<patrick.blaser@unil.ch>
ABSTRACT (246 w.)
Today, the sub-surface Denmark Strait Overflow (DSO) and the Iceland
Scotland Overflow form the starting points of Atlantic Meridional
Overturning Circulation and compensate for the poleward flowing
Norwegian and Irminger branches of the North Atlantic surface current
that drive the ’Nordic Heat Pump’. During peak glacial and early
deglacial times, ice sheets on Iceland and Greenland, and ice-induced
isostatic and eustatic sea-level changes reduced the Denmark Strait
aperture and DSO. Nonetheless, extremely high benthic stable carbon and
oxygen isotope values together with very high ventilation ages of bottom
waters reflect a north-south density gradient of intermediate-waters and
persistent flow of partially Arctic-sourced waters through both the
Denmark Strait and the Faeroe Channel, similar to today. The first
arrival of Heinrich -1 meltwaters northwest of Iceland, arriving from
the southwest around 18.4 cal ka, accompanied a tipping point in DSO
circulation, documented by reduced ventilation and ventilation ages, a
3°C warming, and increased radiogenic Nd isotope signatures in sediments
at luff-side Site PS2644. These records suggest a sudden subsurface
incursion of Atlantic intermediate waters across basaltic sediments from
S.E. of Iceland. Deep-water convection off Norway then was replaced by
weak brine water formation, coeval with a breakdown of the ’Nordic Heat
Pump’ evidenced by a temperature drop on Greenland. After 16.2 cal. ka,
a major meltwater outbreak from the Barents ice shelf led to modified
Heinrich-1-style circulation until ~15.1 cal ka.
Vice-versa, the DSO intensified during interstadial and Holocene times,
then causing sediment hiatuses at site PS2644.
KEY WORDS:
Nordic Seas, Deep-water formation, Denmark Strait Overflow, Age control
of Heinrich stadial 1, Nordic Heat Pump, Atlantic Meridional Overturning
Circulation
KEY POINTS:
The Overflow of Nordic Sea’s deep waters forms the onset of Atlantic
Meridional Overturning Circulation being enhanced at interstadial times
Sediments off N. Iceland show a joint onset of deglacial meltwater
dilution and 3-degree warming of the Denmark Strait Overflow 18400 yr
ago
Deglacial changes in circulation caused Heinrich Stadial 1, a temporary
end of heat advection to European high latitudes 18400–15000 yr ago
PLAIN LANGUAGE SUMMARY (200 w.)
Differences in salt content of North Atlantic surface waters drive
variations in Nordic Seas’ overturning circulation. These form a
switchboard for changes in the oceanic heat transport to North European
high latitudes, the ’Nordic Heat Pump’, and for Atlantic Meridional
Overturning Circulation (AMOC). We deduced changes in the Nordic Seas’
overturning circulation during the peak last glacial and early deglacial
from two high-resolution marine sediment cores with centennial-scale age
resolution based on the technique of radiocarbon plateau tuning (22-15
cal. ka). Sediment data suggest that the salinity of surface water,
advected from the North Atlantic, started to drop about 18 400 years
ago. This drop accompanied a 3°C rise in temperature and a drop in
ventilation and radiocarbon ventilation age of Denmark Strait Overflow
waters feeding the AMOC. Also, it paralleled a massive rise in the
radiocarbon reservoir age of surface waters up to 2000 yr and an abrupt
breakdown of Nordic Seas’ deep-water convection off Norway. Accordingly,
Atlantic waters were replaced by less saline polar waters, marking a
breakdown of the Nordic Heat Pump and start of ’Heinrich Stadial 1’ as
reflected by a coeval cooling documented on top of the Greenland ice
sheet, lasting until ~15.1 cal. ka.
INTRODUCTION
The Denmark Strait Overflow (DSO; Fig. 1) with its exceptionally high
density plays a key role for ~50% of the Atlantic
meridional overturning circulation (AMOC) and thereby, for global
deep-ocean circulation. The DSO is closely associated with the ”Nordic
Heat Pump”, and any changes therefore have major implications for
northern hemispheric climate, moreover, it is crucial for deep-ocean
ventilation and radiocarbon uptake. Today, the flux of DSO is long-term
fairly stable with >3 Sv entering from the southern
Greenland Sea and Arctic Ocean (Brakstad et al., 2023). Further 5 Sv are
entrained into DSO during its overflow in the Irminger Sea to the south
of the Denmark Strait, where a 2.5 km deep descend acts in stabilizing
the DSO (data of Biastoch et al., 2003 and 2021; Kösters, 2004, Kösters
et al., 2005; Kaese & Oschlies, 2000, Kaese et al. 2003 and refs.
therein)
East of Faroe the DSO is paralleled by the Iceland-Scotland Overflow
Water (ISO), also enfolding ~1.4–2.4 Sv, being supplied
to the N.E. North Atlantic and subsequently, after having passed the
Charlie Gibbs Fracture Zone, into the AMOC in the western Atlantic
(Hansen et al., 2000; Sessford et al., 2019; Biastoch et al., 2021).
Important characteristics of modern DSO waters are low bottom water
temperatures near -1°C, high bottom water salinities of 34.9 psu, and
densities (σ) of 27.8-28.05 kg/m3 for Theta=2°C at the
bottle neck of the Denmark Strait (Whitehead et al., 1974; Macrander and
Käse, 2007; Haine, 2021). A slight but persistent north-south density
gradient is indicated for the Last Glacial Maximum (LGM, 19–23 thousand
calibrated yr before present; cal. ka) by δ18O records
of single specimens of epibenthic foraminifera (Millo et al., 2006) and
confirmed by the distribution pattern of εNd isotopes, a
tracer of dominantly continental or basaltic sediments overflown by
bottom waters (Larkin et al., 2022; Blaser et al., 2020).
To monitor short-term variations in the DSO over peak glacial-to-early
deglacial climate change (22 - 15 cal. ka; here ’ka’ is used for ’kilo
yr ago’ in the sense of BP = Before Present = 1950 yr AD) we now employ,
partly refine, and supplement published centennial-scale proxy records
obtained from sediment Core PS2644, a site that marks the southern
margin of the Blosseville Basin, that is the northern funnel to the
Denmark Strait north of westernmost Iceland (Fig. 1; Voelker, 1999;
Millo et al., 2006). We compare the effect and origin of long-term
climate trends and intermittent short, centennial-to-millennial-scale
episodes of major cold spells with episodes of fast climate warming,
regimes in part lasting until today. The very last major Greenland
stadial we redefined as Heinrich Stadial 1a and b (HS-1a and b), in
contrast to a redefinition of stadials proposed by Andrews and Voelker
(2018). On the basis of advanced age control (details are given below)
our definition of HS-1 differs strongly from that previously given by
Hodell et al. (2017) based on sediments from the southern subpolar North
Atlantic.
Moreover, we compare our records with proxy data and circulation models
of stadial and interstadial analogue variations over Marine Isotope
Stage 3 (MIS3) near to the northern entrance of the Denmark Strait as
proposed by Hagen and Hald (2002), Sadatzki et al. (2020), and Sessford
et al. (2018, 2019). The latter authors report on distinct changes in
MIS3 bottom water temperature, which we now try to reproduce for late
MIS2. Sessford et al. (2019) proposed pathways of brine water-induced
intermediate waters that finally might have been funneled into the DSO,
assuming a water column homogenous in the Nordic Seas down to 1500 m.
We test this model by means of a number of multiproxy records from Site
PS2644, compared to pertinent records obtained from outside on the basis
of centennial-scale age resolution (Sarnthein et al., 2020): (i) We
employ records from two neighbor sites at the western margin of the
Vøring Plateau in the eastern Nordic Seas (Fig. 1; GIK23074 and
MD95-211), where brine water formation is widely accepted to have
occurred during HS-1 (Meland et al., 2008; Waelbroeck et al., 2011).
(ii) We compare coeval paleoceanographic records obtained south of
Iceland in the northern North Atlantic (Thornalley et al., 2011; Millo
et al., 2006; Sarnthein et al., 1994).
In this way we focus on minor and major changes in the origin of early
deglacial DSO waters and related changes in the flow geometry of the
eastern Nordic Seas and of North Atlantic intermediate- and deep-water
circulation. Changes in flow geometry were possibly linked to crucial
tipping points in the composition of DSO waters, items to be constrained
in this study. The origin of these changes appears highly important for
a better understanding of past and future trends of the ”Nordic Heat
Pump” and European climate change (sensu Jansen et al., 2020;
Ditlevsen and Ditlevsen, 2023).
OCEANOGRAPHIC WATCHDOG POSITIONS of SITES PS2644 AND GIK23074
Bathymetry and modern patterns of surface and bottom water circulation
in the Denmark Strait (Fig. 1; today: ~630 m w.d.) are
detailed by Kösters et al. (2004). Macrander et al. (2007) present
details of the modern DSO structure being ruled by geostrophic forcing.
Sediment Core PS2644 lies within the lower portion of the modern plume
funneled into the DSO (Käse et al., 2003). Prior to entering the Denmark
Strait, Norwegian Sea Deep Water (NSDW) today is entrained from both the
Greenland Basin and Arctic Ocean (Brakstad etal., 2023), a scenario to
be traced back over last glacial and early deglacial times 22–15 cal.
ka.
Sediments of Core PS2644 are well suited to monitor past
glacial-to-early deglacial variations in the flow of surface and
deep-water masses since the site lies close to the Polar Front, that is
the mean position of the border of perennial sea ice, the boundary
between the East Iceland Current (EIC) and the nearshore North Iceland
Irminger Current (NIIC) (Fig. 1; Voelker, 1999), while not lying too
close to the Denmark Strait where extreme winnowing due to strong bottom
currents largely prevents any sediment deposition. To achieve an
undisturbed and largely continuous sediment record of past changes in
the geometry of ocean water masses, Site PS2644 at 777 m water depth on
top of a narrow hemipelagic sediment ridge was chosen after careful
selection by means of a high-resolution PARASOUND parametric echosounder
system (Hubberten et al., 1995). As detailed in Voelker (1999) the site
shows a sediment drift free from lateral near-bottom sediment input like
sediment slumps and/or turbidity currents (Fig. 2a; Fig. S1).
Core GIK23074 was retrieved from the western margin of the Vøring
Plateau right below the North Atlantic Current (NAC), but far away from
the Norwegian margin (Fig. 1; Voelker, 1999). The site is marked by
exceptionally high, yet undisturbed hemipelagic sedimentation rates
(25-60 cm /ky; Fig. 2b), thus providing a unique high-resolution (17-50
yr per cm sediment depth) record of ocean history in the eastern Nordic
Seas at 1157 m water depth.
AGE CONTROL and PALEOCEANOGRAPHIC PROXIES
Precise age control of marine sediment records PS2644 and
GIK23074 is essential for constraining the objectives of this study. Age
control has been based on various conventional, correlation-based age
tie points by Voelker (1999) and Voelker et al. (1998 and 2000).
Finally, however, we based our age model on the technique of14C plateau tuning (Figs. 2a and b) (Sarnthein et al.,
2020 and 2023). The different approaches are listed below:
• Distinct features in the planktic δ18O records were
correlated to apparently analogous age-calibrated δ18O
oscillations in ice core records GISP2 and NGRIP, where ages are based
on incremental time scales (Grootes and Stuiver, 1997; Svensson et al.,
2008) (Fig. S1; Voelker, 1999).
• Likewise, distinct variations in % Neogloboquadrina pachydermasin (Nps; in ”old” teminology) served as tracer of short-term SST
changes that were used as stratigraphic markers by comparison to
temperature variations dated in Greenland ice core records.
• The Vedde volcanic ash layer was used as tracer dated by land plants
at ~10.31±50 yr 14C ka (= 12.1 cal.
ka) on the basis of ambient plant macrofossils (age revised by Birks et
al., 2017; Bronk Ramsey et al., 2020).
• Initial age control was established by means of a high-resolution
suite of planktic 14C ages in Core PS2644, using the
general assignment of a hypothetical Marine Reservoir Age (MRA) of 400
yr (Voelker, 1999; Voelker et al., 1998). This assumption, however, has
now been subject to major revision (Sarnthein et al., 2015 and 2020) as
shown below.
• On top of a simple conventional stratigraphic alignment of planktic14C ages, we further refined the age control by
defining local planktic 14C age plateaus and plateau
boundaries that were tuned as cal. age tie points to pertinent
age-calibrated structures in the atmospheric 14C
record of Lake Suigetsu (Table 1; Sarnthein et al., 2023). In turn,
Suigetsu age control is based on U/Th-based model ages of Bronk Ramsey
et al., 2020 (details of age derivation in Sarnthein et al., 2020 and
2023) (Figs. 2a and b). We are aware that our approach is in conflict
with allegations of Bard and Heaton (2021) claiming that the approach is
flawed. However, we trust in 14C plateau tuning for
various crucial reasons: (i) Different from Bard and Heaton (op. cit.)
we have clearly shown that past centennial to millennial-scale
fluctuations of the atmospheric 14C record have been
authentic (Sarnthein et al. (2023). (ii) Our technique of plateau tuning
solely relies on the tuning of a whole suite of 14C
plateaus in the atmosphere and a sediment core each (Figs. 2a+b), thus
can clearly distinguish and/or exclude potential fake plateaus in a
sediment core, such as given for Core GIK23074 for the top portion of
HS-1 (Fig. 2b). (iii) Our text (Figs. 4 and 5) is demonstrating two
prominent cases of abrupt deglacial climate change during Heinrich
stadial 1 at 18.4 and 16.2 cal. ka, where the 14C
plateau-based age estimates precisely match pertinent estimates based on
the incremental age scale of the North GRIP ice core with less than 100
years deviation, results that are far from incidental and would
not be revealed by any other stratigraphic method. (iv) On the basis of
numerous details, we have refuted one-by-one the allegations of Bard and
Heaton (op.cit.), as published in the discussion section of their
article (Grootes and Sarnthein, 2021; Sarnthein and Grootes, 2021),
though ignored by Bard and Heaton.
Calendar-age uncertainties of the atmospheric 14C
plateau boundaries employed for our tuning approach hardly exceed
~50 to ~100 yr each. Local MRA of
planktic foraminifers and surface waters were derived from the age
difference between the average 14C age estimated for
paired, that is, coeval atmospheric and marine plateaus based on14C plateau tuning (Sarnthein et al., 2020, and
references therein).
• In addition, a single U/Th-based cal. age was obtained from a solitary
coral in Core MD95-2011. This age was compared with the14C ages of paired planktic and benthic foraminifera
specimens (Fig. 3), thereby largely confirming the paired age estimates
and MRA derived on the basis of 14C plateau tuning of
the planktic 14C record (unpubl. comm. of Dreger,
2000; thorium/uranium data of Lomitschka and Mangini [1999]. The
ratio of thorium to uranium in each sample, which yield the calendar age
of the coral, were measured by thermal ionisation mass spectrometry
(TIMS) at the Heidelberger Akademie der Wissenschaften (Heidelberg,
Germany) according to the method outlined by Bollhöfer et al. (1996) and
Neff et al. (1999).
• Sedimentation rates with multi-centennial time resolution were derived
from the age interpolation of sediment sections using the cal. age of
planktic 14C plateau boundaries (Fig. 2a and b). The
estimates were widely supported by means of ages independent of14C plateau tuning such as the linear age
interpolation between conventional age tie points to pertinent ages of
ice core record GISP2 (Voelker, 1999; Fig. S1).
• To a large extent, Site PS2644 lacks modern and Holocene
reference values for most proxy records employed in this study. Except
for a few cm thick 14C-dated sediment layer that forms
the actual core top, a strong DSO flow has hindered any Holocene
sediment deposition and/or led to sediment erosion prevalent over most
of the last ~15 cal. ka (Voelker, 1999; Kuijpers et al.,
2003).
• The derivation and constraints of various paleoceanographic proxy
values employed in this study are given in Supplementary Text no. 1.
PS2644: FOUR STRATIGRAPHIC TIME SEGMENTS ~22–15 cal. ka
The PS2644 sediment section representing LGM and HS-1 times starts and
ends with major stratigraphic gaps prior to
~21.9 cal. ka and subsequent to ~15.1
cal. ka (Fig. 2a). The gaps occur near the end of Greenland Interstadial
(GI) 2 and close to the end of HS1, shortly prior to the Bølling-Allerød
(B/A) interstadial that is lost by erosion or non-deposition (Fig. S1;
Voelker, 1999, and Voelker et al., 2000). This sediment gap is
documented by X-ray radiography as distinct unconformity at 54-53 cm
composite core depth, right below the Vedde Ash layer, 12.6 cal. ka
(sensu Voelker, 1999, and Voelker and Haflidason, 2015). The
stratigraphic gaps probably result from enhanced sediment winnowing due
to the constriction of an enhanced inflow to the Denmark Strait near to
its northern entrance.
This peak-glacial-to-early-deglacial sediment record of PS2644 for MIS2
is partitioned into four stratigraphic time segments I through IV that
lasted from 21.8–19.8, 19.8–18.4, 18.4–17.2, and
17.2–~15.1 cal. ka, based on epibenthic ventilation
ages, changes in the stable C and O isotope composition of planktic and
epibenthic foraminifera (Nps and mainly Cibicidoides lobatulus ),
and less distinct, planktic MRA. The partitioning is also reflected by
the εNd and Pb isotope records, moreover, by distinct
changes in sedimentation rate (Fig. 4). It is important to note that the
time segments apply to the suite of structures in the proxy records of
both the cores PS2644 near Iceland and GIK23074 from the eastern Nordic
Seas.
Time segment (I), 21.8–19.8 cal. ka, in PS22644 covers the top
portion of the Last Glacial Maximum (LGM; as defined by Mix et al.,
2001), while time segment (II) 19.8–18.4 cal. kaalready reflects the end of the LGM, as suggested by a first major
deglacial rise in eustatic sea level starting at 19.4 cal. ka (based on
purely atmospheric 14C ages of Hanebuth et al., 2009).
During segments I and II, sea ice-covered subsurface waters of the East
Greenland Current (EGC) at PS2644 (Sadatzki et al., 2020) are marked by
minimum temperature and peak salinity values as reflected by fairly
persistent planktic δ18O values of 4.5 ‰ and minimum
SST of 3.7°C based on census counts of planktic foraminifera species
(Pflaumann et al., 2003; Millo et al., 2006). Time segment I shows a
maximum MRA near 2200 yr, that slightly dropped to ~1900
yr after 19.8 cal. ka (Fig. 4).
During time segment I, bottom waters at PS2644 were marked by a
bipartite population of epibenthic δ13C values. One of
them presents the highest δ13C values and, despite all
processes of ocean mixing, the highest deep-water ventilation recorded
in the global ocean (Millo et al., 2006; Duplessy et al., 2002), clearly
predominant over 21.8–20.3 cal. ka, when bottom water ventilation ages
reached up to 2500 14C yr.
Radiogenic isotopes in the detrital sediment fraction indicate a mixture
of Arctic and European sources of the sediment, with a reduced
contribution from nearby Iceland (Fig. 4 and 5; Struve et al., 2019;
Larkin et al., 2021; Blaser et al., 2020). We hypothesize that this
could have been caused by the complete glaciation of Iceland and/or a
change in ocean currents shielding the site from Icelandic input.
Interestingly, the authigenic sediment fraction exhibited significantly
more radiogenic signatures in both Nd and Pb. A similar effect has been
observed in front of the Barents Shelf (Struve et al. 2019) and in the
eastern subpolar North Atlantic (ODP980; Crocket et al. 2013) and
attributed to the glacial erosion of terrestrial metal oxides in
Northwest Europe and their supply to the ocean.
During time segment II, after 19.8 cal. ka, the antecedent high
in bottom water ventilation age was short-term replaced by very low ages
of 100-400 yr. North of Iceland they mark an early deglacial incursion
of waters reflecting a direct contact with the atmosphere nearby,
closely resembling ages that already marked the end of GI 2, and more
important, LGM bottom water ages found at Site GIK23074 in the eastern
Nordic Seas. Moreover, we find a distinct rise in radiogenic εNd values
of the detritus up to -5, which traces an increased contribution by
Icelandic basalts passed by DSO waters prior to reaching Site PS2644
(Fig. 4), or increased supply through an early deglaciation of parts of
Iceland (sensu Crocket et al., 2012).
Time segment III, 18.4–17.2 cal. ka, reflects a phase of
transition, when planktic δ18O values at PS2644 show a
marked gradual drop by more than 1.5 per mil (Fig. 4). The drop
primarily reflects a drop in sea surface salinity induced by a lateral
advection of meltwaters from southwest, from the Irminger Sea, and thus
records the onset of HS-1. Also, foraminifera census counts reflect a
slight SST rise up to 4°C (Sarnthein et al., 2001). Over this time,
local MRA dropped to 1670-1780 yr, a value slightly lower than before,
but still reflecting a reduced carbon exchange of sub-surface waters
with the atmosphere, impeded by ongoing perennial sea ice cover. At the
eastern Site GIK23074 MRAs at the onset of time segment III showed an
impressive sudden rise from 1175 yr up to 1900 yr, a level equating that
found at the western Site PS2644 (Figs. 2b and 3).
Like surface waters, bottom waters document a major change near 18.4
cal. ka: The maxima of benthic δ18O values measured on
single epibenthic specimens reveal a rapid, centennial-scale18O depletion by 0.8 ‰, equivalent to a rise in bottom
water temperature of up to 3.4°C (Fig. 4). At this time, the data
population with extremely high epibenthic δ13C values
has disappeared in favor of medium high values, coeval with a renewed
short-term reduction of bottom water ventilation ages down to 750 yr
near 17.5 cal. ka. Once more, εNd signatures in both the detrital and
authigenic fractions have increased, indicating an elevated supply of
Icelandic volcanogenic material overflown by DSO waters prior to
reaching Site PS2644 (Fig. 4) or delivered to the site directly by
surface currents such as the North Iceland Irminger Current (Fig. 1).
The contemporaneous authigenic206Pb/204Pb -based ratios show a
pronounced short-lasting maximum (Fig. 5), which could result from an
increased delivery of Pb glacially eroded from Northwest Europe
18.4-17.0 cal. ka.
Time segment IV, 17.2–15.1 cal. ka (i.e., until the onset of a
major hiatus), presents the full maturation of the HS-1 sub-ice
meltwater regime as reflected by a persistent minimum in planktic
δ18O. Near the very onset, this interval was marked by
a short-term SST peak of 7.5°C (census counts of planktic foraminifera
species; Fig. 15c of Voelker, 1999). As in antecedent periods, MRA of
1900 years continued. By contrast, paired bottom water ages were as low
as 1100 to 1550 14C years, significantly lower than
during the LGM. As in time segment III, epibenthic
δ13C values of bottom water ventilation were modestly
high (0.6-1.4 ‰).
The onset of time segment IV shows a remarkable sudden rise in bulk
sedimentation rates at PS2644, almost by a factor of three, with
sediments marked by a high in the concentration of ice-rafted
hematite-stained quartz grains originating from (North-) East Greenland
(Voelker, 1999). At the same time, detrital206Pb/204Pb ratios decreased while
εNd became more radiogenic (Fig. 4), which would agree with an increased
delivery of Tertiary basalt sediment. On the expense of the ”European”
source (Fig. 5) the basalt signal then may have come from the nearby
Iceland-Scotland Ridge east of Iceland, overflown by the North Iceland
Jet, when heading west for Site PS2644. Also, Fig. 5 may suggest an
input of basalt debris from major basalt sources in East Greenland south
of 70°N. Conversely, however, we regard this east-west sediment
transport as unlikely, since it would need to cross the (sea-ice
covered) frontal systems of the East Greenland Current flowing toward
southwest (Fig. 1).
Altogether, MRAs and bottom water ventilation ages at Site
GIK23074 (Fig. 3) reveal a suite of changes in differential
stratification of the eastern Nordic Seas contemporary with that in the
west, at PS2644 (Fig. 1). In addition, however, time segment IV has been
dissected near 16.15 cal. ka by a subsequent major drop in both MRA and
bottom water ventilation age (Fig. 2b and 3). The drops are tied to the
great late-deglacial meltwater outbreak from the Barents Shelf as
documented by a short-term extreme low in planktic
δ18O values extending south to the Faeroe Isles
(Weinelt et al, 1991; Voelker, 1999).
DISCUSSION – LINKAGES BETWEEN CHANGING SOURCES OF DSO WATERS AND
SHORT-TERM CHANGES IN SEA SURFACE SALINITY AND CLIMATE