohesion of the snow.
lation between the climate proxies and variables should be As a result, for many sites, there is no overlap between in-
stated, (iii) the record duration should be between 300 and strumental and proxy data, which complicates the proxy cal-
2000 years, (iv) the chronology, certified by the data owner, ibration exercise. To enlarge the calibration data set, we use
should contain at least one chronological control point near the climate field reconstruction from Nicolas and Bromwich
the end (most recent) part of the record and another near the (2014) (hereafter NB2014; http://polarmet.osu.edu/datasets/
oldest part of the record and (v) the resolution should be at Antarctic_recon/). This surface temperature data set provides
least one analysis every 50 years. homogeneous data at 60 km resolution, extends from 1957
In building the Antarctica2k database we also allow to 2013 and includes the revised Byrd temperature record
shorter records to be included, although we request a strati- (Bromwich et al., 2013) that improves the skill of the tem-

Figure 1. (a) Schematic map of the seven Antarctic regions selected for the regional reconstructions. In blue is the East Antarctic Plateau, in
light blue the Wilkes Land Coast, in green the Weddell Sea Coast, in yellow the Antarctic Peninsula, in orange the West Antarctic Ice Sheet,
in red the Victoria Land Coast–Ross Sea and in brown the Dronning Maud Land Coast. The dots show the site locations. The white dots
represent the sites that have been used in the previous continent-scale reconstruction (PAGES 2k Consortium, 2013). (b) Correlation maps
between the regional mean temperature and each grid point using the Nicolas and Bromwich (2014) data set.
perature product over West Antarctica. It covers a longer 2012 atmospheric river event in Greenland has demonstrated
time span than reliable atmospheric reanalysis products for the skill of ECHAM5-wiso to reproduce these events, with
Antarctica (which begin only in 1979 CE) and has a higher a good representation of the water isotope signature (Bonne
spatial resolution than available isotope-enabled general cir- et al., 2014). In Antarctica, model performance was assessed
culation model (GCM) outputs. This data set is used to es- against a compilation of surface data (Masson-Delmotte
timate the spatial representativeness of individual core sites et al., 2008) and recent measurements of vapour and precipi-
(Sect. 3.3.2), to scale the normalized isotopic anomaly data tation (Ritter et al., 2016; Dittmann et al., 2016).
to temperature (Sect. 3.4.2) and to calculate the surface tem- Here, we use a 1958–2014 CE simulation in which
perature reconstructions with the CPS method (Sect. 3.4.4). ECHAM5-wiso was nudged to atmospheric reanalyses from
ERA-40 (Uppala et al., 2005) and ERA-Interim (Dee et al.,
2011) and run using the same ocean surface boundary con-
2.3 Modelling framework
ditions (sea surface temperature and sea ice) as in ERA-40
In order to use model information on isotope–temperature and ERA-Interim. Ocean surface water isotopic values were
relationships in Antarctic precipitation, we use a reference set to constant values using a compilation of observational
simulation performed using the general atmospheric circu- data (Schmidt et al., 2007). Inter-comparisons of reanalysis
lation model ECHAM5-wiso. The initial ECHAM5 model products showed good skills of ERA-Interim for Antarctic
(Roeckner et al., 2003) has been equipped with water sta- precipitation (Wang et al., 2016), surface temperature, and
ble isotopes (Werner et al., 2011), following earlier work on vertical profiles of winds and temperatures. However, com-
ECHAM3 (Hoffmann et al., 1998) and ECHAM4 (Werner parisons with in situ observations reveal an underestimate of
et al., 2001), and accounting for fractionation processes dur- precipitation and a slight cold bias in the surface tempera-
ing phase changes. This model is used here because re- tures in some regions (Thomas and Bracegirdle, 2015).
cent studies, based on model–data comparisons using ob- The ECHAM5-wiso simulations produce a large increase,
servations of precipitation and surface vapour isotopic com- which is not observed in instrumental or ice core data, in the
position on a global scale and in the Arctic (e.g. Werner temperature and the δ 18 O outputs prior to 1979 (Goursaud
et al., 2011; Steen-Larsen et al., 2017) have shown strong et al., 2017). This arises from a discontinuity in the ERA-
model skill of ECHAM5-wiso when it is run in high reso- 40 reanalyses due to the lack of observations available for
lution as in this study (T106, with a mean horizontal grid assimilation and boundary conditions prior to the satellite era
resolution of approximately 1.1◦ × 1.1◦ ). A study of the (e.g. Antarctic sea ice) (Nicholas and Bromwich, 2014). We

therefore use the ECHAM5-wiso simulations only for 1979– 3. Weddell Sea Coast extends eastward from longitude 60
2013 CE. to 30◦ W, and south of 75◦ S, and lies at an altitude
For the analysis of the isotope–temperature relationships < 2000 m. Eastward of the 30◦ W longitude, the 75◦ S
at each individual ice core site, we extracted the grid point latitude defines the boundary with the Dronning Maud
data closest to each site. For the analysis of isotope relation- Land coast region, with the northeastern corner of the
ships on a regional scale, we calculated the area-weighted Weddell Sea Coast region occurring where the 75◦ S lat-
average of model outputs at grid points within the region. itude meets the 2000 m elevation contour. This region
The δ 18 O–temperature relationship was calculated using the includes the Filchner Ice Shelf and most of the Ronne
annual or seasonal average 2 m temperature and annual Ice Shelf.
precipitation-weighted δ 18 O, to mimic deposition processes.
The simulation does not account for post-deposition pro- 4. Antarctic Peninsula encompasses the mountainous
cesses (i.e. diffusion, which is not important on the 5- and Antarctic Peninsula. Between 74 and 70◦ S the longitu-
10-year timescales considered here; e.g. Küttel et al., 2012). dinal boundaries lie between 60 and 80◦ W, while north
of 70◦ S the longitudinal boundaries increase to 50–
80◦ W so as to capture the northern end of the peninsula.
3 Methodology
5. West Antarctic Ice Sheet is bounded by longitudes 60 to
3.1 Defining climatic regions 170◦ W, and north of 85◦ S. In the Antarctic Peninsula
region (60–80◦ W) a northern bound of 74◦ S is also ap-
Earlier work of the PAGES Antarctica2k working group pro-
plied.
duced a continent-scale temperature reconstruction for the
whole of Antarctica, as well as reconstructions for East and 6. Victoria Land Coast–Ross Sea is north of 85◦ S and at
West Antarctica based on a separation approximated by the an altitude < 2000 m, with the exception of some local-
Transantarctic Mountains (PAGES 2k Consortium, 2013). ized peaks within the Transantarctic Mountains. It ex-
These broad-scale groupings mask important regional cli- tends from 160 to 190◦ E (i.e. 170◦ W) and incorporates
matic trends noted in individual studies. In particular, the most of the Ross Ice Shelf.
absence of recent significant warming in the Antarctica2k
continent-scale temperature reconstruction is known to not 7. Dronning Maud Land Coast extends eastward from
be representative of all Antarctic locations (e.g. Steig et al., 30◦ W to 67◦ E (Lambert Glacier). The southernmost
2009, 2013; Mulvaney et al., 2012; Abram et al., 2013). boundary lies at 75◦ S (where this region borders with
In this study, we choose seven climatic reconstruction re- the Weddell Sea Coast region), or at the 2000 m eleva-
gions (Fig. 1). These regions are defined based on our knowl- tion contour elsewhere.
edge of regional climate and snow deposition processes in the
In addition to these seven climatic regions, we also produce
Antarctic region, as well as the availability of ice core isotope
reconstructions for a continent-wide Antarctic region. Broad-
records. In particular, we separated coastal regions (below
scale East Antarctic (incorporating the climatic regions of the
2000 m altitude) from the East Antarctic Plateau: coastal sites
East Antarctic Plateau, as well as the Weddell Sea, Dronning
receive moisture from the high-latitude Southern Ocean and
Maud Land, Wilkes Land and Victoria Land coasts) and West
are affected by the nearby sea ice cover (Masson-Delmotte
Antarctic (incorporating the climatic regions of the West
et al., 2008). In contrast, high-altitude sites receive moisture
Antarctic Ice Sheet and Antarctic Peninsula) reconstructions
that has travelled at higher altitude, originating from further
are also presented. These additional reconstructions facili-
afield, and from clear sky precipitation (Ekaykin et al., 2004).
tate comparisons of our new results, using additional meth-
The regional selections were further validated and refined
ods and an expanded database, with earlier findings of the
by spatial correlation of temperature using the NB2014 data
Antarctica2k working group and subsequent research using
product. The seven climatic regions are defined as follows
the 2013 continent-scale reconstruction for Antarctic temper-
(see Table S1):
ature.
1. For East Antarctic Plateau, all East Antarctic contiguous
regions are at an elevation higher than 2000 m, includ- 3.2 Data preprocessing
ing everything south of 85◦ S. We exclude high peaks of
the Transantarctic Mountains if they belong to the Vic- All ice core records in the Antarctica2k database were as-
toria Land–Ross Sea coast (e.g. Taylor Dome or Her- signed to one of the climatic regions described in Sect. 3.1
cules Névé). (as well as to East vs. West Antarctica and to the Antarctic-
wide classifications). The majority (94 out of 112) of ice core
2. Wilkes Land Coast sits at an altitude < 2000 m and ex- water isotope records in the database are based on oxygen
tends from Lambert Glacier (67◦ E) east to the start isotope ratios (δ 18 O). In cases in which only deuterium iso-
of Victoria Land and the Transantarctic Mountains tope (δD) data are available, the ice core time series were
(160◦ E). converted to an δ 18 O equivalent by dividing by 8, a value

that represents the slope of the global mean meteoric rela- within the same grid cell, their isotopic anomalies (or nor-
tionship of oxygen and deuterium isotopes in precipitation malized data) are averaged to produce a single composite
and is close to the ratio of 7.75 observed in surface Antarctic time series for the grid. This replicates the simple unweighted
snow (Masson-Delmotte et al., 2008). compositing method described in Sect. 3.3.1, but on the grid
The ice core δ 18 O (and δ 18 O equivalent) records were scale to reduce the representation of data-rich areas prior to
compiled on a common annual average age scale. For records the regional compositing. Figures S2 and S3 show the distri-
with sub-annual resolution this involved averaging all data bution of records by region that meet the six-bin (30 years)
from within a calendar year to generate an annual average and nine-bin (90 years) minimum requirements for the 5- and
data set. Pseudo-annual records were generated for the ice 10-year composites, respectively, after the gridded data re-
core δ 18 O records with lower-than-annual resolution. These duction step.
pseudo annuals assume that each low-resolution isotopic
value represents an average of the full-time interval that the 3.3 Compositing methods
sample covers. As such, a nearest-neighbour interpolation
method was used to generate stepped (piecewise constant) We use a suite of reconstruction methods of varying com-
pseudo-annual records that continue the measured isotopic plexity in order to assess robust trends and variability in
value across all of the calendar years that it represents. Antarctic ice core δ 18 O records and temperature.
Records were next binned to 5- and 10-year-average res-
olution and converted to δ 18 O anomalies. This reduction in 3.3.1 Unweighted composites
resolution is designed to reduce the influence of small age un-
Our first reconstruction method involves calculating simple
certainties between the records, as well as the non-climatic
composites of δ 18 O anomalies. For each 5- or 10-year bin
noise induced by post-deposition (e.g. wind erosion, diffu-
we calculate the mean δ 18 O anomaly across all records in the
sion) processes (Frezzotti et al., 2007; Ekaykin et al., 2014).
climatic region, as well as the distribution of δ 18 O anoma-
The 5-year resolution records were converted to anomalies
lies within each bin (Figs. 2 and 3). This basic reconstruc-
relative to their mean over the 1960–1990 CE interval, and
tion method is analogous to that used for the Ocean2k low-
records that do not include a minimum of six bins (30 years)
resolution reconstruction (McGregor et al., 2015). The ben-
of coverage since 1800 CE are excluded based on length.
efit of this simple method is that it requires no weighting
Overall, 79 records in the new Antarctica2k water isotope
or calibration assumptions, which is advantageous for data-
database meet the minimum requirement of having at least
sparse regions such as Antarctica (and the global oceans).
30 years of data coverage since 1800 CE. In some cases,
The disadvantage is that it applies equal weighting to all
records meet this minimum length requirement but do not in-
records within a climatic region, which may introduce biases
clude data for the full 1960–1990 CE reference interval. We
related to record length, location and climatic skill.
adjust the mean value of each of these records by match-
ing the mean δ 18 O of their most recent six bins (30 years)
of data to the mean of all anomaly records from the same 3.3.2 Weighted composites based on site-level
climatic region and over the same six-bin interval. We also temperature regressions
produce normalized records by adjusting the variance in the In order to avoid biases from uneven data sampling, we per-
records using the same reference period and method as for formed, for each region, a multiple regression between each
the anomaly records. site temperature and the relevant regional average temper-
The 10-year resolution anomaly and normalized records ature (Figs. 2 and 3). Most of the ice core records do not
were generated using the same method but using a reference cover the full instrumental period; thus, it is problematic to
period of 1900–1990 CE and a minimum data coverage of use the δ 18 O anomalies directly to determine the regression
nine bins (90 years) since 0 CE. Similarly, records that do vectors required for a weighted temperature reconstruction.
not include the full 1900–1990 CE reference period have the Instead, we use the climate field reconstruction of NB2014
mean of their most recent nine-bin (90 years) interval ad- to estimate the weights: the annual mean temperature time
justed to match the mean of all other anomaly records from series at the grid cell corresponding to each ice core site is
the same region and over the same nine-bin interval. Overall, extracted from the NB2014 product, and the regional average
67 records in the new Antarctica2k water isotope database is also calculated for each reconstruction region. Regression-
meet the minimum requirement of having at least 90 years of based weightings are calculated based on the relationship be-
data coverage since 0 CE. tween site temperature and regional average temperature, and
Some regions of Antarctica, such as coastal and plateau the regression is performed for each combination of ice core
sites in Dronning Maud Land, include dense networks of ice records through time. The weights are then applied to the ice
core data (Fig. S1). To reduce possible bias towards these core δ 18 O anomalies to produce regional, averaged, standard-
data-rich regions, an additional data reduction method was ized anomalies.
used based on a 2◦ latitude by 10◦ longitude grid. Where
multiple ice core records from the same climatic region fall

Figure 2. Regional δ 18 O composite reconstructions over the last 200 years using 5-year-binned anomaly data. Both unweighted composites
and weighted composites (using both NB2014 temperature and ECHAM δ 18 O weighting methods) are shown. For each 5-year bin of the
unweighted data, the mean δ 18 O anomaly across all records in the climatic region, as well as the distribution of δ 18 O anomalies within
each bin, is calculated. All anomalies are expressed relative to the 1960–1990 CE interval. The number of records that contribute to the
reconstructions for each region are displayed in the lower panel.
We further reproduced this regression method using both tion, snow drift, snow–air exchanges, snow metamorphism,
the δ 18 O and temperature field outputs from the ECHAM5- diffusion in ice cores). It is not possible to consider each pro-
wiso experiments. The regression of site δ 18 O compared cess independently because in many cases there are simply
to regional average δ 18 O, or site temperature onto regional no observations to constrain them well enough. However, the
temperature, gave nearly identical weighting factors, sup- atmospheric circulation often leads to several processes to
porting the use of the temperature field to calculate re- be correlated (reduced sea ice, increased precipitation and
gressions. The effects of the different weighting methods warmer temperature, for instance). Here, we follow the clas-
on each regional isotopic composite, as well as the ini- sical approach, which is to perform a linear regression of ice
tial 10-year isotopic anomaly records, are reported in the core δ 18 O with local surface temperature on the regional av-
Supplement (Figs. S4–S10). The small differences between erage products. This method has the advantage of looking at
ECHAM- and NB2014-based regressions were due to the all the climatic processes influencing δ 18 O in “bulk”, and the
lower resolution of ECHAM5-wiso, which does not include use of regional average allows us to limit the influence of
islands and topographic features such as Roosevelt Island small-scale processes.
and Law Dome. For this reason, we preferentially use the The lack of an overlap period between our site δ 18 O
NB2014 data set for the temperature regression reconstruc- records and direct temperature observations makes the proxy
tion method. calibration difficult. The CPS method (Sect. 3.4.4), which
replicates the 2013 PAGES 2k reconstruction method, is lim-
ited to sites where this calibration is possible. To overcome
3.4 Temperature reconstructions this limitation and include the largest number of records,
we also use models to scale the regional isotope compos-
The relationship between δ 18 O and local surface tempera- ites. A first method uses ECHAM5-wiso to determine the re-
ture is complicated by the influence of a large range of pro- gional δ 18 O–temperature relationship in a mechanistic way
cesses (origin of moisture sources, intermittency in precipita-

Figure 3. Regional δ 18 O composite reconstructions over the last 2000 years using 10-year-binned anomaly data. Both unweighted compos-
ites and weighted composites (using both NB2014 temperature and ECHAM δ 18 O weighting methods) are shown. For each 10-year bin of
the unweighted data, the mean δ 18 O anomaly across all records in the climatic region, as well as the distribution of δ 18 O anomalies within
each bin, is calculated. All anomalies are expressed relative to the 1900–1990 CE interval. The number of records that contribute to the
reconstructions for each region are displayed in the lower panel.
(Sect. 3.4.1). A second method uses a more statistical ap- gional ice core composites (δ 18 Oregion anomalies) to scale
proach and simply scales the normalized record to the instru- them from δ 18 O to temperature units (Figs. 4 and 5) and
mental period temperature variance (Sect. 3.4.2). Both ap- produce temperature anomalies (Tregion ). The correlation co-
proaches are equally valid and share the same hypothesis: efficient αregion is calculated using the York et al. (2004)
that the instrumental period (1979–2013) is representative method, taking into account uncertainties both in Tregion and
of the longer-term climate variability. Finally, for the West in δ 18 Oregion , with each prior uncertainty equal to 20 % of the
Antarctic Ice Sheet region, an independent longer-term tem- variance.
perature record is available from borehole temperature mea- X
surements (Orsi et al., 2012). We use this independent tem- Tregion = αregion δ 18 Oregion = αregion wi δ 18 Oi
sites i
perature record to scale the long term 1000–1600 CE temper-
ature trend for the West Antarctic Ice Sheet region to provide In this equation, wi represents the weights assigned to each
our best estimate of temperature change in line with current site i, and δ 18 Oi the site δ 18 O anomalies in 5- or 10-year-
knowledge (Sect. 3.4.3). averaged records. The limited length of the observational pe-
In the figure captions, we refer to the different methods as riod (1979–2013 CE) does not allow us to precisely estimate
“ECHAM”, “NB2014” and “borehole”, respectively. the slope α on 10-year averages, and we preferred to use
1-year anomalies, for which the slopes are significant (Ta-
ble 1), and apply these slopes to the 10-year-binned com-
3.4.1 Scaling using model-based regional
posites. This implies that the interannual δ 18 O–temperature
δ 18 O–temperature relationships
relationship comes from mechanisms that are also applica-
We use the coherent physical framework of the 1979– ble to decadal-scale variability. It is impossible to further
2013 CE simulation performed at T106 resolution with test this hypothesis without longer independent temperature
ECHAM5-wiso to infer constraints on regional δ 18 O– records. The use of the ECHAM5-wiso isotope-enabled cli-
temperature slopes through linear regression analysis be- mate model is the most up-to-date tool we have to quantify
tween regional averages of simulated annual mean temper- the δ 18 O–temperature on broad spatial–temporal scales and
ature and precipitation-weighted δ 18 O (Table 1). These re- is our best tool to infer the δ 18 O–temperature relationship in
gional δ 18 O–temperature regressions were applied to the re- the absence of data. Its main limitation is the model resolu-

Figure 4. Regional temperature (T anomalies in ◦ C, referenced to the 1960–1990 CE interval) reconstructions using 5-year-binned data
for the past 200 years. The weighing method is based on the correlation between site T and regional T from NB2014. The temperature
scaling method is (i) based on the correlation between annual mean regional δ 18 O and regional T from ECHAM5-wiso forced by ERA-
Interim (coloured lines) and (ii) scaled on the NB2014 target over 1960–1990 CE (grey lines), and the (iii) West Antarctic Ice Sheet region
is adjusted to match the temperature trend between 1000 and 1600 CE based on borehole temperature measurements (black line; Orsi et al.,
2012). Linear trends are calculated over the last 100 years of the reconstructions (colours match associated reconstruction methods). The
map at the centre reports regional trends over the last 100-year trend using 5-year data based on the ECHAM method adjusted for the West
Antarctic Ice Sheet region to borehole data. Hatched areas are not significant (p > 0.05).
Table 1. Linear regression analysis (slope with ±1σ uncertainty, correlation coefficient r, and p value) of the simulated δ 18 O–temperature
relationships extracted from the ECHAM5-wiso model for each climatic region, as well as the broad East Antarctic, West Antarctic and
whole of Antarctica regions.
Geographic region Slope (◦ C ‰−1 ) Slope (‰ ◦ C−1 ) r p value

1. East Antarctic Plateau 0.95 ± 0.05 1.05 ± 0.06 0.62 0.0001
2. Wilkes Land Coast 1.91 ± 0.11 0.52 ± 0.03 0.44 0.0084
3. Weddell Sea Coast 1.01 ± 0.06 0.99 ± 0.06 0.34 0.0449
4. Antarctic Peninsula 2.50 ± 0.15 0.40 ± 0.02 0.31 0.0658
5. West Antarctic Ice Sheet 1.04 ± 0.06 0.96 ± 0.05 0.59 0.0002
6. Victoria Land Coast 0.83 ± 0.05 1.21 ± 0.07 0.49 0.0027
7. Dronning Maud Land Coast 1.08 ± 0.06 0.93 ± 0.05 0.39 0.0217
West Antarctica 1.03 ± 0.06 0.97 ± 0.05 0.62 0.0001
East Antarctica 1.00 ± 0.05 1.00 ± 0.05 0.58 0.0002
All Antarctica 1.02 ± 0.06 0.98 ± 0.05 0.56 0.0004
tion: it is missing some coastal topographical features, no- and cannot faithfully represent regions where these sites are
tably James Ross Island, Roosevelt Island, and Law Dome, important.

Figure 5. Regional temperature (T anomalies in ◦ C, referenced to the 1900–1990 CE period) reconstructions using 10-year data for the
past 2000 years. Weighing method based on the correlation between site T and regional T from NB2014 forced by ERA-Interim. The
temperature scaling method is (i) based on the correlation between annual mean regional δ 18 O and regional T from ECHAM5-wiso forced
by ERA-Interim (coloured lines) and (ii) scaled on the NB2014 target over 1960–2010 CE (grey lines), and the (iii) West Antarctic Ice Sheet
region is adjusted to match the temperature trend between 1000 and 1600 CE based on borehole temperature measurements (black line; Orsi
et al., 2012). Linear trends are calculated over the period 0–1900 CE using 10-year-binned data. The map at the centre reports the trend values
calculated between 0 and 1900 CE using 10-year data based on the ECHAM method adjusted to borehole data for the West Antarctic Ice
Sheet region (the hatched area is not significant).
All regional δ 18 O–temperature relationships produced by 64◦ W) (slope = 0.63 ± 0.58 ‰ ◦ C−1 , r = 0.13, p = 0.03),
the ECHAM5-wiso output are statistically significant (at and the overall low δ 18 O–temperature slope is largely at-
95 % confidence) with the exception of the Antarctic Penin- tributable to model resolution. We expect that the ECHAM
sula. Weak correlations are also found for the Weddell Sea scaling will produce a temperature reconstruction with
Coast (r = 0.34). Stronger correlation coefficients are ob- a high-amplitude bias in the Antarctic Peninsula.
tained inland, for the larger-scale East and West Antarctic High slopes similar to the Victoria Land are identi-
sectors, and maximum values (r = 0.62) are identified for the fied in the inland East Antarctic Plateau, Weddell Sea
East Antarctic Plateau. Coast and West Antarctic Ice Sheet regions (1.05, 0.99 and
Similarly, the simulated regional δ 18 O–temperature slopes 0.96 ‰ ◦ C−1 , respectively), together with intermediate val-
are highest for Victoria Land (1.21 ‰ ◦ C−1 ) and lowest for ues in coastal Dronning Maud Land with a 0.93 ‰ ◦ C−1
the Antarctic Peninsula (0.40 ‰ ◦ C−1 ). This low slope for mean slope. On the scale of the whole Antarctic ice sheet,
the Antarctic Peninsula does not agree with the temporal the overall temporal slope is dominated by inland regions
δ 18 O–temperature relationship that has been reported for the and simulated at 0.98 ‰ ◦ C−1 . This analysis is more thor-
highly resolved James Ross Island ice core (0.86 ‰ ◦ C−1 ; oughly examined in a study comparing an isotopic data set
Abram et al., 2011), while it is similar to one reported from surface snow, snowfalls and ice cores (Sentia Goursaud,
for the Gomez ice core (0.5 ‰ ◦ C−1 ; Thomas et al., 2009) personal communication, 2017).
and precipitation samples collected at the O’Higgins Sta-
tion (0.41 ‰ ◦ C−1 ; Fernandoy et al., 2012). ECHAM finds
that the only site on the peninsula with a significant correla-
tion between δ 18 O and temperature is Bruce Plateau (66◦ S,

3.4.2 Scaling based on NB2014 variance updated ice core isotope data collection. The method is a sim-
ple CPS approach (Jones et al., 2009), updated from Schnei-
In addition, we used an independent method of scaling the der et al. (2006) and implemented similarly to Neukom
normalized δ 18 O anomalies to the SD (σ (T )) of the regional et al. (2014). We apply this method to all subregions de-
temperature from NB2014, over the 1960–1990 CE inter- fined above, the broad East and West Antarctic divisions and
val for the 5-year-binned averages, and the period 1960– to the Antarctic-wide database, replicating the Antarctic re-
2010 CE for the 10-year-binned averages. This scaling is constructions presented in the PAGES 2k Consortium (2013)
similar to the one used for the CPS method (see next sec- study.
tion). First, the annual-average records are allocated to the cli-
matic regions as defined above. The following steps are then
Tregion = σ (T )region δ 18 Oregion (normalized) repeated for each region separately. Second, only the records
with no missing values in the 1961–1991 CE calibration pe-
This scaling method implies that the δ 18 O–temperature re- riod are selected. These records are then scaled to mean zero
lationship can be inferred from the ratio of temperature to and unit SD over their common interval of data availability.
δ 18 O SD, which would be true if the relationship between Next, the normalized records are correlated with the NB2014
the two were purely linear. If some of the δ 18 O variance is regional mean temperatures over 1961–1991 CE. Between 0
due to something other than temperature, this scaling will and 33 % of the ice core records within each region have neg-
underestimate temperature variations. This method also as- ative correlations (physically implausible) with the target and
sumes that the last 30 to 50 years provide a good estimate are removed from the proxy matrix. A composite of the re-
of the 5- or 10-year temperature variance. In the absence of maining records is then calculated by creating a weighted av-
longer temperature reconstructions, this is the best estimate erage, where the weighting of each ice core record is based
of σ (T )region that we can provide. on its temperature correlation from the previous step. The
composite is then scaled to the mean and SD of the NB2014
3.4.3 Scaling based on borehole temperature for the regional temperatures over the 1961–1991 CE period. The
WAIS region. compositing and scaling steps are carried out with a nested
approach, i.e. repeated for all periods with different proxy
In the West Antarctic Ice Sheet region, the approaches de- data availability.
scribed above give different results (Figs. 4 and 5), with In the reconstruction of the PAGES 2k Consortium (2013),
the first method (temperature scaling from the δ 18 O–T re- three records were infilled with neighbouring sites to have no
lationship in ECHAM5-wiso) giving a smaller amplitude. At missing data in the calibration window: WDC06A was in-
WAIS Divide, there is an independent temperature record, filled with data from WDC05A, and Siple Station and Plateau
which can be used to scale the long-term temperature evo- Remote records were infilled by a least median of squares
lution. We used the borehole temperature reconstruction at multiple linear regression using nearby records (PAGES 2k
WAIS Divide (Orsi et al., 2012) to adjust the amplitude of Consortium, 2013). To allow comparison, we also used the
temperature variations, matching the cooling trend over the infilled data for these records. Thus, the only difference
period 1000–1600 CE (−1.1 ◦ C 1000 yr−1 ), with a correction from the reconstruction of PAGES 2k Consortium (2013) is
factor relating the WAIS Divide site to the West Antarctic re- that we use an extended proxy database with more records
gion (c = σWAIS /σWDC = 0.65), with σWAIS the SD of the and an updated temperature target (NB2014 instead of Steig
annual NB2014 WAIS regional mean data set, and σWDC the et al., 2009).
SD of the annual NB2014 time series at the WAIS Divide While this CPS approach allows a quantitative calibra-
site, which gives a 1000–1600 CE cooling trend of −0.76 ◦ C. tion to the NB2014 temperature data, it has some limitations
This scaling is actually in line with the other two scalings: compared to the methods above. First, in this implementa-
1000–1600 CE slope of −0.65 ◦ C for ECHAM and −1.01 ◦ C tion it allows only the inclusion of data covering the cali-
for NB2014 scaling. The temperature calibration presented bration period, thereby removing more than half of the avail-
here is the best estimate we can provide with current knowl- able records (62 out of 112). Second, the calibration period is
edge, but we expect it to be revised in the future, with more extremely short and therefore individual years (for example
precise δ 18 O modelling and more independent quantitative with outliers) can significantly bias the reconstruction, and
temperature reconstructions. reasonable validation of the reconstruction is hardly possible.
The main difference from the other compositing methods de-
3.4.4 Replication of the 2013 Antarctica2k scribed above is the weight of each record and the interval
reconstruction method over which the data are standardized.
To facilitate comparison with the preceding Antarctica2k

temperature reconstruction published by the PAGES 2k Con-
sortium (2013), we apply their reconstruction method to the

4 Results and discussion ture field (Fig. S8). The temperature reconstructions obtained
with the different methods are further shown in Fig. S15.
We use the varying reconstruction methods to identify robust We assess trends for the seven climatic regions (and in
trends in the Antarctic ice core database. We present results the larger-scale Antarctic groupings) for the reconstructions
based on isotopic trends, as well as temperature reconstruc- prior to 1900 CE (up to 1900 years length) (Sect. 4.1.2), and
tions, and examine these for the seven climatic regions and since 1900 CE (up to 110 years length) (Sect. 4.1.3), and we
for the larger-scale Antarctic regions (Sect. 4.1), compare our estimate the significance of the most recent 100-year trend
results to the previous Antarctic temperature reconstruction relative to natural variability (Sect. 4.1.4).
(Sect. 4.2), and investigate the link between temperature and
volcanic activity (Sect. 4.3).
4.1.1 Trend significance in unweighted composites
4.1 Regional-scale δ 18 O and temperature We first use a Monte Carlo approach to assess the signifi-
reconstructions cance of trends in the unweighted composites. This test is
designed to test the significance of trends in relation to the
For each of the seven climatic regions in Fig. S2 (unweighted distribution of data within each bin of the isotopic compos-
isotope anomalies) and Fig. 2 (weighted and unweighted ites. For each bin in which two or more ice cores contribute
data), 5-year-binned δ 18 O composite records since 1800 CE data, we scale random Gaussian data about the median value
are reported. The unweighted composites are shown with and ±2σ distribution of isotopic data within that bin. We
respect to the distributions of data within each bin and ex- then sample from this scaled Gaussian data to produce 10 000
pressed relative to the 1960–1990 CE interval. Figure 2 also simulations of each regional composite. We then assess the
shows the reconstructions that are obtained by weighting the proportion of ensemble members that produce trends of the
records within each region based on the NB2014 tempera- same sign as the mean composite and the proportion of en-
ture field and by the ECHAM5-wiso δ 18 O field. Figure 3 semble members for which the trend is of the same sign
(and Fig. S3) shows equivalent data, but for 10-year averages as the mean composite and also significant at greater than
since 0 CE relative to the 1900–1990 CE interval. 95 % confidence. These trend analyses are based on 10-year-
The highest density of ice core records is present in the last binned isotopic anomalies for trends prior to 1900 CE and
century, but these are not evenly distributed over Antarctica 5-year-binned data for trends over the last 100 years of the
(Fig. S1), with most of the records in the plateau and coastal composites (although equivalent results are found if 10-year-
areas of Dronning Maud Land and across the West Antarctic binned data are used to assess trends in the last 100 years).
Ice Sheet. Conversely, only one and three records are present Results of unweighted composite trend analysis are sum-
in the Weddell Sea and Wilkes Land coastal areas, respec- marized in Table 2. This analysis shows that for the un-
tively. weighted composites the long-term cooling trend from 0 to
In order to separate the uncertainties due to the stacking 1900 CE is only significant for the East Antarctic Plateau.
procedure from uncertainties in the temperature scaling, we Visual examination of the unweighted composites (Fig. 3)
first discuss the main features of the unweighted regional suggests that many of the other climate regions appear to
δ 18 O anomalies (Sect. 4.1.1) and then proceed to discuss the also have a negative unweighted isotopic trend over part of
weighted regional δ 18 O anomalies, and finally the temper- the last 2000 years (e.g. Wilkes Land and West Antarctic Ice
ature reconstructions. The weighted and unweighted com- Sheet regions), but these trends are not significant in the un-
posites produce similar results for the seven climatic regions weighted composites when calculated across the full interval
(see Figs. S4–S10 in the Supplement), suggesting that our from 0 to 1900 CE. The Victoria Land Coast trend prior to
reconstructions are robust, and the composites do not de- 1900 CE is not significant in the mean or median unweighted
pend on the exact methodology used (Figs. 2 and 3). There composites, but over the interval in which two or more ice
are two exceptions to this: the Antarctic Plateau (Fig. S4), cores contribute to the composites the Monte Carlo assess-
where the many sites in Dronning Maud Land perhaps still ments indicate that negative isotopic trends are produced in
bias the simple average towards this area even with the grid- all 10 000 ensemble members and are significant (p < 0.05)
ded data-reduction process, and the West Antarctic Ice Sheet more often than can be explained by chance alone.
(Fig. S8), where the two long records from WAIS Divide The significant trend that is produced in the unweighted
and Roosevelt Island (RICE) have diverging trends for much composite for the East Antarctic Plateau climate region is
of the last 2000 years. The weighted reconstruction gives also evident in the broader East Antarctic compilation and
a higher weight to WAIS Divide, which maintains a long- for the Antarctic continent-scale composite. The continent-
term negative isotopic trend over the last 2 kyr in this region scale cooling trend produced in unweighted composites us-
(see Sect. 4.1.2). We further checked that WAIS Divide is ing the expanded Antarctica2k database is in agreement with
indeed more representative of temperatures averaged across the PAGES 2k Consortium (2013) results for which a long-
the West Antarctic Ice Sheet region than RICE, looking at term cooling trend over the Antarctic continent was identi-
temperature correlation maps that use the NB2014 tempera- fied. It is also consistent with robust pre-industrial cooling

Table 2. Summary of trend statistics for isotopic anomalies using unweighted composites. Trends are expressed as isotopic anomalies, in per
mille δ 18 O per decade units. Trends and their significance (p; based on the Student’s t statistic) are calculated using a linear additive model
and reported for the mean and median composites. Monte Carlo testing is applied to 10 000 ensemble members based on the unweighted
composite distributions, which are assessed to determine the percentage of trends with the same sign as the mean trend and the percentage
with the same sign and a significance of p < 0.05. Bold values indicate mean and median trends with a significance of p < 0.05 and Monte
Carlo simulations where significant; same-signed trends exceed 5 % of the ensemble.
Pre-1900 CE trends Last-100-year trends

Region Mean trend Median trend Monte Carlo Mean trend Median trend Monte Carlo
‰ 10 yr−1 (p) ‰ 10 yr−1 (p) % (%p < 0.05) ‰ 10 yr−1 (p) ‰ 10 yr−1 (p) % (%p < 0.05)
1. East Antarctic Plateau −0.003 (0.000) −0.002 (0.000) 100 (99.2) 0.011 (0.675) 0.040 (0.096) 96.2 (14.1)
2. Wilkes Land Coast 0.017 (0.444) 0.017 (0.444) 80.6 (2.3) 0.098 (0.001) 0.098 (0.001) 100 (90.5)
3. Weddell Sea Coast −0.002 (0.376) −0.002 (0.376) – −0.060 (0.318) −0.060 (0.318) –
4. Antarctic Peninsula 0.006 (0.188) 0.004 (0.361) 84.3 (2.9) 0.112 (0.000) 0.111 (0.000) 100 (98.2)
5.West Antarctic Ice Sheet 0.000 (0.435) 0.000 (0.428) 85.4 (1.4) 0.033 (0.226) 0.042 (0.091) 97.0 (14.4)
6. Victoria Land Coast −0.006 (0.118) −0.006 (0.101) 100 (12.76) −0.054 (0.220) −0.074 (0.102) 99.9 (13.6)
7. Dronning Maud Land Coast −0.032 (0.366) −0.027 (0.448) 95.3 (0.1) 0.147 (0.000) 0.158 (0.001) 100 (99.8)
West Antarctica 0.000 (0.437) −0.001 (0.118) 95.57 (11.2) 0.054 (0.021) 0.082 (0.001) 100 (79.8)
East Antarctica −0.003 (0.000) −0.002 (0.000) 100 (100) 0.037 (0.035) 0.064 (0.003) 99.8 (55.1)
All Antarctica −0.002 (0.000) −0.002 (0.000) 100 (99.5) 0.044 (0.005) 0.073 (0.000) 99.9 (76.2)
trends that have been identified in other continental recon- for coastal regions of Antarctica and the West Antarctic Ice
structions (PAGES 2k Consortium, 2013) and in the global Sheet. It should be stressed that these results are based only
oceans (McGregor et al., 2015). upon the simple unweighted compositing of the regional iso-
Isotopic trends in the last 100 years of the unweighted topic data, and the significance of trends is further assessed
composites show significant positive trends across a number in the following section after weighting of the individual ice
of regions. In particular, significant isotopic trends, indica- core records based on how representative they are of isotopic
tive of climate warming, are evident in the unweighted com- and temperature variability within each climatic region.
posites for the Antarctic Peninsula and the Wilkes Land and
Dronning Maud Land coasts. The West Antarctic Ice Sheet
region does not display a significant (p < 0.05) positive trend 4.1.2 Long-term trends in weighted reconstructions
in the mean or median of the unweighted composites, but To extend this trend analysis further, we next assess the pre-
the Monte Carlo analysis across the distribution of isotopic 1900 CE (Fig. 5) trends in the temperature reconstructions
data within each 10-year bin suggests that positive trends produced by scaling the ice core data based on our differ-
are produced in 99.5 % of the 10 000 simulations and are ent approaches (see Sect. 3.3.3). We estimate the uncertainty
significant (p < 0.05) more often than can be explained by of the slope based on the ±2σ uncertainty in the regression
chance alone (20.5 % of simulations). Similarly, the Victo- parameters. The robustness of the slope estimation to indi-
ria Land–Ross Sea region shows a negative but insignificant vidual data points was further checked by taking out 10 %
trend in the mean and median composites, but in the Monte of data points randomly and calculating the slope again, but
Carlo simulations this negative (cooling) trend is produced the uncertainty estimate this yields is much smaller than the
in 99.9 % of ensemble members and is significant in 13.6 % uncertainty based on regression parameters. The slopes ob-
of ensemble members. The apparent inverse isotopic trends tained by each of the temperature scaling methods are pre-
during the last century between the Victoria Land–Ross Sea sented in Table 3. The uncertainty in the amplitude of the
region and the West Antarctic Ice Sheet and Antarctic Penin- 0–1900 CE trend is dominated by the uncertainty in the tem-
sula regions may be indicative of dynamical processes in the perature scaling of the composite. To make the discussion
Amundsen Sea sector, which on interannual timescales are clear, we first focus on the trend in terms of normalized δ 18 O
known to cause opposing climate anomalies between these anomalies, with respect to the 1900–1990 CE periods, which
regions. circumvent the temperature scaling issues.
The PAGES 2k Consortium (2013) study concluded that The period 0–1900 CE exhibits significant negative trends
Antarctica was the only continent-scale region to not see the in most regions, from −0.4 to −1.3σ 1000 yr−1 . The
long-term cooling trend of the past 2000 years reverse to re- trend is largest in the West Antarctic Ice Sheet (−1.3 ±
cent significant warming. However, on the regional scale ex- 0.2σ 1000 yr−1 ), Victoria Land (−0.9±0.4σ 1000 yr−1 ) and
amined in this study recent significant warming is suggested the Antarctic Plateau (−0.8 ± 0.3σ 1000 yr−1 ) regions. It
by many of the unweighted isotopic composites, particularly is smaller but still significant for the Wilkes Land Coast

Figure 6. Histograms showing the distributions of all 100-year trends on normalized, weighted composites over the last 2000 years. Distribu-
tions are shown for each climatic region, as well as for East, West and whole of Antarctica composites, and are calculated on 10-year-binned
composites. The solid vertical lines represent the most recent 100-year trend in each reconstruction, and grey shading corresponds to the
5–95 % uncertainty range of the last 100-year trends. Only for the Antarctic Peninsula does the most recent 100-year trend stand out as
unusual compared to the natural range of century-scale trends over the last 2000 years.
Table 3. Trend analysis of the period 0–1900 CE (or shorter depending on maximum record length) for the various temperature scaling
approaches detailed in Sect. 3.4.
ECHAM NB2014 Borehole

Geographic region Start End Slope p value Slope p value Slope p value
date date (◦ C 1000 yr−1 ) (◦ C 1000 yr−1 ) (◦ C 1000 yr−1 )
1. East Antarctic Plateau 0 1900 −0.38 ± 0.14 < 0.0001 −0.32 ± 0.12 < 0.0001 NaN NaN
2. Wilkes Land Coast 170 1900 −0.24 ± 0.17 0.0072 −0.16 ± 0.13 0.0161 NaN NaN
3. Weddell Sea Coast 1000 1900 −0.24 ± 0.54 0.3763 −0.12 ± 0.27 0.3763 NaN NaN
4. Antarctic Peninsula 0 1900 −0.52 ± 0.23 < 0.0001 −0.20 ± 0.09 < 0.0001 NaN NaN
5. West Antarctic Ice Sheet 0 1900 −0.47 ± 0.10 < 0.0001 −0.92 ± 0.16 < 0.0001 −0.55 ± 0.11 < 0.0001
6. Victoria Land Coast 0 1900 −0.34 ± 0.18 0.0003 −0.29 ± 0.12 < 0.0001 NaN NaN
7. Dronning Maud Land Coast 1530 1900 5.96 ± 3.27 0.0007 0.97 ± 0.62 0.0032 NaN NaN
West Antarctica 0 1900 −0.24 ± 0.07 < 0.0001 −0.44 ± 0.10 < 0.0001 −0.55 ± 0.10 < 0.0001
East Antarctica 0 1900 −0.30 ± 0.10 < 0.0001 −0.32 ± 0.10 < 0.0001 NaN NaN
All Antarctica 0 1900 −0.36 ± 0.08 < 0.0001 −0.40 ± 0.08 < 0.0001 −0.26 ± 0.06 < 0.0001
Normalized CPS
Geographic region Slope p value Slope p value Start End
(σ 1000 yr−1 ) (◦ C 1000 yr−1 ) date date
1. East Antarctic Plateau −0.76 ± 0.28 0 −0.15 ± 0.06 0 10 1900
2. Wilkes Land Coast −0.59 ± 0.48 0.0161 −0.10 ± 0.07 0.008 180 1900
3. Weddell Sea Coast −0.41 ± 0.92 0.3763 −0.09 ± 0.27 0.4935 1000 1900
4. Antarctic Peninsula −0.50 ± 0.23 0 −0.09 ± 0.09 0.0479 0 1900
5. West Antarctic Ice Sheet −1.32 ± 0.23 0 −0.59 ± 0.08 0 0 1900
6. Victoria Land Coast −0.89 ± 0.39 0 −0.54 ± 0.58 0.0661 1140 1900
7. Dronning Maud Land Coast 4.98 ± 3.20 0.0032 NaN NaN 1890 1900
West Antarctica −0.76 ± 0.16 0 −0.55 ± 0.07 0 0 1900
East Antarctica −0.59 ± 0.19 0 −0.18 ± 0.06 0 10 1900
All Antarctica −0.76 ± 0.15 0 −0.38 ± 0.05 0 0 1900
(−0.6±0.5σ 1000 yr−1 , p = 0.007) and Antarctic Peninsula ing trend over the continent, which is comparable in ampli-
(−0.5 ± 0.2σ 1000 yr−1 ). It is insignificant on the Weddell tude to the variance over the past 100 years. As previously
Sea Coast (p = 0.4), and the data set is not long enough mentioned, the 0–1900 CE negative trend is largest in the
to estimate millennial-scale trends on the Dronning Maud normalized data sets for the West Antarctic Ice Sheet region,
Land Coast. These observations indicate a broad-scale cool- but this feature masks subregional-scale differences, with

the RICE record indicating increasing rather than decreasing the Antarctic Peninsula (+2.65σ 100 yr−1 with respect to
anomalies. This result is puzzling because both WAIS Divide the 1960–1990 CE normalization period), Dronning Maud
(to the west) and Victoria Land cores (e.g. Victoria Lower Land Coast (+2.51σ 100 yr−1 ) and the West Antarctic Ice
Glacier, to the east of RICE) show a clear cooling trend prior Sheet (+1.17σ 100 yr−1 ) regions. The trends in other re-
to 1900 CE, and together this points to a more complex pic- gions are not significant (Table 4). In temperature units, the
ture of the evolution of the atmospheric circulation around NB2014 method gives a scaling within 30 % of ECHAM,
the Ross Sea than can be captured with the current network while CPS is generally 50 % lower (Table 4). Since 1900 CE,
of cores. The RICE δ 18 O record, situated on Roosevelt Island the reconstructions indicate that the Antarctic Peninsula has
on the Ross Ice Shelf, is more influenced by air masses from been warming by 1.14–2.87 ◦ C 100 yr−1 , the West Antarc-
the eastern Ross Sea sector than the rest of the West Antarc- tic Ice Sheet by 0.46–1.32 ◦ C 100 yr−1 and the Dronning
tic Ice Sheet region, which is influenced predominantly by air Maud Land Coast by 0.59 to 1.33 ◦ C 100 yr−1 . The bore-
masses from the Amundsen Sea (Neff et al., 2015; Emanuels- hole temperature adjustment needed to match the long-term
son et al., 2017). Moreover, borehole temperature and δ 15 N trend leads to an overestimation of the 100-year trend in
thermal fractionation records at RICE highlight some notable the WAIS region. Indeed, the same borehole temperature
differences in the isotope temperature reconstruction, which record finds a warming trend of 0.70 ◦ C 100 yr−1 for the past
suggest that sea ice extent exerts an important control, per- 100 years and 1.22 ◦ C 50 yr−1 for the past 50 years. This ex-
haps masking aspects of the longer-term temperature trends ample shows that a simple linear scaling cannot be valid for
in the region (Bertler et al., 2017). all timescales and that another approach will be needed to
Over the period from 0 to 1900 CE the ECHAM tem- improve quantitative temperature reconstructions for Antarc-
perature scaling suggests that the mean cooling trends are tica.
−0.38 ◦ C 1000 yr−1 for the East Antarctic Plateau, −0.47 ◦ C Despite these uncertainties on absolute scaling, our anal-
1000 yr−1 for the West Antarctic Ice Sheet, −0.52 ◦ C yses underline that the sustained warming of the Antarctic
1000 yr−1 for the Antarctic Peninsula and −0.34 ◦ C Peninsula over the last century stands out as being a ro-
1000 yr−1 for Victoria Land. Coastal East Antarctic regions bust feature across all methods. Moreover, while the West
do not show significant trends (Weddell Sea, Dronning Maud Antarctic Ice Sheet and the Antarctic Peninsula regions have
Land Coast and Wilkes Land Coast). The NB2014 temper- now seen reversal of the long-term cooling trend of the past
ature scaling produces lower trends overall (Table 3), ex- 2000 years, this is not the case for the rest of the continent,
cept for the West Antarctic Ice Sheet where the trend dou- where temperatures changes over the last century have not
bles (−0.92 ◦ C 1000 yr−1 ). The trend in the West Antarc- been significant (Fig. 4, Table 4).
tic Ice Sheet region can be assessed independently by com-
paring the WAIS composite to the borehole temperature 4.1.4 Significance of most recent 100-year trends
record of Orsi et al. (2012) at WAIS Divide for the pe- relative to natural variability
riod 1000–1600 CE. The borehole-derived trend at WAIS Di-
vide is −0.55 ◦ C 1000 yr−1 for the 0–1900 CE period, in be- Finally, we assess how significant the trends in the most re-
tween the −0.92 ◦ C 1000 yr−1 value found with the NB2014 cent 100 years of the regional temperature reconstructions
method and the −0.47 ◦ C 1000 yr−1 value found with the are relative to the range of all other 100-year trends in the
ECHAM scaling. The CPS method finds a slope of −0.59 ◦ C reconstructions since 0 CE (Fig. 6). The most recent 100-
1000 yr−1 for the period 0–1900 CE, very close to the bore- year trend and its (±1σ ) uncertainty range are compared to
hole method. the distribution of all other 100-year trends for that region.
Over all of Antarctica, we find a cooling trend of −0.26 to This assessment method uses overlapping trend analysis in-
−0.4 ◦ C 1000 yr−1 between 0 and 1900 CE across the vari- tervals that are not fully independent, but it captures the full
ous scaling approaches (Table 3). This trend is comparable range of possible trends in all 100-year intervals of the re-
in magnitude to the −0.22 ± 0.06 ◦ C 1000 yr−1 found for gional reconstructions. In this analysis, the significant recent
the Arctic region (Kaufman et al., 2009). This trend was at- positive trends for the Dronning Maud Land Coast and West
tributed to a decrease in June–August solar insolation due to Antarctic Ice Sheet do not emerge as unusual in the context of
precession of the equinoxes. similar length trends of the last 2 millennia, consistent with
the findings of Steig et al. (2013) based on the WAIS Di-
4.1.3 Trends of the last 100 years in weighted
vide ice core. However, the most recent 100-year warming of
reconstructions
the Antarctic Peninsula is unusual compared to the range of
natural century-scale warming trends of the last 2 millennia
Studies based on individual ice cores have identified signifi- (Mulvaney et al., 2012). Thus, across all of our trend tests,
cant positive trends in the last century (Mulvaney et al., 2012; the warming of the Antarctic Peninsula over the last century
Steig et al., 2013). Similar to the findings for unweighted stands out as being robust to the binning of multiple isotopic
composites, significant warming trends in the weighted anomaly records for the region, significant after weighting
anomalies for the last 100 years (Table 4) are evident for and scaling of the regional isotopic records to represent re-

Table 4. Regression of the last 100-year slope based on the 5-year averages.
ECHAM NB2014 Borehole

Geographic region Start End Slope p value Slope p value Slope p value
date date (◦ C 100 yr−1 ) (◦ C 100 yr−1 ) (◦ C 100 yr−1 )
1. East Antarctic Plateau 1915 2010 −0.49 ± 0.73 0.174 −1.28 ± 1.71 0.133 NaN NaN
2. Wilkes Land Coast 1915 1995 1.70 ± −0.55 0.034 0.47 ± 0.60 0.115 NaN NaN
3. Weddell Sea Coast 1915 1995 −0.79 ± 1.75 0.351 −0.50 ± 1.12 0.351 NaN NaN
4. Antarctic Peninsula 1915 2010 2.87 ± 1.12 < 0.001 1.99 ± 0.75 < 0.001 NaN NaN
5. West Antarctic Ice Sheet 1915 2015 1.12 ± 0.92 0.020 0.97 ± 0.75 0.014 1.32 ± 1.08 0.020
6. Victoria Land Coast 1915 2005 −0.55 ± 0.97 0.247 −0.64 ± 1.00 0.195 NaN NaN
7. Dronning Maud Land Coast 1915 2010 1.33 ± 0.90 0.006 0.98 ± 0.73 0.012 NaN NaN
West Antarctica 1915 2015 1.25 ± 0.86 0.007 1.12 ± 0.62 0.001 0.97 ± 0.78 0.019
East Antarctica 1915 2010 0.09 ± 0.58 0.743 0.06 ± 1.23 0.924 NaN NaN
All Antarctica 1915 2015 0.79 ± 0.82 0.060 0.99 ± 0.94 0.040 0.21 ± 0.38 0.262
Normalized CPS
Geographic region Slope p value Slope p value Start End
(σ 100 yr−1 ) (◦ C 100 yr−1 ) date date
1. East Antarctic Plateau −0.85 ± 1.14 0.133 0.77 ± 0.46 0.004 1900 2000
2. Wilkes Land Coast 1.08 ± 1.37 0.115 0.05 ± 0.49 0.820 1900 1980
3. Weddell Sea Coast −1.10 ± 2.45 0.351 −0.25 ± 0.93 0.544 1900 1980
4. Antarctic Peninsula 2.65 ± 0.99 < 0.001 1.14 ± 0.60 0.002 1900 2000
5. West Antarctic Ice Sheet 1.17 ± 0.90 0.014 0.46 ± 0.63 0.132 1900 2000
6. Victoria Land Coast −0.79 ± 1.23 0.195 0.22 ± 0.52 0.364 1900 1990
7. Dronning Maud Land Coast 2.51 ± 1.88 0.012 0.59 ± 0.61 0.057 1900 1990
West Antarctica 1.50 ± 0.83 0.001 0.60 ± 0.61 0.052 1900 2000
East Antarctica 0.04±−0.85 0.924 0.43 ± 0.33 0.017 1900 2000
All Antarctica 0.72 ± 0.68 0.040 0.43 ± 0.32 0.013 1900 2000
gional temperature, and unusual compared with the range of contributing records in the unweighted composites. This is
natural century-scale temperature variability in this region also in agreement with what was recently found by Hakim
over the last 2000 years. The companion Antarctica2k pa- et al. (2016) using a data assimilation approach to identify
per (Thomas et al., 2017) that examines snow accumulation a very strong SH cooling trend through the last 2000 years,
across Antarctica over the last 1000 years also concludes that up to about 1850 CE.
the most robust recent changes in snowfall are evident for the
Antarctic Peninsula. Temperature and precipitation increases
over the Antarctic Peninsula in recent decades have previ- 4.2 Continent-scale temperature reconstructions
ously been linked to the SAM (Abram et al., 2014) and the
effects of the coupling between SAM and El Niño–Southern Figure 7 shows the results obtained by applying the CPS
Oscillation on the strength and position of the Amundsen Sea method published in 2013 (PAGES 2k Consortium, 2013) to
Low pressure and sea ice conditions offshore of West Antarc- the expanded Antarctica2k database to reconstruct temper-
tica and the Antarctic Peninsula (Thompson and Solomon, ature across the whole Antarctic continent. A good agree-
2002; Ding et al., 2011; Thomas and Abram, 2016). The ment is observed between the new reconstruction and the
variability in these driving forcings over the past 2000 years previously published one (Fig. 7), on both annual and
could also explain part of the opposite regional climate vari- decadal scales (r = 0.89 and 0.91, respectively). There is
ability that appears to be present on decadal and centennial a slightly stronger negative temperature trend, which is re-
scales between our reconstructions for the Antarctic Penin- lated mainly to a slightly cooler temperature estimate for
sula and the Victoria Land–Ross Sea regions. the 1600–1900 CE interval using the expanded database, in
Prior to this recent warming, which has been most sig- the new Antarctic continent-scale reconstruction. Figures in
nificant for the Antarctic Peninsula region of Antarctica, the Supplement similarly repeat this CPS comparison for the
long-term cooling is evident for many regions of Antarc- West Antarctica and East Antarctica regions and find simi-
tica. In particular, it is the long-term cooling of the East lar agreement between the 2013 results and those using the
Antarctic Plateau that emerges as significant in the weighted expanded ice core database (Figs S11, S12 and S13).
reconstructions and also robust across the distribution of Figure 8 shows the continental-scale (all of Antarctica),
East Antarctica and West Antarctic temperature reconstruc-

Figure 7. Comparison of CPS reconstructions of Antarctic mean temperatures over 167–2010 CE. Red shows results from PAGES 2k
Consortium (2013). Blue shows updated results using the new ice core isotope data collection described herein and the NB2014 temperature
target. Blue shading shows 2RMSE uncertainties of the updated reconstruction. The top panel shows unfiltered interannual reconstructions.
The second panel shows the 10-year running mean of reconstructions. The third (fourth) panel shows reduction of error skill from a split
calibration verification exercise using 1961–1976 for calibration (verification) and 1977–1991 for verification (calibration). The bottom panel
shows the 2RMSE reconstruction uncertainty range.
tions that have been obtained from the different tempera- in which cooling trends of −0.18 and −0.46 ◦ C 1000 yr−1
ture scaling approaches in comparison with the CPS method. were calculated, for the same interval, for East and West
The period between 1200 and 1900 CE appears as the cold- Antarctica, respectively. By comparison, cooling trends of
est interval in the East Antarctic reconstruction, and this is −0.25 and −0.56 ◦ C 1000 yr−1 are calculated from the 10-
also reflected in the reconstruction on the whole continen- year-binned data obtained with the same CPS methods but
tal scale. The warmest interval is identified between 300 and applied to this new database.
1000 CE, especially for West Antarctica. The comparison be- East Antarctica is one of the last places on the globe where
tween the CPS method and our new reconstruction methods the long-term cooling trend of the last millennium has not
shows a better agreement in East Antarctica than in West been inverted (PAGES 2k Consortium, 2013; Abram et al.,
Antarctica when considering the ECHAM temperature scal- 2016). This feature of delayed onset of industrial warming
ing (Fig. 8). However, a better agreement between the CPS is clear in our reconstruction, but is not captured by climate
method and the temperature reconstruction obtained with the models (PAGES 2k–PMIP3 group, 2015). Climate models
borehole scaling method is observed for West Antarctica. are known to overestimate inter-hemispheric synchronicity
Note that ECHAM and NB2014 use two independent data (Neukom et al., 2014). The model–data mismatch may come
sets for scaling the δ 18 O signal to temperature and that the from an overestimation of the response to external forcing
difference between them is a measure of the uncertainty in in climate models, an underestimation of the role of inter-
the temperature reconstruction. nal ocean–atmosphere dynamics in models or incorrect infer-
Over the period from 165 to 1900 CE (the reconstruction ence from proxy data. This quantitative reconstruction will
interval of the 2013 PAGES 2k reconstruction for Antarc- provide an important constraint on the forced vs. unforced
tica), a cooling trend is common to all three broad-scale re- nature of trends observed for the last century.
construction regions and across all reconstruction methods
(see Table S2). Cooling trends range between −0.21 and 4.3 Response to volcanic forcing
−0.45 ◦ C 1000 yr−1 for the ECHAM method and between
−0.34 and −0.53 ◦ C 1000 yr−1 for the borehole method. The radiative forcing caused by major volcanic eruptions is
These values are in the range of those previously reported known to drive hemispheric-scale cooling on subdecadal to
by the 2013 reconstruction (PAGES 2k Consortium, 2013) decadal timescales (e.g. Sigl et al., 2015; Stoffel et al., 2015).
Previous model-based assessments have suggested that the

Figure 8. Composite temperature reconstructions (T anomalies in ◦ C) for East, West and the whole of Antarctica using 10-year averages
and the different temperature scaling approaches: dotted coloured lines for the CPS method (2013 method applied to the new database); grey
lines for the method that uses the NB2014 variance for scaling (different weighting method compared to CPS); coloured lines for the method
based on the correlation between annual mean regional δ 18 O and regional T from ECHAM5-wiso forced by ERA-Interim; black lines for
the method adjusted to match the temperature trend between 1000 and 1600 CE based on borehole temperature measurements at WAIS
Divide (Orsi et al., 2012). Linear trends are calculated between 165 and 1900 CE. The grey horizontal segments correspond to volcanic–solar
downturn intervals as defined in PAGES 2k Consortium (2013) and corresponding to the following periods: 1251–1310 CE, 1431–1520 CE,
1581–1610 CE, 1641–1700 CE and 1791–1820 CE.
Figure 9. Reconstructed regional surface atmospheric temperature anomalies during the periods of 1802.5–1827.5 CE (a) and 1872.5–
1897.5 CE (b) overlapping with the three major tropical volcanic eruptions of 1808/09, Mount Tambora 1815 and Krakatoa (1883). The
reconstruction segments for each region were centred on the means of the corresponding intervals. Stars mark the bin centres of the recon-
structions, and the years of the eruptions are highlighted in grey.

long-term cooling trend over Antarctica during the 850– mate response to volcanic forcing (e.g. Fischer et al., 2007;
1850 CE interval may be attributed to volcanic forcing of the Stoffel et al., 2015). Figure 9 simply displays the regional
climate (Goosse et al., 2012). Volcanic forcing may also af- surface atmospheric temperature anomalies for the two peri-
fect atmospheric and oceanic dynamics, including modes of ods centred at 1815 CE and 1883 CE.
variability, causing substantial seasonal and regional varia- Following the Mount Tambora eruption, cooling is
tions in the climate footprint of volcanic forcing (e.g. Ot- recorded in five out of our seven regional reconstructions,
terå et al., 2010; Schleussner et al., 2015; Ortega et al., 2015; namely in the Dronning Maud Land Coast, Wilkes Land
Swingedouw et al., 2015). For example, the long-term cool- Coast, Victoria Land Coast, East Antarctic Plateau and possi-
ing trend of the global oceans during the last 2 millennia bly West Antarctic Ice Sheet regions. Following the 1808/09
has been attributed to the effects of short-term cooling from eruption, cooling appears in three out of our seven regional
episodic eruptions being cycled into subsurface layers of the reconstructions (West Antarctic Ice Sheet, Wilkes Land coast
oceans (McGregor et al., 2015). regions and to a lesser extent the Antarctic Peninsula). Fol-
We use the composite reconstructions of Antarctic tem- lowing the Krakatoa eruption, regional cooling emerges for
perature anomalies to explore whether signatures and/or im- four out of seven regions (East Antarctic Plateau, Antarctic
prints of past major eruptions can be identified. Since dif- Peninsula, Wilkes Land coast and West Antarctic Ice Sheet
ferent normalizations applied during the reconstruction pro- regions). Potential response time lags should be interpreted
cedure are not expected to affect a qualitative interpretation with caution due to effects of the data binning procedure as
of the results, we present the analysis based on the recon- well as age scale uncertainties.
struction that fits 5-year annual δ 18 O to regional temperature Within the limitations associated with our reconstructions
anomalies from ECHAM5-wiso forced by ERA-Interim. and with the small number of large eruptions in the time pe-
A robust identification of the climate impacts of volcan- riod studied, we can only report a lack of consistent pan-
ism from Antarctic ice cores faces a number of challenges. In Antarctic cooling and a variable regional response. Such
addition to chronological uncertainties due to low accumula- features have also been identified for the largest volcanic
tion and post-depositional alterations, processes other than eruptions in the Northern Hemisphere (e.g. Guillet et al.,
local temperature may affect water stable isotopes in Antarc- 2017). Local negative anomalies following major eruptions
tic precipitation, therefore making the detection of a tem- may arise from non-temperature drivers of water stable iso-
perature response to volcanic forcing more challenging. Fi- tope records and be artefacts of inconsistencies in chronolo-
nally, stacking the individual core records acts as a low-pass gies; they may also reflect an actual regional cooling. Further
filter, suppressing the variability in the shorter, interannual studies will require a focus on ice core records with a solid
timescales, thereby eliminating potentially short-lived cool- chronological control and sufficient regional records to as-
ing signals triggered by volcanic forcing. sess the common climatic signal from the deposition noise.
Despite these limitations, it should be possible to record Second-order isotopic parameters such as deuterium excess,
a volcanic cooling signal from well-dated cores from high- which preserves a signal associated with moisture source
accumulation areas and/or around the time periods with clear characteristics, may also provide insights into the response
markers of volcanic origin, widely used as precise age con- of the Southern Ocean surface state to volcanic forcing.
trol points for constructing ice core chronologies. For ex- This analysis represents only a preliminary attempt to as-
ample, the chemical fingerprints of two consecutive tropical sess the possible short-term climatic response of Antarctic
eruptions of 1808/09 (Volcanic Explosivity Index (VEI) 6– climate to volcanic forcing, and we suggest that this is a good
7) and Mount Tambora 1815 (VEI 7) are detected in many avenue for further detailed study using the expanded Antarc-
Antarctic ice cores as a pronounced double peak in a number tica2k isotope database assembled here.
of measured ice parameters such as electrical conductivity or
non-sea salt SO2− 4 profiles (Sigl et al., 2015).
Due to the aforementioned difficulties in the identifica- 5 Conclusions and implications
tion of a potential Antarctic temperature response to vol-
canic forcing, we restrict our analysis to the eruptions asso- We have presented a new enlarged database of high- and
ciated with the estimated strongest volcanic forcing, which low-resolution ice core water stable isotope records from
are more likely to leave a clear footprint in ice core records. Antarctica, which has been compiled in the framework of
The highest-resolution reconstruction (5-year averages) is the PAGES Antarctica2k working group. To further develop
available for a period that experienced only three eruptions our scientific understanding based on continent-scale recon-
with a VEI of 6–7, namely the tropical eruptions of 1808/09, structions of Antarctic temperature, we define seven climatic
Mount Tambora 1815 and Krakatoa 1883. The signatures of regions based on our knowledge of the regional Antarctic cli-
the first two eruptions are commonly used as reference hori- mate and further supported by spatial correlation of temper-
zons when constructing the core chronologies. Due to a low ature using the NB2014 reanalysis data. In our reconstruc-
number of replicates, we cannot use a superposed epoch anal- tions of both regional- and continent-scale climate, we use
ysis, commonly applied for detecting the signature of the cli- 5- and 10-year-averaged data in order to limit our depen-

dence on annually precise dating, which faces real problems, trends: the Antarctic Peninsula, the West Antarctic Ice Sheet
especially in interior Antarctica, due to the very low snow and the coastal Dronning Maud Land. However, currently
accumulation there. Similar conclusions are reached using only for the Antarctic Peninsula is the last 100-year trend un-
different approaches (ECHAM, NB2014 and borehole meth- usual in the context of natural century-scale trends over the
ods), giving support to our temperature reconstructions. Fur- last 2000 years. Although, if sustained and significant warm-
thermore, a replication of the CPS approach used in a pre- ing of Dronning Maud Land Coast and the West Antarctic
vious continent-scale reconstruction but adapted to the new Ice Sheet continues during the coming decades (as would
database also produces consistent results. The range of the be expected based on climate model simulations), then these
values obtained from these four methods provides informa- areas may soon also see the recent warming emerge as un-
tion on the uncertainties related to the temperature recon- usual in the context of natural variability. A companion pa-
structions on regional and continental scales. per (Thomas et al., 2017), which examines the snow ac-
Our new continent-scale reconstructions, based on the cumulation rate variability over Antarctica during the last
extended database, corroborate previously published find- millennium, reaches the same conclusion, showing that in
ings for Antarctica from the PAGES 2k Consortium (2013): the Antarctic Peninsula region the late 20th century snow-
(1) temperatures over the Antarctic continent show an over- fall increase is unusual in the context of the past 300 years.
all cooling trend during the period from 0 to 1900 CE, which The regional- to local-scale patterns of climate changes, and
appears strongest in West Antarctica, and (2) no continent- the processes that these illustrate, are another target for fu-
scale warming of Antarctic temperature is evident in the last ture research that could be explored using the new Antarc-
century. tica2k database. Future work will also consider the compari-
The robust and significant long-term cooling trend of son of the results obtained in this study with those by Thomas
Antarctic-wide temperatures from 0 to 1900 CE is also re- et al. (2017), assessing the snow accumulation rate variabil-
flected in the broader East Antarctic and West Antarctic re- ity in the same regions, with the aim of exploring a long-
gions. In East Antarctica, the coldest interval occurs from standing question about the relationship between tempera-
1200 to 1900 CE. During the same time period, five intervals ture and precipitation in Antarctica.
of volcanic–solar downturns have been previously linked to Our new regional- and continent-scale reconstructions of
cooling in temperature reconstructions for other continental Antarctic temperatures again highlight the urgent need to in-
regions (PAGES 2k Consortium, 2013). These intervals of crease the spatial coverage of the sparse data for this con-
negative radiative forcings are in agreement with our lower tinent with new temperature records, particularly focusing
temperature estimates during this time (Fig. 8). in the coastal areas. These regions are typically seen to
The absence of significant continent-scale warming of be data-poor areas in our extended database. However, the
Antarctica over the last 100 years is in clear contrast with the coastal areas are also the places where high-resolution ice
significant industrial-era warming trends that are evident in cores can be retrieved, where strong climatic differences can
reconstructions for all other continents (except Africa) and occur over relatively short distances, where there is some
the tropical oceans (Abram et al., 2016). As noted in other evidence for non-significant but emerging development of
studies (e.g. Abram et al., 2016; Jones et al., 2016) the ab- warming trends, where even small amounts of warming can
sence of continent-scale warming over Antarctica is not in rapidly move the local climate beyond the threshold and
agreement with climate model simulations, which consis- where surface melting of snow occurs, potentially leading to
tently produce a 20th century warming trend over Antarc- widespread impacts of any future Antarctic warming.
tica in response to greenhouse gas forcing. The high interan-
nual climate variability with respect to the magnitude of cli-
mate trends over Antarctica, and the small number of ice core Data availability. The target data set of NB2014 is archived at
records that have contributed to previous continent-scale re- http://polarmet.osu.edu/datasets/Antarctic_recon/. Table S1 (in the
constructions of Antarctic temperature, have been suggested Supplement) contains the data citations, as well as the primary bib-
as possible sources for the lack of detection of continent- liographic reference, for each record used in this study. We used the
URL links for existing published data from public repositories.
scale warming trends in palaeoclimate records from Antarc-
The regional- and continent-scale isotopic composites, includ-
tica. However, we replicate this finding of a continent-scale ing the statistical information for the 5- and 10-year unweighted
absence of 20th century warming using our greatly expanded composites, and temperature reconstructions generated during the
ice core database and with composites and reconstructions current study and Table S1 are available in the NOAA World Data
based on a range of methods. This suggests that the ab- Center for Paleoclimatology (WDC Paleo) at the following link:
sence of continent-scale warming over Antarctica during the https://www.ncdc.noaa.gov/paleo/study/22589.
last century is a robust result and alternate reasons for data–
model divergences need to be investigated.
Despite the lack of significant warming over the last
The Supplement related to this article is available online
100 years on the continent scale, there are three regions
at https://doi.org/10.5194/cp-13-1609-2017-supplement.
that show positive and significant isotopic and temperature

Author contributions. BS led Phase 2 of the Antarctica2k Bertler, N. A. N., Conway, H., Dahl-Jensen, D., Emanuelsson, D.
project, and MAJC was data manager. The data analysis presented B., Winstrup, M., Vallelonga, P. T., Lee, J. E., Brook, E. J., Sev-
in this paper was carried out by NJA, AO, SG, RN and DD. BS, NJA eringhaus, J. P., Fudge, T. J., Keller, E. D., Baisden, W. T., Hind-
and AO led the writing of the paper, with contributions by MAJC, marsh, R. C. A., Neff, P. D., Blunier, T., Edwards, R., Mayewski,
SG, RN, VM-D, HG, DD and EJS. All authors contributed to dis- P. A., Kipfstuhl, S., Buizert, C., Canessa, S., Dadic, R., Kjær,
cussions on the analysis design and paper. H. A., Kurbatov, A., Zhang, D., Waddington, E. D., Baccolo,
G., Beers, T., Brightley, H. J., Carter, L., Clemens-Sewall, D.,
Ciobanu, V. G., Delmonte, B., Eling, L., Ellis, A. A., Ganesh,
Special issue statement. This article is part of the special issue S., Golledge, N. R., Haines, S. A., Handley, M., Hawley, R.
“Climate of the past 2000 years: regional and trans-regional synthe- L., Hogan, C. M., Johnson, K. M., Korotkikh, E., Lowry, D.
ses”. It is not associated with a conference. P., Mandeno, D., McKay, R. M., Menking, J. A., Naish, T. R.,
Noerling, C., Ollive, A., Orsi, A., Proemse, B. C., Pyne, A. R.,
Pyne, R. L., Renwick, J., Scherer, R. P., Semper, S., Simon-
Acknowledgements. This paper is the Antarctic contribution to sen, M., Sneed, S. B., Steig, E. J., Tuohy, A., Ulayottil Venu-
the international effort to characterize global climate variability dur- gopal, A., Valero-Delgado, F., Venkatesh, J., Wang, F., Wang,
ing the last 2000 years, coordinated within the Past Global Changes S., Winski, D. A., Winton, V. H. L., Whiteford, A., Xiao, C.,
PAGES 2k network working group (http://www.pages-igbp.org/ini/ Yang, J., and Zhang, X.: The Ross Sea Dipole – Temperature,
wg/2k-network/intro) and the planning group International Partner- Snow Accumulation and Sea Ice Variability in the Ross Sea Re-
ship for Ice Core Science, IPICS (http://www.pages-igbp.org/ini/ gion, Antarctica, over the Past 2700 Years, Clim. Past Discuss.,
end-aff/ipics/intro). We thank all of the scientists who made their https://doi.org/10.5194/cp-2017-95, in review, 2017.
ice core data available for this study. Bonne, J.-L., Masson-Delmotte, V., Cattani, O., Delmotte, M.,
This is a contribution to the PAGES 2k Network (through the Risi, C., Sodemann, H., and Steen-Larsen, H. C.: The iso-
Antarctica2k working group). Past Global Changes (PAGES) is topic composition of water vapour and precipitation in Ivit-
supported by the US and Swiss National Science Foundations. Ner- tuut, southern Greenland, Atmos. Chem. Phys., 14, 4419–4439,
ilie J. Abram acknowledges funding from the Australian Research https://doi.org/10.5194/acp-14-4419-2014, 2014.
Council through Discovery Project DP140102059. Barbara Stenni Bromwich, D. H., Nicolas, J. P., Monaghan, A. J., Lazzara, M. A.,
acknowledges funding from the PNRA (Italian Antarctic Research Keller, L. M., Weidner, G. A., and Wilson, A. B.: Central West
Programme) IPICS-2kyr-It project. Raphael Neukom is funded by Antarctica among the most rapidly warming regions on Earth,
the Swiss NSF (Ambizione grant PZ00P2_154802). Nat. Geosci., 6, 139–145, https://doi.org/10.1038/ngeo1671,
2013.
Edited by: Christian Turney Casado, M., Landais, A., Masson-Delmotte, V., Genthon, C., Ker-
Reviewed by: two anonymous referees stel, E., Kassi, S., Arnaud, L., Picard, G., Prie, F., Cattani, O.,
Steen-Larsen, H.-C., Vignon, E., and Cermak, P.: Continuous
measurements of isotopic composition of water vapour on the
East Antarctic Plateau, Atmos. Chem. Phys., 16, 8521–8538,
https://doi.org/10.5194/acp-16-8521-2016, 2016.
References Dee, D., Uppala, S., Simmons, A., Berrisford, P., Poli, P.,
Kobayashi, S., Andrae, U., Balmaseda, M., Balsamo, G., and
Abram, N. J., Mulvaney, R., and Arrowsmith, C.: Environmen- Bauer, P.: The ERA-Interim reanalysis: configuration and perfor-
tal signals in a highly resolved ice core from James Ross mance of the data assimilation system, Q. J. Roy. Meteor. Soc.,
Island, Antarctica, J. Geophys. Res.-Atmos., 116, D20116, 137, 553–597, 2011.
https://doi.org/10.1029/2011JD016147, 2011. Ding, Q., Steig, E. J., Battisti, D. S., and Küttel, M.: Winter warm-
Abram, N. J., Mulvaney, R., Wolff, E. W., Triest, J., Kipfs- ing in West Antarctica caused by central tropical Pacific warm-
tuhl, S., Trusel, L. D., Vimeux, F., Fleet, L., and Arrow- ing, Nat. Geosci., 4, 398–403, https://doi.org/10.1038/ngeo1129,
smith, C.: Acceleration of snow melt in an Antarctic Peninsula 2011.
ice core during the twentieth century, Nat. Geosci., 6, 404–411, Dittmann, A., Schlosser, E., Masson-Delmotte, V., Powers, J. G.,
https://doi.org/10.1038/ngeo1787, 2013. Manning, K. W., Werner, M., and Fujita, K.: Precipitation
Abram, N. J., Mulvaney, R., Vimeux, F., Phipps, S. J., Turner, J., regime and stable isotopes at Dome Fuji, East Antarctica, At-
and England, M. H.: Evolution of the Southern Annular Mode mos. Chem. Phys., 16, 6883–6900, https://doi.org/10.5194/acp-
during the past millennium, Nat. Clim. Change, 4, 564–569, 16-6883-2016, 2016.
https://doi.org/10.1038/NCLIMATE2235, 2014. Ekaykin, A. A., Lipenkov, V. Y., Kuzmina, I. N., Petit, J. R.,
Abram, N. J., McGregor, H. V., Tierney, J. E., Evans, M. N., Masson-Delmotte, V., and Johnsen, S. J.: The changes in isotope
McKay, N. P., Kaufman, D. S., and the PAGES 2k composition and accumulation of snow at Vostok Station, East
Consortium: Early onset of industrial-era warming Antarctica, over the past 200 years, Ann. Glaciol., 39, 569–575,
across the oceans and continents, Nature, 536, 411–418, 2004.
https://doi.org/10.1038/nature19082, 2016. Ekaykin, A. A., Kozachek, A. V., Lipenkov, V. Y., and
Armour, K. C., Marshall, J., Scott, J. R., Donohoe, A., and New- Shibaev, Y. A.: Multiple climate shifts in the Southern Hemi-
som, E. R.: Southern Ocean warming delayed by circumpolar sphere over the past three centuries based on central Antarc-
upwelling and equatorward transport, Nat. Geosci., 9, 549–554,
https://doi.org/10.1038/ngeo2731, 2016.

tic snow pits and core studies, Ann. Glaciol., 55, 259–266, and Oppenheimer, C.: Climate response to the Samalas volcanic
https://doi.org/10.3189/201AoG66A189, 2014. eruption in 1257 revealed by proxy records, Nat. Geosci., 10,
Emanuelsson, D. B., Bertler, N. A. N., Neff, P., Renwick, J., 123–128, https://doi.org/10.1038/ngeo2875, 2017.
Markle, B., Baisden, W. T., Keller, E.: The role of Amundsen- Hakim, G. J., Emile-Geay, J., Steig, E. J., Noone, D., An-
Bellingshausen Sea anticyclonic circulation in forcing marine derson, D. M., Tardif, R., Steiger, N., and Perkins, W. A.:
air intrusions into West Antarctica, Clim. Dynam., CLDY-D-16- The last millennium climate reanalysis project: framework
00858, submitted, 2017. and first results, J. Geophys. Res.-Atmos., 121, 6745–6764,
Fegyveresi, J. M., Alley, R. B., Spencer, M. K., Fitzpatrick, J. J., https://doi.org/10.1002/2016JD024751, 2016.
Steig, E. J., White, J. W. C., McConnell, J. R., and Taylor, K. C.: Hoffmann, G., Werner, M., and Heimann, M.: Water isotope module
Late-Holocene climate evolution at the WAIS Divide site, West of the ECHAM atmospheric general circulation model: a study
Antarctica: bubble number-density estimates, J. Glaciol., 57, on timescales from days to several years, J. Geophys. Res.-
629–638, 2011. Atmos., 103, 16871–16896, 1998.
Fernandoy, F., Meyer, H., Oerter, H., Wilhelms, F., Graf, W., and Hoshina, Y., Fujita, K., Nakazawa, F., Iizuka, Y., Miyake, T.,
Schwander, J.: Temporal and spatial variation of stable-isotope Hirabayashi, M., Kuramoto, T., Fujita, S., and Motoyama, H.: Ef-
ratios and accumulation rates in the hinterland of Neumayer Sta- fect of accumulation rate on water stable isotopes of near-surface
tion, East Antarctica, J. Glaciol., 56, 673–687, 2010. snow in inland Antarctica, J. Geophys. Res.-Atmos., 119, 274–
Fernandoy, F., Meyer, H., and Tonelli, M.: Stable water isotopes of 283, https://doi.org/10.1002/2013JD020771, 2014.
precipitation and firn cores from the northern Antarctic Peninsula Jones, P. D., Briffa, K. R., Osborn, T. J., Lough, J. M.,
region as a proxy for climate reconstruction, The Cryosphere, 6, van Ommen, T. D., Vinther, B. M., Luterbacher, J.,
313–330, https://doi.org/10.5194/tc-6-313-2012, 2012. Wahl, E. R., Zwiers, F. W., Mann, M. E., Schmidt, G. A.,
Fischer, E. M., Luterbacher, J., Zorita, E., Tett, S. F. B., Casty, C., Ammann, C. M., Buckley, B. M., Cobb, K. M., Es-
and Wanner, H.: European climate response to tropical volcanic per, J., Goosse, H., Graham, N., Jansen, E., Kiefer, T.,
eruptions over the last half millennium, Geophys. Res. Lett., 34, Kull, C., Küttel, M., Mosley-Thompson, E., Overpeck, J. T.,
L05707, https://doi.org/10.1029/2006GL027992, 2007. Riedwyl, N., Schulz, M., Tudhope, A. W., Villalba, R.,
Flato, G., Marotzke, J., Abiodun, B., Braconnot, P., Chou, S. C., Wanner, H., Wolff, E., and Xoplaki, E.: High-resolution
Collins, W., Cox, P., Driouech, F., Emori, S., Eyring, V., For- palaeoclimatology of the last millennium: a review of cur-
est, C., Gleckler, P., Guilyardi, E., Jakob, C., Kattsov, V., Rea- rent status and future prospects, The Holocene, 19, 3–49,
son, C., and Rummukainen, M.: Evaluation of climate models, https://doi.org/10.1177/0959683608098952, 2009.
in: Climate Change 2013: The Physical Science Basis. Con- Jones, J. M., Gille, S. T., Goosse, H., Abram, N. J., Canziani, P. O.,
tribution of Working Group I to the Fifth Assessment Report Charman, D. J., Clem, K. R., Crosta, X., de Lavergne, C., Eisen-
of the Intergovernmental Panel on Climate Change, edited by: man, I., England, M. H., Fogt, R. L., Frankcombe, L. M., Mar-
Stocker, T. F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K., shall, G. J., Masson-Delmotte, V., Morrison, A. K., Orsi, A. J.,
Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P. M., Cam- Raphael, M. N., Renwick, J. A., Schneider, D. P., Simp-
bridge University Press, Cambridge, UK and New York, NY, kins, G. R., Steig, E. J., Stenni, B., Swingedouw, D., and
USA, 2013. Vance, T. R.: Assessing recent trends in high-latitude Southern
Frezzotti, M., Urbini, S., Proposito, M., Scarchilli, C., and Gan- Hemisphere surface climate, Nat. Clim. Change, 6, 917–926,
dolfi, S.: Spatial and temporal variability of surface mass bal- https://doi.org/10.1038/NCLIMATE3103, 2016.
ance near Talos Dome, East Antartica, J. Geophys. Res.-Earth, Jouzel, J.: Water stable isotopes: atmospheric composition and ap-
112, F02032, https://doi.org/10.1029/2006JF000638, 2007. plications in polar ice core studies, in: Treatise on Geochem-
Goosse, H., Braida, M., Crosta, X., Mairesse, A., Masson- istry, 2nd Ed., Vol. 5, The Atmosphere, edited by: Keeling, R.
Delmotte, V., Mathiot, P., Neukom, R., Oerter, H., Philippon, G., and Russel, L., Elsevier, 213–256, https://doi.org/10.1016/B978-
Renssen, H., Stenni, B., van Ommen, T., and Verleyen, E.: 0-08-095975-7.00408-3, 2014.
Antarctic temperature changes during the last millennium: eval- Jouzel, J. and Merlivat, L.: Deuterium and oxygen 18 in precipita-
uation of simulations and reconstructions, Quaternary Sci. Rev., tion: modeling of the isotopic effects during snow formation, J.
55, 75–90, https://doi.org/10.1016/j.quascirev.2012.09.003, Geophys. Res.-Atmos., 89, 11749–11757, 1984.
2012. Kaufman, D. S., Schneider, D. P., McKay, N. P., Ammann, C. M.,
Goosse, H.: Reconstructed and simulated temperature asymmetry Bradley, R. S., Briffa, K. R., Miller, G. H., Otto-Bliesner, B. L.,
between continents in both hemispheres over the last centuries. Overpeck, J. T., Vinther, B. M., and Lakes, A.: Recent warm-
Clim. Dynam., 48, 1483–1501, https://doi.org/10.1007/s00382- ing reverses long-term Arctic cooling, Science, 325, 1236–1239,
016-3154-z, 2017. 2009.
Goursaud, S., Masson-Delmotte, V., Favier, V., Preunkert, S., Fily, Küttel, M., Steig, E. J., Ding, Q., Monaghan, A. J., and Bat-
M., Gallée, H., Jourdain, B., Legrand, M., Magand, O., Minster, tisti, D. S.: Seasonal climate information preserved in West
B., and Werner, M.: A 60-year ice-core record of regional climate Antarctic ice core water isotopes: relationships to temperature,
from Adélie Land, coastal Antarctica, The Cryosphere, 11, 343– large-scale circulation, and sea ice, Clim. Dynam., 39, 1841–
362, https://doi.org/10.5194/tc-11-343-2017, 2017. 1857, https://doi.org/10.1007/s00382-012-1460-7, 2012.
Guillet, S., Corona, C., Stoffel, M., Khodri, M., Lavigne, F., Ortega, Masson-Delmotte, V., Hou, S., Ekaykin, A., Jouzel, J., Aris-
P., Eckert, N., Sielenou, P. D., Daux, V., Churakova (Sidorova), tarain, A., Bernardo, R., Bromwich, D., Cattani, O., Del-
O. V., Davi, N., Edouard, J. L., Zhang, Y., Luckman, B. H., motte, M., Falourd, S., Frezzotti, M., Gallée, H., Genoni, L.,
Myglan, V. S., Guiot, J., Beniston, M., Masson-Delmotte, V., Isaksson, E., Landais, A., Helsen, M., Hoffman, G., Lopez, J.,

Morgan, V., Motoyama, H., Noone, D., Oerter, H., Petit, J., PAGES 2k-PMIP3 group: Continental-scale temperature variabil-
Royer, A., Uemura, R., Schmidt, G., Schlosser, E., Simões, J., ity in PMIP3 simulations and PAGES 2k regional temperature
Steig, E., Stenni, B., Stievenard, M., Van den Breoke, M., Van reconstructions over the past millennium, Clim. Past, 11, 1673–
de Wak, R., Van de Berg, W., Vimeux, F., and White, J.: A re- 1699, https://doi.org/10.5194/cp-11-1673-2015, 2015.
view of Antarctic surface snow isotopic composition: observa- Plummer, C. T., Curran, M. A. J., van Ommen, T. D., Rasmussen,
tions, atmospheric circulation, and isotopic modeling, J. Climate, S. O., Moy, A. D., Vance, T. R., Clausen, H. B., Vinther, B.
21, 3359–3387, https://doi.org/10.1175/2007JCLI2139.1, 2008. M., and Mayewski, P. A.: An independently dated 2000-yr vol-
Masson-Delmotte, V., Buiron, D., Ekaykin, A., Frezzotti, M., canic record from Law Dome, East Antarctica, including a new
Gallée, H., Jouzel, J., Krinner, G., Landais, A., Motoyama, perspective on the dating of the 1450s CE eruption of Kuwae,
H., Oerter, H., Pol, K., Pollard, D., Ritz, C., Schlosser, E., Vanuatu, Clim. Past, 8, 1929–1940, https://doi.org/10.5194/cp-
Sime, L. C., Sodemann, H., Stenni, B., Uemura, R., and 8-1929-2012, 2012.
Vimeux, F.: A comparison of the present and last interglacial Ritter, F., Steen-Larsen, H. C., Werner, M., Masson-Delmotte,
periods in six Antarctic ice cores, Clim. Past, 7, 397–423, V., Orsi, A., Behrens, M., Birnbaum, G., Freitag, J., Risi, C.,
https://doi.org/10.5194/cp-7-397-2011, 2011. and Kipfstuhl, S.: Isotopic exchange on the diurnal scale be-
McGregor, H. V., Evans, M. N., Goosse, H., Leduc, G., Martrat, B., tween near-surface snow and lower atmospheric water vapor
Addison, J. A., Mortyn, P. G., Oppo, D. W., Seidenkrantz, M.- at Kohnen station, East Antarctica, The Cryosphere, 10, 1647–
S., Sicre, M.-A., Phipps, S. J., Selvaraj, K., Thirumalai, K., Fil- 1663, https://doi.org/10.5194/tc-10-1647-2016, 2016.
ipsson, H. L., and Ersek, V.: Robust global ocean cooling trend Roeckner, E., Bäuml, G., Bonaventura, L., Brokopf, R., Esch, M.,
for the pre-industrial Common Era, Nat. Geosci., 8, 671–677, Giorgetta, M., Hagemann, S., Kirchner, I., Kornblueh, L.,
https://doi.org/10.1038/NGEO2510, 2015. and Manzini, E.: The atmospheric general circulation model
Mulvaney, R., Abram, N. J., Hindmarsh, R. C. A., Arrow- ECHAM 5. PART I: Model description, Report 349, Max-
smith, C., Fleet, L., Triest, J., Sime, L. C., Alemany, O., Planck-Institut für Meteorologie, Hamburg, 2003.
and Foord, S.: Recent Antarctic Peninsula warming relative to Schleussner, C. F., Divine, D. V., Donges, J. F., Miettinen, A., and
Holocene climate and ice-shelf history, Nature, 489, 141–144, Donner, R. V.: Indications for a North Atlantic ocean circulation
https://doi.org/10.1038/nature11391, 2012. regime shift at the onset of the Little Ice Age, Clim. Dynam., 45,
Neff, P. D. and Bertler, N. A. N.: Trajectory modeling 3623–3633, https://doi.org/10.1007/s00382-015-2561-x, 2015.
of modern dust transport to the Southern Ocean and Schlosser, E., Reijmer, C., Oerter, H., and Graf, W.: The influence
Antarctica, J. Geophys. Res.-Atmos., 120, 9303–9322, of origin of precipitation on the δ 18 O-T relationship at Neumayer
https://doi.org/10.1002/2015JD023304, 2015. Station, Ekströmisen, Antarctica, Ann. Glaciol., 39, 41–48, 2004.
Neukom, R., Gergis, J., Karoly, D. J., Wanner, H., Curran, M., El- Schlosser, E., Reijmer, C., Oerter, H., and Graf, W.: The influence
bert, J., González-Rouco, F., Linsley, B. K., Moy, A. D., Mundo, of origin of precipitation on the δ 18 O-T relationship at Neumayer
I., Raible, C. C., Steig, E. J., van Ommen, T., Vance, T., Villalba, Station, Ekströmisen, Antarctica, Ann. Glaciol., 39, 41–48, 2004.
R., Zinke, J., and Frank, D.: Inter-hemispheric temperature vari- Schmidt, G. A., LeGrande, A. N., and Hoffmann, G.: Water iso-
ability over the past millennium, Nat. Clim. Change, 4, 362–367, tope expressions of intrinsic and forced variability in a cou-
https://doi.org/10.1038/nclimate2174, 2014. pled ocean–atmosphere model, J. Geophys. Res.-Atmos., 112,
Nicolas, J. P. and Bromwich, D. H.: New reconstruction of D10103, https://doi.org/10.1029/2006JD007781, 2007.
antarctic near-surface temperatures: multidecadal trends and Schneider, D. P., Steig, E. J., and van Ommen, T.: Interpretation
reliability of global reanalyses, J. Climate, 27, 8070–8093, of high resolution ice core stable isotope records from Antarc-
https://doi.org/10.1175/JCLI-D-13-00733.1, 2014. tica: towards interannual climate reconstruction, Ann. Glaciol.,
Orsi, A. J., Cornuelle, B. D., and Severinghaus, J. P.: Little Ice Age 41, 63–70, 2005.
cold interval in West Antarctica: evidence from borehole temper- Schneider, D. P., Steig, E. J., van Ommen, T. D., Dixon, D. A.,
ature at the West Antarctic Ice Sheet (WAIS) Divide, Geophys. Mayewski, P. A., Jones, J. M., and Bitz, C. M.: Antarctic tem-
Res. Lett., 39, L09710, https://doi.org/10.1029/2012GL051260, peratures over the past two centuries from ice cores, Geophys.
2012. Res. Lett., 33, L16707, https://doi.org/10.1029/2006GL027057,
Ortega, P., Lehner, F., Swingedouw, D., Masson-Delmotte, V., 2006.
Raible, C. C., Casado, M., and Yiou, P.: A model-tested North Sigl, M., McConnell, J. R., Toohey, M., Curran, M., Das, S. B., Ed-
Atlantic Oscillation reconstruction for the past millennium, Na- wards, R., Isaksson, E., Kawamura, K., Kipfstuhl, S., Krüger, K.,
ture, 523, 71–74, https://doi.org/10.1038/nature14518, 2015. Layman, L., Maselli, O. J., Motizuki, Y., Motoyama, H., Pas-
Otterå, O. H., Bensen, M., Drange, H., and Suo, L.: External forc- teris, D. R., and Severi, M.: Insights from Antarctica on volcanic
ing as a metronome for Atlantic multidecadal variability, Nat. forcing during the Common Era, Nat. Clim. Change, 4, 693–697,
Geosci., 3, 688–694, https://doi.org/10.1038/NGEO955, 2010. https://doi.org/10.1038/NCLIMATE2293, 2014.
PAGES 2k Consortium: Continental-scale temperature variability Sigl, M., Winstrup, M., McConnell, J. R., Welten, K. C.,
during the past two millennia, Nat. Geosci., 6, 339–346, Pub- Plunkett, G., Ludlow, F., Büntgen, U., Caffee, M., Chell-
lished online 21 April 2013, https://doi.org/10.1038/NGEO1797, man, N., Dahl-Jensen, D., Fischer, H., Kipfstuhl, S., Kostick, C.,
2013. Maselli, O. J., Mekhaldi, F., Mulvaney, R., Muscheler, R., Pas-
PAGES 2k Consortium: A global multiproxy database for temper- teris, D. R., Pilcher, J. R., Salzer, M., Schüpbach, S., Stef-
ature reconstructions of the Common Era, Sci. Data, 4, 170088, fensen, J. P., Vinther, B. M., and Woodruff, T. E.: Timing and
https://doi.org/10.1038/sdata.2017.88, 2017. climate forcing of volcanic eruptions for the past 2500 years, Na-
ture, 523, 543–549, 2015.

Sigl, M., Fudge, T. J., Winstrup, M., Cole-Dai, J., Ferris, D., Mc- since 1702 A.D., Geophys. Res. Lett., 43, 5309–5317,
Connell, J. R., Taylor, K. C., Welten, K. C., Woodruff, T. E., https://doi.org/10.1002/2016GL068130, 2016.
Adolphi, F., Bisiaux, M., Brook, E. J., Buizert, C., Caffee, M. Thomas, E. R. and Bracegirdle, T. J.: Precipitation pathways for five
W., Dunbar, N. W., Edwards, R., Geng, L., Iverson, N., Koff- new ice core sites in Ellsworth Land, West Antarctica, Clim. Dy-
man, B., Layman, L., Maselli, O. J., McGwire, K., Muscheler, nam., 44, 2067–2078, https://doi.org/10.1007/s00382-014-2213-
R., Nishiizumi, K., Pasteris, D. R., Rhodes, R. H., and Sowers, 6, 2015.
T. A.: The WAIS Divide deep ice core WD2014 chronology – Thomas, E. R., Dennis, P. F., Bracegirdle, T. J., and Franzke, C.:
Part 2: Annual-layer counting (0–31 ka BP), Clim. Past, 12, 769– Ice core evidence for significant 100 year regional warming
786, https://doi.org/10.5194/cp-12-769-2016, 2016. on the Antarctic Peninsula, Geophys. Res. Lett., 36, L20704,
Sime, L. C., Marshall, G. J., Mulvaney, R., and Thomas, E. R.: Inter- https://doi.org/10.1029/2009GL040104, 2009.
preting temperature information from ice cores along the Antarc- Thomas, E. R., Bracegirdle, T. J., Turner, J., and
tic Peninsula: ERA40 analysis, Geophys. Res. Lett., 36, L18801, Wolff, E. W.: A 308 year record of climate variability
https://doi.org/10.1029/2009GL038982, 2009. in West Antarctica, Geophys. Res. Lett., 40, 5492–5496,
Sodemann, H. and Stohl, A.: Asymmetries in the moisture ori- https://doi.org/10.1002/2013GL057782, 2013.
gin of Antarctic precipitation, Geophys. Res. Lett., 36, L22803, Thomas, E. R., van Wessem, J. M., Roberts, J., Isaksson, E.,
https://doi.org/10.1029/2009GL040242, 2009. Schlosser, E., Fudge, T. J., Vallelonga, P., Medley, B., Lenaerts,
Steen-Larsen, H. C., Masson-Delmotte, V., Hirabayashi, M., Win- J., Bertler, N., van den Broeke, M. R., Dixon, D. A., Frezzotti,
kler, R., Satow, K., Prié, F., Bayou, N., Brun, E., Cuffey, K. M., Stenni, B., Curran, M., and Ekaykin, A. A.: Regional Antarc-
M., Dahl-Jensen, D., Dumont, M., Guillevic, M., Kipfstuhl, S., tic snow accumulation over the past 1000 years, Clim. Past, 13,
Landais, A., Popp, T., Risi, C., Steffen, K., Stenni, B., and 1491-1513, https://doi.org/10.5194/cp-13-1491-2017, 2017.
Sveinbjörnsdottír, A. E.: What controls the isotopic compo- Thompson, D. W. J. and Solomon, S.: Interpretation of recent
sition of Greenland surface snow?, Clim. Past, 10, 377–392, Southern Hemisphere climate change, Science, 296, 895–899,
https://doi.org/10.5194/cp-10-377-2014, 2014. https://doi.org/10.1126/science.1069270, 2002.
Steen-Larsen, H. C., Risi, C., Werner, M., Yoshimura, K., and Thompson, D. W. J., Solomon, S., Kushner, P. J., England, M. H.,
Masson-Delmotte, V.: Evaluating the skills of isotope-enabled Grise, K. M., and Karoly, D. J.: Signatures of the Antarctic
general circulation models against in situ atmospheric water va- ozone hole in Southern Hemisphere surface climate change, Nat.
por isotope observations, J. Geophys. Res.-Atmos., 122, 246– Geosci., 4, 741–749, https://doi.org/10.1038/ngeo1296, 2011.
263, https://doi.org/10.1002/2016JD025443, 2017. Touzeau, A., Landais, A., Stenni, B., Uemura, R., Fukui, K., Fu-
Steig, E. J., Schneider, D. P., Rutherford, S. D., Mann, M. E., jita, S., Guilbaud, S., Ekaykin, A., Casado, M., Barkan, E.,
Comiso, J. C., and Shindell, D. T.: Warming of the Antarctic ice- Luz, B., Magand, O., Teste, G., Le Meur, E., Baroni, M.,
sheet surface since the 1957 International Geophysical Year, Na- Savarino, J., Bourgeois, I., and Risi, C.: Acquisition of iso-
ture, 457, 459–462, https://doi.org/10.1038/nature07669, 2009. topic composition for surface snow in East Antarctica and the
Steig, E. J., Ding, Q., White, J. W. C., Küttel, M., Rupper, S. B., links to climatic parameters, The Cryosphere, 10, 837–852,
Neumann, T. A., Neff, P. D., Gallant, A. J. E., Mayewski, P. A., https://doi.org/10.5194/tc-10-837-2016, 2016.
Taylor, K. C., Hoffmann, G., Dixon, D. A., Schoenemann, S. W., Uppala, S. M., Kållberg, P., Simmons, A., Andrae, U., Bechtold, V.
Markle, B. R., Fudge, T. J., Schneider, D. P., Schauer, A. J., d., Fiorino, M., Gibson, J., Haseler, J., Hernandez, A., and
Teel, R. P., Vaughn, B. H., Burgener, L., Williams, J., and Ko- Kelly, G.: The ERA-40 re-analysis, Q. J. Roy. Meteor. Soc., 131,
rotkikh, E.: Recent climate and ice-sheet changes in West Antarc- 2961–3012, 2005.
tica compared with the past 2000 years, Nat. Geosci., 6, 372–375, Wang, Y., Ding, M., van Wessem, J. M., Schlosser, E., Altnau, S.,
https://doi.org/10.1038/ngeo1778, 2013. van den Broeke, M. R., Lenaerts, J. T. M., Thomas, E. R.,
Stenni, B., Scarchilli, C., Masson-Delmotte, V., Schlosser, E., Cia- Isaksson, E., Wang, J., and Sun, W.: A comparison of Antarc-
rdini, V., Dreossi, G., Grigioni, P., Bonazza, M., Cagnati, A., tic Ice Sheet surface mass balance from atmospheric climate
Karlicek, D., Risi, C., Udisti, R., and Valt, M.: Three year models and in situ observations, J. Climate, 29, 5317–5337,
monitoring of stable isotopes of precipitation at Concordia https://doi.org/10.1175/JCLI-D-15-0642.1, 2016.
Station, East Antarctica, The Cryosphere, 10, 2415–2428, van Ommen, T. D. and Morgan, V. I.: Calibrating the ice core pale-
https://doi.org/10.5194/tc-10-2415-2016, 2016. othermometer using seasonality, J. Geophys. Res.-Atmos., 102,
Stoffel, M., Khodri, M., Corona, C., Guillet, S., Poulain, V., 9351–9357, 1997.
Bekki, S., Guiot, J., Luckman, B., Oppenheimer, C., Van Wessem, J. M., Reijmer, C. H., Morlighem, M., Mouginot, J.,
Lebas, N., Beniston, M., and Masson-Delmotte, V.: Esti- Rignot, E., Medley, B., and Van Meijgaard, E.: Improved rep-
mates of volcanic-induced cooling in the Northern Hemi- resentation of East Antarctic surface mass balance in a re-
sphere over the past 1500 years, Nat. Geosci., 8, 784–788, gional atmospheric climate model, J. Glaciol., 60, 761–770,
https://doi.org/10.1038/ngeo2526, 2015. https://doi.org/10.3189/2014JoG14J051, 2014.
Swingedouw, D., Ortega, P., Mignot, J., Guilyardi, E., Masson- Werner, M., Heimann, M., and Hoffmann, G.: Isotopic composition
Delmotte, V., Butler, P. G., Khodri, M., and Séférian, R.: and origin of polar precipitation in present and glacial climate
Bidecadal North Atlantic ocean circulation variability con- simulations, Tellus B, 53, 53–71, 2001.
trolled by timing of volcanic eruptions, Nat. Commun., 6, 6545, Werner, M., Langebroek, P. M., Carlsen, T., Herold, M., and
https://doi.org/10.1038/ncomms7545, 2015. Lohmann, G.: Stable water isotopes in the ECHAM5 gen-
Thomas, E. R. and Abram, N. J.: Ice core reconstruc- eral circulation model: toward high-resolution isotope model-
tion of sea ice change in the Amundsen-Ross Seas

ing on a global scale, J. Geophys. Res.-Atmos., 116, D15109, York, D., Evensen, N. M., Lopez Martinez, M., and De Basabe Del-
https://doi.org/10.1029/2011JD015681, 2011. gado, J.: Unified equations for the slope, intercept, and standard
Winstrup, M., Vallelonga, P., Kjær, H. A., Fudge, T. J., Lee, J. errors of the best straight line, Am. J. Phys., 72, 367–375, 2004.
E., Riis, M. H., Edwards, R., Bertler, N. A. N., Blunier, T.,
Brook, E. J., Buizert, C., Ciobanu, G., Conway, H., Dahl-Jensen,
D., Ellis, A., Emanuelsson, B. D., Keller, E. D., Kurbatov, A.,
Mayewski, P., Neff, P. D., Pyne, R., Simonsen, M. F., Svens-
son, A., Tuohy, A., Waddington, E., and Wheatley, S.: A 2700-
year annual timescale and accumulation history for an ice core
from Roosevelt Island, West Antarctica, Clim. Past Discuss.,
https://doi.org/10.5194/cp-2017-101, in review, 2017.

Antarctic Climate Variability On Regional and Cont

Uploaded by

Copyright:

Available Formats

Antarctic Climate Variability On Regional and Cont

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Antarctic Climate Variability On Regional and Cont

Uploaded by

Copyright:

Available Formats

Clim.

Past, 13, 1609–1634, 2017

Antarctic climate variability on regional and continental scales

91191 Gif-sur-Yvette CEDEX, France

38041 Grenoble, France

1348 Louvain-la-Neuve, Belgium

9037, Tromsø, Norway

Correspondence to: Barbara Stenni ([email protected])

Received: 28 February 2017 – Discussion started: 22 March 2017

Published by Copernicus Publications on behalf of the European Geosciences Union.

Abstract. Climate trends in the Antarctic region remain 1 Introduction

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Geographic region Slope (◦ C ‰−1 ) Slope (‰ ◦ C−1 ) r p value

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

To facilitate comparison with the preceding Antarctica2k

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Pre-1900 CE trends Last-100-year trends

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

ECHAM NB2014 Borehole

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

ECHAM NB2014 Borehole

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

www.clim-past.net/13/1609/2017/ Clim. Past, 13, 1609–1634, 2017

Clim. Past, 13, 1609–1634, 2017 www.clim-past.net/13/1609/2017/

You might also like