See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/248809565

Derivation of δ 18 O from sediment core log data: Implications for millennial-

scale climate change in the Labrador Sea

Article  in  Paleoceanography and Paleoclimatology · October 2001

DOI: 10.1029/2000PA000560

CITATIONS READS
16 433

7 authors, including:

Michael E Weber Larry A Mayer

University of Bonn University of New Hampshire
381 PUBLICATIONS   2,076 CITATIONS    965 PUBLICATIONS   7,818 CITATIONS   

SEE PROFILE SEE PROFILE

Claude Hillaire-Marcel Guy Bilodeau

Université du Québec à Montréal Laval University
476 PUBLICATIONS   14,119 CITATIONS    25 PUBLICATIONS   2,122 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Volatiles in the mantle. View project

IODP Expedition 354: Bengal Fan View project

All content following this page was uploaded by Michael E Weber on 30 April 2014.

The user has requested enhancement of the downloaded file.

 PALEOCEANOGRAPHY, VOL. 16, NO. 0, PAGES 1 – 12, MONTH 2001

Derivation of d 18O from sediment core log data:

Implications for millennial-scale climate change
in the Labrador Sea
M. E. Weber,1 L. A. Mayer,1,2 C. Hillaire-Marcel,3 G. Bilodeau,3
F. Rack,4 R. N. Hiscott,5 and A. E. Aksu5

Abstract. Sediment core logs from six sediment cores in the Labrador Sea show millennial-scale climate variability
during the last glacial by recording all Heinrich events and several major Dansgaard-Oeschger cycles. The same millennial-
scale climate change is documented for surface water d18O records of Neogloboquadrina pachyderma (left coiled); hence
the surface water d18O record can be derived from sediment core logging by means of multiple linear regression, providing a
paleoclimate proxy record at very high temporal resolution (70 years). For the Labrador Sea, sediment core logs contain
important information about deepwater current velocities and also reflect the variable input of ice-rafted debris from
different sources as inferred from grain-size analysis, the relation of density and P wave velocity, and magnetic
susceptibility. For the last glacial, faster deepwater currents, which correspond to highs in sediment physical properties,
occurred during iceberg discharge and lasted from several centuries to a few millennia. Those enhanced currents might have
contributed to increased production of intermediate waters during times of reduced production of North Atlantic Deep
Water. Hudson Strait might have acted as a major supplier of detrital carbonate only during lowered sea level (greater ice
extent). During coldest atmospheric temperatures over Greenland, deepwater currents increased during iceberg discharge in
the Labrador Sea, then surface water freshened shortly thereafter, while the abrupt atmospheric temperature rise happened
after a larger time lag of 1 kyr. The correlation implies a strong link and common forcing for atmosphere, sea surface, and
deep water during the last glacial at millennial timescales but decoupling at orbital timescales.

1. Introduction: Climate Variations During the last glacial, large volumes of freshwater were released
in the Labrador Sea by iceberg calving and subglacial meltwater outflow into the North
Atlantic every 7 – 10 kyr [Bond et al., 1992]. These events pertur-
The Labrador Sea is an important ocean basin for studying the bated deepwater circulation [Paillard and Labeyrie, 1994] and were
glacial discharge into the North Atlantic of both the Laurentide associated with ice sheet instabilities that gave rise to massive
and the Greenland ice sheets. It is also important because it is discharge of ice-rafted debris (IRD) into the Labrador Sea and the
one of the major areas where intermediate waters are renewed deposition of detrital carbonate material from the Hudson Strait
by winter mixing [Lazier, 1988] and the formation of Labrador [Andrews and Tedesco, 1992; Dowdeswell et al., 1995]. Icebergs
Seawater, which extends to 2 km water depth below the that left the Laurentide ice sheet moved along the Labrador coast
surface mixed layer. Temporal changes in the water mass and released large quantities of IRD that have been identified as
structure, however, are less well studied for the Labrador Sea Heinrich events [Heinrich, 1988], which are marked by light d18O
than for the open North Atlantic [Hillaire-Marcel and Bilodeua, values in planktonic foraminifera [Bond et al., 1992] and low
2000]. The modern (Holocene) circulation is characterized by an productivity in surface waters [Hillaire-Marcel et al., 1994]. Hein-
anticlockwise gyre of North Atlantic Deep Water (NADW) rich events are identified between 40°N and 60°N [Kissel et al.,
associated with a strong Western Boundary Undercurrent (>2 1999] and may result from internal oscillations of the Laurentide ice
km water depth). The major surface currents are the Labrador sheet [McAyeal, 1993]. Intense input glacial freshwater may have
Current (<1.5 km water depth), which is a near-surface, western led to reduced NADW formation during the events [Broecker et al.,
boundary current, and the northeast directed Gulf Stream south 1989]. Hillaire-Marcel et al. [1994] reported reduced NADW
of the Labrador Sea. formation for the Labrador Sea from a stable isotope study that
documents a vertically almost homogenous water column during
glacials, in contrast to strong stratification during interglacials.
1 At higher frequencies, millennial-scale air temperature fluctua-
Ocean Mapping Group, Department of Geodesy and Geomatics tions over Greenland as recorded in the isotopic signature of air
Engineering, University of New Brunswick, Fredericton, New Brunswick,
Canada.
bubbles trapped in the ice sheet [Dansgaard et al., 1993], corre-
2
Now at Center for Coastal and Ocean Mapping, University of New spond to changes in the isotopic composition of planktonic
Hampshire, Durham, New Hampshire, USA. foraminifera in North Atlantic glacial sediment [Bond and Lotti,
3
Center for Research in Isotopic Geochemistry and Geochronology, 1995]. Accordingly, the atmospheric record and the sea surface
Université du Quebec à Montréal, Montréal, Quebec, Canada. record were linked. There is also evidence for a link between
4
Joint Oceanographic Institutions, Inc., Washington, D. C., USA. atmospheric variability and paleocirculation and possibly to the
5
Earth Sciences Department, Memorial University of Newfoundland, production of NADW as indicated in sediment physical properties
St. John’s, Newfoundland, Canada. of North Atlantic sediment [Rasmussen et al., 1996; Moros et al.,
1997; Kissel et al., 1999]. Here we will use a sediment core log-
Copyright 2001 by the American Geophysical Union. derived climate proxy to provide evidence that the atmospheric
Paper number 2000PA000560. variability over Greenland is linked to both surface water and
0883-8305/01/2000PA000560$12.00 deepwater variability in Labrador Sea sediment.

1
2 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS

MD95 in 1995 with the exception of core MD99-2242, which is

2. Climate Proxy Derivation From Sediment Core from cruise MD99 in 1999. Core locations are given in Table 1.
Logs: Background Site MD95-2024 is located off the shelf of Newfoundland on the
Near continuous sediment core logs of physical and optical slope of Orphan Knoll and was taken on the same location as core
properties have become substantial and persuasive tools in paleo- 91-045-094 studied by Hillaire-Marcel et al. [1994]. It spans the
ceanography. Core logs can be gathered rapidly and at millimeter last glacial cycle and documents a sequence of eight Heinrich
to centimeter resolution. Data quality is improving and currently events. Core MD95-2025 is located 75 km to the southwest at
benefits from of increased attention paid to common calibration Orphan Basin. It is strongly influenced by the Western Boundary
and data processing procedures [e.g., Weber et al., 1997; Best and Undercurrent and spans the last three glacial cycles, documenting
Dunn, 1999]. 13 Heinrich events. Cores MD95-2024 and MD95-2025 were both
Physical properties are playing an increasingly important role in studied in great detail for stable isotopes and geochemistry. Core
stratigraphic studies of marine sediment [e.g., Shackleton et al., MD95-2026 is from the northeast flank of the Sackville Spur
1995], with high-resolution timescales being constructed by relat- sediment drift and provides insight into variations in the strength of
ing variations of physical properties to variations of, for example, the Labrador Current. Therefore it was studied for grain-size
orbital parameters or isotopic reference curves. For longer time- distribution. Core MD99-2242 is located on the Greenland Rise
scales, deep-sea cores are used as the reference section; for shorter directly in the path of the Western Boundary Undercurrent and
timescales, ice cores are used. Further application includes spectral should hence have low interglacial sedimentation rates. Core
analyses in order to study the climate response to forcing factors MD95-2028 is from Fogo Seamount, south of the Grand Banks
[Mayer et al., 1996]. Underlying this approach is the ability to of Newfoundland, and yields a record of meltwater discharges
derive and predict the variation of climatically relevant proxies from the Laurentian Channel over several glacial cycles, whereas
from sediment core logging since it is the establishment of this core MD95-2029 from the eastern levee of the Laurentian Fan and
relationship that allows the collection of climate proxy data at high core MD95-2033 from the Laurentian Margin provide the same
resolution [Mayer, 1991]. In Pacific biogenic sediment, for record during the Last Glacial Maximum and the following
instance, carbonate contains the major information about produc- deglaciation. Core MD95-2031 is from the continental slope off
tivity in surface waters and dissolution in deep waters [Pisias et al., Whale Bank, southwest of the Grand Bank, and provides a high-
1995]. Therefore a number of successful, sediment core log-based resolution record of deglaciation.
predictions exist for deep-sea carbonate content [e.g., Mayer, 1991; Density, P wave velocity, and magnetic susceptibility were
Hagelberg et al., 1995; Weber, 1998]. Further predictions have determined nondestructively onboard using a MultiSensor Core
been proposed for deep-sea carbonate grain-size distribution Logger (Geotek, United Kingdom) at 2-cm increments. This
[Mayer et al., 1993] and biogenic opal contents [Weber, 1998]. system provides three sensors: a pair of compressional wave
For continental margin settings, grain-size distribution [Weber et transducers to determine the velocity of compressional waves in
al., 1997] and magnetite content [Harris et al., 1997] have been the core (P wave velocity); a gamma ray source and detector to
derived from sediment core logs. measure the attenuation of gamma rays through the core (density);
Here we demonstrate an approach for the derivation of d18O and a magnetic susceptibility sensor loop to determine the amount
values for Neogloboquadrina pachyderma (left coiled) during the of magnetic material present in the sediment. Densities were
last glacial cycle from variations of gamma ray density (hereinafter calculated using varying attenuation coefficients and an iteration
density), compressional wave velocity (hereinafter P wave veloc- procedure described by Weber et al. [1997]. In addition, a Minolta
ity), magnetic susceptibility, and sediment color (gray scale) chromatometer was used to measure gray scale at 5-cm increments
measured on sediment cores retrieved on the slope off Atlantic (for method, see Weber [1998]). For core MD95-2024, carbonate,
Canada. The Labrador Sea cores of this study yield a detailed stable isotopes, and organic carbon were analyzed at 5-cm incre-
record of the late Quaternary response of the eastern Laurentide ice ments at Center for Research in Isotopic Geochemistry and Geo-
sheet to sub-Milankovitch climate forcing. Therefore millennial- chronology (GEOTOP), Montreal, using the procedures described
scale climate signals such as Heinrich events [Bond et al., 1992] by Hillaire-Marcel et al. [1994]. For core MD95-2025, stable
and Dansgaard-Oeschger cycles [Dansgaard et al., 1993] can be isotopes and carbonate content were measured at Memorial Uni-
identified and studied in great detail by high-resolution physical versity of Newfoundland using the procedures described by Hiscott
and optical sediment properties. et al. [2001]. For core MD95-2026, grain size distribution was
In this paper, we will provide evidence that the surface water determined at Bedford Institute of Oceanography. For the fraction
d18O signal and variations recorded in core logs of sediment >63 mm, settling tubes were applied, and for the fraction <63 mm, a
physical properties are highly correlated for Labrador Sea sedi- Sedigraph was used.
ment, implying a strong link that allows the use of sediment core Site MD95-2024 is the focus of this study since it contains both a
logs as rapid and nondestructive stratigraphic tools for studying high-resolution stratigraphy and a complete set of sediment core
millennial-scale climate variability. Furthermore, we will demon- log and sample data at high temporal resolution (70 and 210 years,
strate that variations recorded in sediment core logs provide respectively). The directly determined stratigraphy of this site is
important information about current intensities in deep water and based on accelerator mass spectrometry (AMS) 14C dates to 30
that both deepwater and shallow water processes responded to the ka, which were measured at core 91-045-094 from the same
variable input in the amount of detrital carbonate and IRD. Argu- location [Hillaire-Marcel et al., 1994]. Sites 91-045-094 and
ments will be obtained from grain-size analysis, the relation of MD95-2024 were correlated using d18O records from N. pachy-
density and P wave velocity, and magnetic susceptibility variations. derma [Hillaire-Marcel and Bilodeua, 2000]. For the actual age
model we used the high-resolution paleointensity record of Stoner
et al. [2000], which relies on correlation of site MD95-2024 to the
3. Material, Methods, and Stratigraphy d18O record of the Greenland Ice Sheet Project 2 (GISP2) ice core.
All sample data from cores 91-045-094 and MD95-2024 are
The sediment cores presented here were collected and analyzed available at the open GEOTOP database (www.geotop.uqam.ca/
as part of the Canadian Climate System History and Dynamics geotop/paleoceanographicDatabase/eng/database.html). All
Project (CSHD) and the International Marine Global Change Study onboard nondestructive data are available at the IMAGES web
(IMAGES). Cores were collected in the Labrador Sea (Figure 1) site (www.images.cnrs-gif.fr) and a revised version including d18O
with R/V Marion Dufresne II during the first IMAGES cruise derivations will be made available within the Paleoclimate Data
 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS 3

-70 -60 -50 GISP 2 -40 -30

00 GREENLAND
30

0
200
60 60

200
MD99-2242

Nain

55 55

0
200
MD95-2024
CANADA
40
00

50 MD95- 50
Quebec St. Lawrence 2025

0
100
NF DSDP
Site
NB 609
La Chan

MD95-2026
ure ne

Ottawa Fredericton MD95-

nti l

NS 2033 MD95-
45 45
an

2031
Halifax 0m
200 m
1000 m
Boston MD95- 2000 m
2029 MD95- 3000 m
0
400 2028 4000 m
-70 -60 -50 -40 -30

Figure 1. Location map from the Labrador Sea. Cores of this study are from the International Marine Global
Change Study (IMAGES) cruise MD95, collected with R/V Marion Dufresne II in 1995. Arrow marks the path of
North Atlantic Deep Water (NADW) and Labrador Current. Directions for the Greenland Ice Sheet Project 2 (GISP2)
ice core and Deep Sea Drilling Program (DSDP) Site 609 are indicated. WBU is Western Boundary Undercurrent; NB
is New Brunswick; NS is Nova Scotia; NF is New Foundland.

Network (SEPAN) at the Alfred Wegener Institute for Polar and relate to the planktonic d18O signal. This, in turn, enables us to
Marine Research (www.pangaea.de /home/mweber/). better extract paleoclimate information from sediment core logs,
that is, to study the link between surface water processes repre-
sented by planktonic d18O and sediment core logs.
4. The d 18O Derivation Strategy for Labrador Sea Reference core MD95-2024 from Orphan Knoll is located at the
Sediment outlet of the NADW gyre into the open Atlantic (Figure 1). Core
MD95-2024 contains marine isotope stages (MIS) 1 – 5 and thus a
Derivation of the surface water d18O signal from sediment core detailed record of glacial climate history for the last 120 kyr
logs serves several purposes. First, we are able to gather important (Figure 2). Icebergs that left Hudson Strait probably moved south
paleoceanographic proxy data rapidly, at low costs, and at very along the Labrador coast before they entered the open North
high temporal resolution. Second, it helps us to explore the nature Atlantic. They passed between the continental shelf and Orphan
of sediment core logs and to understand how these log parameters Knoll and released material rich in detrital carbonate and IRD

Table 1. Core Locationsa

Core Latitude Longitude Water Depth, m Core Length, m
MD95-2024 50°130N 45°410W 3539 29.52
MD95-2025 49°470N 46°420W 3009 35.12
MD95-2026 48°140N 47°400W 878 27.91
MD95-2028 42°060N 55°450W 3368 34.20
MD95-2029 43°070N 53°160W 4156 35.10
MD95-2031 44°190N 53°440W 1570 27.72
MD95-2033 44°400N 55°370W 1412 29.68
MD99-2242 58°550N 47°070W 2895 35.36
184KL 6°330S 90°310W 4102 12.55
40KL 7°330N 85°300W 3810 8.46
a
MD cores are from the Labrador Sea (cruises MD95 in 1995 and MD99 in 1999), core 184KL is from the eastern equatorial Pacific [Weber, 1998], and
core 40KL is from the Bay of Bengal [Weber et al., 1997].
4 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS

δ18O N. pachy. [‰] Carbonate[%] Density [g/cm3] Velocity [m/s] Mag. Sus.[Si] Color L δ18O
MIS
3.5 2.5 20 40 1.7 2.1 1520 1620 300 600 38 48
0
MD95-
2 2024
Orphan 1
4
Knoll H0
H1
6
H2 2
8
H3
10
Depth [m]

H4
12 3
H5
14
H6
16 4

18 H7
5
a-c
20

22
H8 5
24 d-e
a b c d e f
26

Figure 2. Downcore variations in core MD95-2024. (a) The d18O from N. pachyderma (left coiled), (b) carbonate
content, (c) density, (d) P wave velocity, and (e) magnetic susceptibility determined using a MultiSensor Core Logger,
and (f) gray scale, determined with a Minolta chromatometer. All records are resampled at 5-cm increments using
linear interpolation for derivation purposes shown in Figure 3. H0 through H8 mark Heinrich events. Marine isotopic
stages (MIS) are indicated on the right.

while they were melting. Accordingly, Heinrich layers 1 – 8 are core log data sets the correlation with d18O varies with depth; that
identified at site MD95-2024 as negative peaks in surface water is, at specific depth intervals, d18O may correlate very well with
d18O values and as maxima in carbonate content, sediment physical one logged property but not necessarily with another one.
properties, and grain size (>125 mm content; Figure 2). We should Thus, instead of deriving the d18O signal in a more conservative
point out that we measured only bulk carbonate content, but in way from a single sediment core log, we used multiple linear
most cases, the carbonate minerals in the Labrador Sea are ice- regression to detect the contribution of individual sediment core
rafted limestone and dolostone fragments and the biogenic com- log parameters to the overall correlation to the d18O signal. This
ponent (foraminifer shells) is usually very low because of the low strategy diminishes abrupt offsets of the correlation coefficient, and
glacial productivity in this region. Peaks in bulk carbonate might the derivation still yields reliable results where individual sediment
be related to biogenic carbonate only during some interglacial core logs may fail or data gaps may occur.
periods [Hiscott et al., 2001]. First, we adjusted all sediment core log records and the d18O
All sediment core log parameters for site MD95-2024 correlate record to a common variation scale by normalizing them by their
in various ways to the d18O record of N. pachyderma (left coiled). variance. This procedure allows a better calculation of the con-
This is indicative of the close relationship among surface water tribution of individual sediment core logs to the derivation and
processes (documented by the stable isotope composition) and avoids the heavily biased contributions of the original values (e.g.,
deepwater processes such as increased current strength and/or the simple product of density and P wave velocity, the acoustic
increased turbidity current activity (documented by sediment impedance, would reflect 80 – 95% density variation, depending on
physical properties; see discussion in section 6) and chemical lithology). The control of the contribution of individual sediment
parameters (Figure 2). Light d18O values are associated with highs core logs is recommended when focusing on the study of the
in carbonate, density, P wave velocity, magnetic susceptibility, and physical process that may relate d18O and sediment core log, rather
light colors. Since different parameters were measured at different than predicting it absolutely from original sediment core log
core depths and at different spatial resolution, we first resampled values. Least squares multiple linear regression shows that P wave
all data sets at 5-cm increments by linear interpolation in order to velocity and magnetic susceptibility each contribute 31% to the
achieve a common depth scale. correlation, whereas density and gray value provide only 18%
The comparison reveals that (1) at specific depth intervals (e.g., each. We used these proportions to create a stacked record which
5 – 14 m core depth) the correlation of individual sediment core will be referred to as the ‘‘combined’’ sediment core log. Corre-
logs and d18O is very good (r is up to 0.9), (2) at short depth lation of the normalized and linearly detrended d18O record and the
intervals below 14 m core depth, correlations are weaker, and (3) normalized combined sediment core log of site MD95-2024
sediment core logs and d18O are completely out of phase (i.e., they reveals coefficients from r = 0.9 (5 – 14 m) to 0 (above 5 m),
oppose each other) above 5 m core depth. Furthermore, among the averaging r = 0.65 (Figure 3e). Only d18O values show a linear
 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS 5

Variance Variance δ18O [‰] δ18O [‰] Correlation δ18O

4 3 2 1 4 2 0 3.9 2.9 1.9 3.5 2.5 -1 0 1 MIS
0
MD95-2024
2 Orphan Knoll
1
4 H0
H1
6
H2 2
8
H3
Depth [m]

10
H4
12 3
H5
14
H6
16 4
18 H7
5
20 N. Combined a-c
22
pachyderma Log
H8 5
24
2 3 4a b c d e d-e
26

Figure 3. Derivation strategy for d18O of N. pachyderma in core MD95-2024. Four steps are required to derive the
isotopic variation from a combined sediment core log of density, P wave velocity, magnetic susceptibility, and gray
value. (a) Variance-normalized and linearly detrended d18O record of N. pachyderma (dashed line) and variance-
normalized record of the combined sediment core log (solid line). (b) Combined sediment core log converted to the
variance scale of d18O of N. pachyderma by linear regression. (c) Linear regression used to correlate the combined
sediment core log of Figure 3b to the d18O record of Figure 3b. (d) Comparison of the combined sediment core log of
Figure 3c, corrected for the linear trend of d18O and calculated on the true d18O% scale and the original d18O record of
N. pachyderm. (e) Correlation coefficient determined with moving window of 50 data points (2.5 m). The offset of the
window from one analysis to another is 1 data point (5 cm). H0 through H8 mark Heinrich events. Marine isotopic
stages (MIS) are indicated on the right.

trend (cooling) from isotopic stages 5 though 2 that had to be MD95-2024 and MD95-2025 demonstrate that a relationship can
detrended before the relation between the combined sediment core be derived between d18O values, recorded at the surface mixed
log and the isotopic record could be explored. layer of the Labrador Sea, and sediment core logs, although the
In order to calculate the true d18O% value from the combined correlation coefficients are not as high as for, for example,
sediment core log we calculated a linear regression between the carbonate predictions for biogenic open ocean environments.
normalized and detrended d18O record and the normalized com- In order to apply this relationship to the prediction of d18O
bined sediment core log of Figure 3a, resulting in a common variation from sediment core logging alone, we conducted a
variance scale for both records (Figure 3b). Then a linear regres- second set of experiments. Using site MD95-2024, d18O values
sion adjusted the combined sediment core log to the d18O variance were obtained directly from original core log values (without
scale (Figure 3c). In the next step the linear trend was applied to the scaling to variance) by least squares multiple linear regression.
combined sediment core log, which was then calculated on the true This procedure provides a prediction algorithm that can be applied
d18O% scale (Figure 3d). Derivations are quite successful with directly at other sites without further conversion. A comparison of
correlation coefficients of up to r = ÿ0.9 for many intervals below both derivation and prediction methods at site MD95-2025 reveals
5 m core depth, i.e., during the last glacial cycle. Above 5 m core that, on average, slightly higher correlation coefficients for the site-
depth, both climate proxy and sediment core log values are specific derivation (Figure 4e – g). This is not surprising consider-
decoupled during the Holocene. ing the fact that the site-specific d18O data are used to obtain the
In the next step we applied the derivation method to neighboring correlation instead of the MD95-2024 data set. Nonetheless, the
core MD95-2025, where basically a similar set of isotopic and independent prediction also yields adequate results for these
sediment core log data was obtained [Hiscott et al., 2001]. Site neighboring locations and thus proves to be a powerful tool to
MD95-2025 shows clear indications of Heinrich layers 1 – 13 estimate at high resolution the surface water isotopic variation for
(Figure 4) with a core base age of roughly 340 ka. As for the this part of the Labrador continental margin.
site-specific derivation of the d18O signal, both sediment core log The correlation suggests that the processes affecting the surface
and isotopic data of site MD95-2025 were treated with the same set of the ocean (as manifested by the variability of planktonic d18O)
of methods described for core MD95-2024 except that no linear and the processes affecting the bottom of the ocean (as expressed
detrending had to be applied to the d18O data because that site by changes in physical properties that responded to variable deep-
extends further back in time and comprises several glacials with sea current strength; see discussion in section 6) must have been
heavy d18O values. The correlation coefficient between d18O and linked during the last glacial. This link is also indicated by the fact
combined sediment core log is, on average, lower than that in core that spectral analysis identifies major Dansgaard-Oeschger fre-
MD95-2024 (r = 0.59) but is more stable through time. Sites quencies of the GISP2 ice core (1.4 – 1.5 kyr, 1 kyr, 0.75 kyr)
 6
Density [g/cm3] Velocity [m/s] Mag. Sus. [Si] Color L δ18O N. pachy. [‰] δ18O N. pachy. [‰] Correlation δ18O
1.8 2.3 1550 1650 200 350 40 45 50 4.5 3.0 4.5 3.0 -1 0 1 MIS
0
MD95-2025 1
2 H1
Orphan

WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS

2
4 Basin H2

6 H3
E 3
8 H4

10 H5 F
H6
12 4
H7
Depth [m]

14
H8 5
16 predicted

18 H9
measured
20 6

22
derived
24 7
H10
26 H11
8
28
H12
9
30
a b c d e f g 10
H13
10 20 30
CaCO3 [%]

Figure 4. Downcore variations in core MD95-2025. (a) Density, (b) P wave velocity, (c) magnetic susceptibility, and (d) gray scale
(solid line) and calcium cabonate content (dashed line). (e) The d18O variations of N. pachyderma (left coiled; dashed line) and the high-
resolution prediction of the same record from a combined sediment core log (solid line) of Figures 4a – 4c, using equations developed
independently for site MD95-2024. (f) The same N. pachyderma record (dashed line) and the site-specific d18O derivation from a
combined sediment core log (solid line) of Figures 4a – 4d. For methodology see Figure 3 and refer to text. (g) Correlation coefficient of
Figures 4e and 4f as described in Figure 3. H1 through H13 mark Heinrich events. Marine isotopic stages (MIS) are indicated on the right.
 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS 7

MD95-2024 MD95-2025 MD95-2026 MD99-2242 MD95-2029 MD95-2033

δ18O [‰] δ18O [‰] δ18O [‰] δ18O [‰] δ18O [‰] δ18O [‰]
4 3 4 3 4.5 3.5 4.0 3.3 2.6 3.2 2.2 3.5 2.5 1.5
0

2 HOLOCENE

10

12
Core Depth [m]

14

16
*
LGM
18

20 LGM
22

24

26
MIS 6/5
28 boundary
30

32 δ18O N.p.l. δ18O N.p.l. δ18O N.p.l.

34 measured predicted derived

3539 m 3009 m 878 m 2895 m 4156 m 1412 m

GREEN-
LABRADOR SLOPE ST. LAWRENCE
LAND RISE

Figure 5. The d18O variability of five sites in the Labrador Sea. Shaded curve is d18O measured in N. pachyderma
(left coiled); solid curve is the same d18O record predicted from a combined sediment core log using the reference
equation developed at site MD95-2024; dashed curve is d18O derived site specifically from a combined sediment core
log. Note that measured d18O record at site MD99-2242 (asterisk) is from twin site 90-013-012P [Hillaire-Marcel et al.,
1990], which had to be stretched for depth matching. MIS is marine isotopic stage; LGM is Last Glacial Maximum.

in both sediment core logs and d18O values. Also, times when this time (MIS boundary 6/5 is at roughly 6.5 m), and the orbitally
link was not established are indicated by offsets in the correlation induced, large-amplitude variations of the d18O signal are not well
that contain an important climate signal in itself. captured by the physical property-based prediction. Strong millen-
Two things should be pointed out. First, the derivation does not nial-scale variability (although at lower amplitude than at the
capture the huge isotopic amplitude change at MIS boundaries 6/5 Labrador continental slope) and coupling of surface- and deep-
and 2/1 which indicates that global signals such as sea level rise water proxies is again observed for Greenland Rise core MD99-
(i.e., decrease in global ice volume) affect the surface water 2242, where the glacial-to-interglacial amplitude is again partially
isotopic composition to a much greater extent than deepwater decoupled (e.g., the upper 3 m).
processes. Therefore there is strong glacial coupling at millennial Sites farther south, outside the Labrador Sea at the St. Lawrence
timescales between surface-ocean and deep-ocean properties but outlet, penetrate the last glacial and thus provide a higher-reso-
decoupling at orbital timescales. Second, freshwater peaks in the lution record of oceanic variability. There, oceanic processes must
isotopic record are overpredicted by the log of core MD95-2025 be very different because heavy d18O values correspond to highs in
compared to the log of core MD95-2024. The higher amplitude of sediment physical properties and high-amplitude features are
the deepwater signal at site MD95-2025 (3009 m) points to the missing for both surface- and deepwater proxies. Consequently,
proximity to the present high-velocity zone of NADW (2500 – the strong meltwater pulses that affected the surface and deep water
3000 m water depth according to Hillaire-Marcel et al. [1994]), in the Labrador Sea did not penetrate that area. Site-specific
whereas site MD95-2024 (3539 m) is located 500 m deeper and derivations of the d18O signal, which, of course, invert the relation
might have been less affected by velocity variations of major established for Labrador Sea sediment, show that variability is very
NADW fluctuations. low during the last glacial for both the d18O signal and the
The next step was to predict or, where necessary, to derive site- combined log; in this respect, both signals are also coupled.
specific d18O signal for all sites from the Labrador Sea that have The direct link of surface water proxies such as the isotopic
the required data sets (cores MD95-2026, MD95-2029, MD95- composition of planktonic foraminifera to atmospheric change and
2033, and MD99-2242; Figure 5). On the Labrador continental ice sheet dynamics is not necessarily given for deepwater proxies
slope, all records show the high-amplitude pattern with light d18O such as sediment core log parameters. Since both proxies appa-
values corresponding to highs in sediment physical properties as rently participated in millennial-scale climate change in the Lab-
described before. Core MD95-2026 reaches even further back in rador Sea, a physical link has to exist between the two. To explore
8 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS

Velocity [m/s]
1500 1550 1600 1650 1700 1750

4800
1750 a gravel b
4400
gravel
1700

Impedance [g/cm2/s*102]
sand 4000
Velocity [m/s]

1650 3600 sand

silt
3200
1600
opal- silt 40KL
rich mud 2800 184KL
MD95-2024
1550 mud c MD95-2025
2400
opal- MD95-2026
rich MD95-2028
1500 2000 carbonate-rich MD95-2029
MD95-2031
carbonate-rich
1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8
Density [g/cm3] Density [g/cm3]

Figure 6. Physical property relations among the CSHD cores MD95-2024, MD95-2025, MD95-2026, MD95-2028,
MD95-2029, and MD95-2031 from IMAGES cruise 101 with R/V Marion Dufresne II. (a) P wave velocity versus
density; (b and c) acoustic impedance (the product of density and P wave velocity) versus P wave velocity and versus
density, respectively. For comparison, core 40KL from the Bengal Fan [Weber et al., 1997] and core 184KL from the
equatorial Pacific [Weber, 1998] are displayed (for location, see Table 1 and Figure 1). Note that physical property
relations among sediments containing primarily terrigenous components, provide substantial information about
differences in grain size.

the nature of this link, we first studied the relation of sediment core The trough-shaped P wave velocity versus density curve
log variation and grain-size distribution. (bottom left of Figure 6a) has a minimum P wave velocity of
1480 m/s, where density ranges from 1.35 to 1.5 g/cm3. In
biogenic settings, P wave velocity increases up to 1530 m/s at
5. Density and P Wave Velocity as Facies very low densities of 1.1 g/cm3. In terrigenous settings, P wave
Indicator velocity and density may increase up to 1800 m/s and 2.8 g/
cm3, respectively, depending on grain size. Values are absolute
The relationship between density and P wave velocity is one of on a global scale, and thus lithology can be derived indirectly
the most important indirect facies indicators for unconsolidated based on this relation. Also, the biogenic and terrigenous end-
marine sediment in general [e.g., Hamilton, 1970, 1971; Mienert et members of the P wave velocity versus density relation allow
al., 1988; Mayer, 1991; Weber et al., 1997; Weber, 1998]. In cores quantitative assessment of carbonate and opal contents of bio-
containing predominantly terrigenous material, density increases genic environments [Weber, 1998] and grain-size distribution in
with P wave velocity [Wood, 1941]. Here, the relationship is terrigenous environments [Weber et al., 1997] from sediment
primarily a function of grain-size distribution at relatively high core logs.
and stable grain densities of 2.5 – 2.7 g/cm3 [Weber et al., 1997]. Accordingly, sediments from the Labrador Sea are composed
High density (2.0 g/cm3) and high P wave velocity (1600 m/s) primarily of terrigenous components with minor proportions of
values occur in coarse-grained sediment, where porosity and water biogenic carbonate and opal, and they mainly reflect varying grain
content are low. Low density (1.6 g/cm3) and low P wave sizes (Figure 6c). Using the relationship described in Figure 6,
velocity (1520 m/s) values occur in fine-grained terrigenous insight into past variations of the current regime and/or the
sediment (e.g., from the Bengal Fan, core 40KL, Figure 6), where occurrence of IRD may be provided [McCave et al., 1995]. A
porosity and water content are higher. fine-grained, hemipelagic end-member is present at all sites,
For unconsolidated biogenic sediment (e.g., the southern Atlan- whereas a coarse-grained end-member progressively coarsens from
tic [Weber et al., 1997] and the equatorial Pacific [Weber, 1998]), cores MD95-2028 and MD95-2029 (relatively fine grained and
density decreases with increasing P wave velocity (core 184KL, muddy throughout) to cores MD95-2024, MD95-2031, and MD95-
Figure 6). This relationship depends on the ratio of biogenic 2025 (silty-sandy) and finally to core MD95-2026 (gravelly).
carbonate and opal. P wave velocity is high (density is low) in Provided that grain size yields relative estimates of current
biogenic opal-rich sediment because grain density is low (2.2 – 2.4 strength, at least if not deposited as IRD, core MD95-2026 would
g/cm3) and intraporosity and rigidity of siliceous skeletons are high indicate the largest variability in current strength as well as the
[e.g., Schön, 1996]. Conversely, P wave velocity is low (density is highest average current velocities. Of course, this is an over-
high) in carbonate-rich sediment because grain density is high simplification since core MD95-2026 contains gravel as a source
(2.5 – 2.8 g/cm3) and intraporosity and shear resistance of carbonate signal, which indicates that iceberg activity and deposition is
skeletons are relatively low [Weber, 1998]. Hemipelagic sediment important, too. Grain-size distribution is, in part, a function of
shows neither a clear positive nor a clear negative relation of P water depth with finer sediment at greater water depth and lower
wave velocity and density because they consist primarily of current velocities. Accordingly, the shallowest site MD95-2026 at
terrigenous material with minor contents of biogenic components. 826 m water depth has the largest variation in grain size and is thus
 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS 9

Density [g/cm3] Velocity [m/s] Mag. Sus. [SI] Mag. Sus. [SI] Density [g/cm3] Density [g/cm3] Velocity [m/s]
2.0 2.4 2.8 1600 1800 150 300 450 150 300 450 2.0 2.4 2.8 2.0 2.4 2.8 1600 1800
0
solid = MD95-
top 2026
4
Sackville
Spur
Core Depth [m]

12

16

dashed =
20 bottom

a b c d e f g
24
5.7 5.4 5.7 5.4 5.7 5.4 60 40 20 20 40 10 20 30 0 10 20 30
Mean Sort Silt [ Phi] Clay [%] Sortable Silt [%] CaCO3 [%] Gravel [%]

Figure 7. Downcore variations in core MD95-2026. Solid curves (for legend, see top axis) are sediment core log
measurements; dashed lines are grain-size parameters (for legend, see bottom axis) and carbonate content. Note that
the mean grain size of the sortable silt fraction (Figures 7a – 7c) and clay content (Figure 7d) are well correlated to
sediment core logs, whereas silt content (Figure 7e) and, especially, carbonate content (Figure 7f) and gravel content
(Figure 7g) show no or only weak correlation to sediment core logss.

the logical choice to explore the relationship of grain size and increased not only in the Labrador Current but also in the Western
sediment core log data in detail (Figure 7). Boundary Undercurrent or, alternatively, that the Labrador Current
deepened with increased production of Labrador Seawater. The
latter possibility is also indicated by the isotopic study of fora-
6. Possible Link of Surface- and Deepwater minifer assemblages by Hillaire-Marcel and Bilodeua [2000].
Processes They show that glacial periods generally have reduced production
of intermediate waters, and the Holocene as well as the short
Core MD95-2026 is from Sackville Spur, a sediment drift that is interstadials which follow Heinrich events and the successive
located in intermediate waters. The age control for the younger part freshening of surface waters show enhanced production of inter-
is rather limited since there is only one date at 6 m core depth (52 mediate waters, comparable to the present-day situation.
ka), which is beyond the reliability of the 14C method. However, on As for the timing of events, we should point out that according to
the basis of the isotopic record (Figure 5), the MIS boundary 6/5 is the high-resolution paleointensity stratigraphy of Stoner et al.
at roughly 6.5 m core depth. One important grain-size parameter is [2000] for site MD95-2024 (Figure 8), magnetic susceptibility
the mean grain size (given in 0Phi) of the size fraction ‘‘sortable and detrital carbonate peaks, which are associated with most
silt’’ (63-10 mm) which reflects most confidently the current Heinrich events, match peak cold times in the GISP2 ice core
strength with stronger currents associated with coarser sizes (lower record. At site MD95-2024, sediment physical properties vary in
 values) [McCave et al., 1995]. In core MD95-2026, coarser phase, whereas surface water d18O lags slightly behind. Coherency
mean grain sizes of the sortable silt fraction are clearly associated is established between the two proxies at various millennial
with higher densities, velocities, and magnetic susceptibilities frequencies, indicating a 10° – 40° phase shift that translates into,
(Figures 7a – 7c); that is, physical properties at site MD95-2026 e.g., 70 – 230 year lag time for surface water d18O relative to
clearly trace variations in current strength faster currents at higher sediment core log for the major Dansgaard-Oeschger period of
values. Neither gravel nor detrital carbonate content, both of which 1500 years. Thus, during coldest atmospheric temperatures over
contain direct information about the source and amount of material Greenland, deepwater currents increased during iceberg discharge
released by icebergs, correlates in detail to the sediment core log in the Labrador Sea, then surface water freshened shortly after,
measurements (Figures 7e and 7f). This is a further indication that while the abrupt atmospheric temperature rise, which introduced
sediment core log signals are governed by current activity rather the next interstadial over Greenland, occurred after a larger time
than by iceberg activity. Faster deepwater currents during times of lag of 1 kyr.
increased iceberg discharge argue for either enhanced turbidity Whether or not increased deepwater currents in the Labrador Sea
current activity or enhanced production of Labrador Seawater. during peak cold times (mainly, Heinrich events) and into the
Highs in physical properties that correspond to times of faster following interstadial interval might have been linked to enhanced
currents at site MD95-2026 are associated with freshening of production of NADW is unclear. Oppo and Lehman [1995] showed
surface water at neighboring deepwater sites MD95-2024 (Figure that the production of NADW has generally been reduced during
8) and MD95-2025. Although grain size was not studied in detail at the last glacial. At the Faeroe Margin, a strategic site for monitor-
core MD95-2024, it is possible that during iceberg discharge, most ing the production of NADW, peak cold times showed reduced
significantly during Heinrich events, current velocities were NADW and increased intermediate water influence on faunal
10 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS

δ18O [‰] N. pachy. δ18O [‰] Carbonate [%] Density [g/cm3] Velocity [m/s] Mag. Sus [Si] > 125 µm [g] δ18O
-42 -38 3.6 2.6 20 40 1.7 2.0 1530 1590 500 800 0.6 1.2 MIS
0
GISP2 MD95- Orphan
2024 Knoll 1
10
H0
H1
20 2
H2
30 H3

40 910 H4
11 3
12
13
Age [ka]

50
1514 H5
1617
60 H6
18

70 19
4
20
H7
80
21
90
5
100

110 H8

a b c d e f g
120

Figure 8. Record of millennial-scale climate change during the last glacial cycle, determined at the summit of
Greenland (GISP2 ice core, Figure 8a [Grootes et al., 1993]) and in the Labrador Sea (core MD95-2024, Figures 8b –
8h). (a) The d18O of air bubbles trapped in ice, (b) density, (c) P wave velocity, (d) magnetic susceptibility, (e) d18O of
N. pachyderma (left coiled), (f) carbonate content, and (g) amount of sand >125 mm. H1 through H8 mark Heinrich
events. Italic numbers in Figure 8a mark Dansgaard-Oeschger cycles. Marine isotopic stages (MIS) are indicated on
the right.

assemblages [Rasmussen et al., 1996]. Therefore the contribution during stadials and Heinrich events. Although Labrador Sea sites
of the Labrador Sea to the formation of intermediate waters seems were not included in their study, their findings would argue
more likely. Labrador Seawater is currently transported from its against an impact of increased deepwater currents during Heinrich
formation region in the south central Labrador Sea, to the east, events on the production of NADW; instead, there would be
where it becomes stratified and finally involved in convective indirect evidence that the intermediate water production was
mixing in the Irminger Basin as part of the formation of NADW enhanced.
[e.g., Sy et al., 1997]. An eastern transport path has to be assumed Detrital carbonate and magnetic susceptibility show negative
for the last glacial as well because intensified deepwater currents correlation for many North Atlantic sites [Kissel et al., 1999;
are not documented for the sites south of the Labrador Sea (cores Rasmussen et al., 1996], including some sites from the Labrador
MD95-2029 and MD95-2033). Sea reported by Andrews and Tedesco [1992]. This also means that
For the Labrador Sea, further evidence for increased deepwater Heinrich layers are low in magnetite [e.g., Moros et al., 1997],
currents during freshwater injection into surface waters is provided partly because of simple dilution by detrital carbonate and partly
by magnetic susceptibility, which mainly describes the amount of because the IRD carried a low-magnetite signal. However, most
magnetite in sediments [Stoner et al., 1996]. If neither the source of sites used in our study show a positive and in-phase correlation of
sediment supply nor the current intensity in deep water changes, detrital carbonate and magnetic susceptibility (e.g., Figure 2),
magnetic susceptibility should be higher in fine-grained sediment pointing to an additional source signal. Detrital carbonate-rich
because magnetite is usually concentrated in the very fine silt and bedrock is located at Hudson Strait [Andrews and Tedesco,
clay fractions. In core MD95-2026, magnetic susceptibility and 1992], whereas siliciclastic bedrock that is richer in magnetite,
clay content are inversely correlated (Figure 7d). Thus changing e.g., gneisses, is located at the coastal Labrador Sea and at Baffin
source areas with high (e.g., basaltic rocks, volcanic regions) and Bay [Hiscott et al., 2001]. Interestingly, Heinrich layers 8 and 7,
low (sedimentary basins) amounts of magnetite and/or variations in both of which were deposited at higher sea level at site MD95-
current intensity caused variations in the signal. 2024 (Figure 8), are low in detrital carbonate, whereas younger
Kissel et al. [1999] studied magnetic susceptibility in a variety Heinrich events are associated with distinct peaks in detrital
of cores that trace the path of NADW between the Nordic Sea and carbonate. Therefore ice dynamics incorporating Hudson Strait as
the Bermuda Rise. They concluded that varying magnetic inputs a major sediment supplier might have only operated at lower sea
determine the signal, i.e., basaltic material (higher susceptibility) level (greater ice extent). This interpretation is corroborated by site
prevailed during interstadials, while continental material (lower MD95-2025, which covers three glacial cycles. There, Heinrich
susceptibility) dominated during stadials and Heinrich events. layers 9, 10, and 12, all of which were deposited at higher sea
Thus periodic changes in the efficiency of transport should explain level, are also low in detrital carbonate (Figure 4), whereas the
changes in the relative amount of magnetic material with stronger Heinrich layers of glacial MIS 6 [see Hiscott et al., 2001] are high
bottom currents during interstadials and weaker bottom currents in detrital carbonate.
 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS 11

Hillaire-Marcel et al. [1994] reported that layers rich in detrital For the Labrador Sea, sediment core logs contain important
carbonate are, in part, decoupled from IRD deposition. Stoner et al. information about deepwater current velocities and also reflect
[1996] noted that the correspondence of detrital carbonate content the variable input of IRD from different sources as inferred from
with increased grain size of well-sorted magnetite, but not with grain-size analysis, benthic d18O, the relation of density and P
coarse fraction, precludes ice rafting as the primary depositional wave velocity, and magnetic susceptibility. The grain-size dis-
mechanism. They proposed that the ice advances that produced the tribution shows that highs in physical properties correspond to
IRD in the twin core of MD95-2024 (site 91-045-094) also coarser mean grain sizes in the sortable silt fraction, a clear
triggered turbiditic flows down the North Atlantic Mid-Ocean indication of faster deepwater currents. During coldest atmos-
Channel (NAMOC), where material rich in detrital carbonate pheric temperatures over Greenland, deepwater currents
was transported and subsequently deposited at site MD95-2024 increased during iceberg discharge in the Labrador Sea, then
from suspension clouds. Therefore one possible link of surface surface water freshened shortly thereafter. The abrupt atmos-
water and deepwater variability would be ice sheet advance, pheric temperature rise over Greenland, which is associated with
triggering turbidity currents in the deep ocean and meltwater the beginning of an interstadial, happened after a larger time lag
injection into the surface ocean. However, the fact that the of 1 kyr. The correlation implies a strong link and common
deposition of these layers took 0.3 – 1.3 kyr, based on excess forcing for atmosphere, sea surface, and deep water for the last
230
Th [Francois and Bacon, 1994] and AMS 14C data [Hillaire- glacial at millennial timescales. Times of noncorrelation between
Marcel et al., 1994], favors continuing deep-sea currents rather sediment core log and surface water d18O occur on orbital
than a distinct event as transport mechanism. Also, the shapes of timescales and contain a climate signal of decoupling of forcing
density and P wave velocity curves, which provide a grain-size factors; for example, during times of substantial sea level rise,
proxy (see section 5), do not resemble typical turbidites with a surface water d18O values changed dramatically, whereas deep-
coarser-grained lag followed by a fining upward bed; instead, a water properties did not.
variety of shapes is observed from plateau-like highs to sym- Contrary to many North Atlantic sites, Labrador Sea Heinrich
metrical peaks and fining upward beds. Therefore our conclusion is events usually show high magnetic susceptibility associated with
that times of increased iceberg discharge led to increased deep- high detrital carbonate content, pointing to different sources and
water currents that continued over hundreds of years to millennia, transport mechanisms. Heinrich layers deposited at higher sea level
most likely as contour-following currents along the western Lab- are relatively low in detrital carbonate, whereas those Heinrich
rador slope. layers deposited at lower sea level are associated with distinct
Figure 8 shows that variations in the glacial surface water peaks in detrital carbonate. Therefore Hudson Strait, as a major
isotopic composition in the Labrador Sea were linked to atmos- sediment supplier, might have only been active at lower sea level
pheric processes documented in Greenland ice on millennial time- (greater ice extent). The correspondence of detrital carbonate with
scales. It also shows that deepwater current velocities are linked to increased grain size of well-sorted magnetite but not with the
freshwater pulses during peak cold times. Thus deepwater varia- coarse fraction precludes ice rafting as the primary depositional
bility is also linked to atmospheric variability. This implies that all mechanism. Instead, enhanced bottom currents that may have
three parts of the climate system corresponded to the same sub- lasted several hundred years to a few millennia are associated with
Milankovitch climate forcing during the last glacial. The nature of Heinrich events.
this forcing is as yet unknown, and further research will have to be Variations in current strength along the west coast of the
conducted in a variety of fields in order to unravel it. Labrador Sea should be associated with the production of NADW
because most of the sites investigated are located within the
present-day strong Western Boundary Undercurrent. The fact that
7. Summary and Conclusions we infer faster deepwater currents during coldest air temperatures
over Greenland seems to contradict observations that at least
Records of surface water isotopic composition in the Labrador during the stadials of the last glacial period, NADW production
Sea show millennial-scale climate change signals during the last should have been reduced. However, there are several lines of
glacial cycle, all Heinrich events, and several major Dansgaard- evidence that increased current strength might have been associ-
Oeschger cycles. The same millennial-scale climate change is ated with increased production of intermediate waters in the
documented for deepwater processes by peaks in physical and
Labrador Sea.
optical sediment properties that mainly correspond to changes in
sediment composition. Heinrich events are marked by light d18O
values of N. pachyderma and highs in density, P wave velocity, and Acknowledgments. We wish to thank the reviewers M. Lyle, J.
magnetic susceptibility. Thus either the variation of surface d18O Mienert, and C. Kissel for their helpful suggestions and comments. We
values can be predicted from sediment core logging using the are also grateful to David Piper for providing grain-size data from core
procedure developed at site MD95-2024, or it can be derived by MD95-2026. This study is a contribution to the Climate System History
and Dynamics project (CSHD), supported by the National Science and
incorporating site-specific d18O measurements as shown for site Engineering Research Council of Canada (NSERC). M.E.W. received
MD95-2025. Both prediction and derivation provide a paleoclimate additional support from the Deutsche Forschungsgemeinschaft (DFG; grant
proxy record at very high temporal resolution (several decades). We 2039/2-1).

References
Andrews, J. T., and K. Tedesco, Detrital carbo- icebergs into the North Atlantic Ocean during ice sheet during the Younger Dryas cold epi-
nate-rich sediments, northwest Labrador Sea: the last glacial period, Nature, 360, 245 – 249, sode, Nature, 34, 318 – 321, 1989.
Implications for ice-sheet dynamics and ice- 1992. Dansgaard, W., Evidence for general instability
berg rafting Heinrich events in the North Bond, G. C., and R. Lotti, Iceberg discharges of past climate from a 250 kyr ice core record,
Atlantic, Geology, 20, 1087 – 1090, 1992. into the North Atlantic on millennial time Nature, 364, 218 – 220, 1993.
Best, A. I., and D. E. Dunn, Calibration of mar- scales during the last glaciation, Science, Dowdeswell, J. A., M. A. Maslin, J. T. Andrews,
ine sediment core loggers for quantitative 267, 1005 – 1010, 1995. and I. N. McCave, Iceberg production, debris
acoustic impedance studies, Mar. Geol., 160, Broecker, W. S., J. P. Kennett, B. P. Flower, J. T. rafting, and the extend and thickness of Hein-
137 – 146, 1999. Teller, S. Trumbore, G. Bonani, and W. Woel- rich layers (H1, H2) in North Atlantic sedi-
Bond, G., Evidence for massive discharges of fli, Routing of meltwater from the Laurentide ments, Geology, 23, 301 – 304, 1995.
 12 WEBER ET AL.: DERIVATION OF d18O FROM LABRADOR SEA CORE LOGS

Francois, R., and M. P. Bacon, Heinrich events in Lazier, J. R., Temperature and salinity changes sites, Proc. Ocean Drill. Program Sci. Results,
the North Atlantic: Radiochemical evidence, in the deep Labrador Sea 1962-1986, Deep 138, 289 – 319, 1995.
Deep Sea Res., Part I, 41, 315 – 334, 1994. Sea Res., Part I, 35, 1247 – 1253, 1988. Stoner, J. S., J. E. T. Channell, and C. Hillaire-
Grootes, P. M., M. Stuiver, J. W. C. White, S. Mayer, L. A., Extraction of high-resolution car- Marcel, The magnetic signature of rapidly de-
Johnsen, and J. Jouzel, Comparison of oxygen bonate data for palaeoclimate reconstruction, posited detrital layers from the deep Labrador
isotope records from the GISP2 and GRIP Nature, 352, 148 – 150, 1991. Sea: Relationship to North Atlantic Heinrich
Greenland ice cores, Nature, 366, 552 – 554, Mayer, L. A., E. Jansen, J. Backman, and T. layers, Paleoceanography, 11, 309 – 325, 1996.
1993. Takayama, Climatic cyclicity at Site 806: The Stoner, J. S., J. E. T. Channell, C. Hillaire-Marcel,
Hagelberg, T. K., N. G. Pisias, L. A. Mayer, N. J. GRAPE record, Proc. Ocean Drill. Program and C. Kissel, Geomagnetic paleointensity and
Shackleton, and A. C. Mix, Spatial and tem- Sci. Results, 130, 623 – 639, 1993. environmental record from Labrador Sea core
poral variability of late Neogene equatorial Pa- Mayer, L. A., C. Gobrecht, and N. G. Pisias, MD95-2024: Global marine sediment ice core
cific carbonate: Leg 138, Proc. Ocean Drill. Three-dimensional visualization of orbital for- chronostratigraphy for the last 110 kyr, Earth
Program Sci. Results, 138, 321 – 336, 1995. cing and climatic response: Interactively ex- Planet. Sci. Lett., 183, 161 – 177, 2000.
Hamilton, E. L., Sound velocity and related ploring the pacemaker of the ice ages, Geol. Sy, A., M. Rhein, J. R. N. Lazier, K. P. Kolter-
properties of marine sediments, North Pacific, Rundsch., 85, 505 – 512, 1996. mann, J. Meincke, A. Putzka, and M. Bersch,
North Pacific, J. Geophys. Res., 75, 4423 – McAyeal, D. R., Binge/purge oscillations of the Surprisingly rapid spreading of newly formed
4447, 1970. Laurentide ice sheet as a cause of North Atlan- intermediate waters across the Atlantic Ocean,
Hamilton, E. L., Prediction of in-situ acoustic tic’s Heinrich events, Paleoceanography, 8, Nature, 386, 675 – 679, 1997.
and elastic properties of marine sediments, 775 – 784, 1993. Weber, M. E., Estimation of biogenic carbonate
Geophysics, 36, 266 – 284, 1971. McCave, I. N., B. Manighetti, and S. G. Robin- and opal by continuous non-destructive mea-
Harris, S. E., A. C. Mix, and T. King, Biogenic son, Sortable silt and fine sediment size/com- surements in deep-sea sediments from the east-
and terrigenous sedimentation at Ceara Rise, position slicing: Parameters for paleocurrent ern equatorial Pacific, Deep Sea Res., Part I,
western tropical Atlantic, supports Pliocene- speed and paleoceanography, Paleoceanogra- 45, 1955 – 1975, 1998.
Pleistocene deepwater linkage between hemi- phy, 10, 593 – 610, 1995. Weber, M. E., F. Niessen, G. Kuhn, and M. Wie-
spheres, Proc. Ocean Drill. Program Sci. Re- Mienert, J., W. B. Curry, and M. Sarnthein, So- dicke, Calibration and application of marine
sults, 154, 331 – 345, 1997. nostratigraphic records from equatorial Atlan- sedimentary physical properties using a mul-
Heinrich, H., Origin and consequences of cyclic tic deep-sea carbonates: Paleoceanographic ti-sensor core logger, Mar. Geol., 136, 151 –
ice rafting in the northeast Atlantic Ocean dur- and climatic relationships, Mar. Geol., 83, 172, 1997.
ing the past 130,000 years, Quat. Res., 29, 9 – 20, 1988. Wood, A. B., A Textbook of Sound, G. Bell, Lon-
142 – 152, 1988. Moros, M., R. Endler, K. S. Lackschewitz, H. J. don, 1941.
Hillaire-Marcel, C., and G. Bilodeua, Instabil- Wallrabe-Adams, J. Mienert, and W. Lemke,
ities in the Labrador Sea water mass structure Physical properties of Reykjanes Ridge sedi-
during the last climatic cycle, Can. J. Earth ments and their linkage to high resolution
Sci., 37, 795 – 809, 2000. Greenland Ice Sheet Project 2 ice core data, ÿÿÿÿÿÿÿÿÿÿÿ
Hillaire-Marcel, C., A. Rochon and On-board Paleoceanography, 12, 687 – 695, 1997. A. E. Aksu and R. N. Hiscott, Earth Sciences
Participants, The Labrador Sea, CSS-Hudson Oppo, D. W., and S. J. Lehman, Suborbital time- Department, Memorial University of Newfound-
cruise 90-013 report and onboard studies, open scale variability of North Atlantic Deep Water land, St. John’s, Newfoundland, A1B 3X5
file report, Bedford Inst. of Oceanogr., Dart- during the past 200,000 years, Paleoceanogra- Canada. (aaksu@sparky2.esd.mun.ca; rhiscott@
mouth, N. S., Canada, 1990. phy, 10, 901 – 910, 1995. sparky2.esd.mun.ca)
Hillaire-Marcel, C., A. de Vernal, G. Bilodeau, Paillard, D., and L. Labeyrie, Role of the ther- G. Bilodeau and C. Hillaire-Marcel, GEOTOP,
and G. Wu, Isotope stratigraphy, sedimentation mohaline circulation in the abrupt warming Université du Quebec à Montréal, C.P. 8888,
rates, deep circulation, and carbonate events in after the Heinrich events, Nature, 372, 162 – Montréal, H3C 3P8 Canada. (bilodeau.guy@
the Labrador Sea during the last 200 ka, Can. 164, 1994. uqam.ca)
J. Earth Sci., 31, 63 – 89, 1994. Pisias, N. G., L. A. Mayer, T. R. Janecek, A. L. A. Mayer, Center for Coastal and Ocean
Hiscott, R. N., E. A. Aksu, P. Mudie, and D. F. Palmer-Julson, and T. H. von Andel, Proceed- Mapping, University of New Hampshire, Jere
Parsons, A 340,000 year record of ice rafting, ings of the Ocean Drilling Program, Scientific Chase Ocean Engineering Bldg., Durham, NH
palaeoclimatic fluctuations, and shelf-crossing Results, vol. 138, 960 pp., Ocean Drill. Pro- 03824, USA. (lmayer@unh.edu)
glacial advances in the southwestern Labrador gram, College Station, Tex., 1995. F. Rack, Joint Oceanographic Institutions, Inc.,
Sea, Global Planet. Change, 28, 227 – 240, Rasmussen, T. L., E. Thomsen, T. C. E. van 1755 Massachusetts Ave, NW, Suite 800, Wa-
2001. Weering, and L. Labeyrie, Rapid changes in shington, DC 20036, USA. (frack@brook.edu)
Holler, P., Geotechnical properties of Antarctic surface and deep water conditions at the Faer- M. E. Weber, Ocean Mapping Group, Depart-
deep sea sediments, Meteor Forschungser- oe Margin during the last 58,000 years, Paleo- ment of Geodesy and Geomatics Engineering,
geb., Reihe C, 39, 23 – 36, 1985. ceanography, 11, 757 – 771, 1996. University of New Brunswick, P. O. Box 4400,
Kissel, C., C. Laj, L. Labeyrie, T. Dokken, A. Schön, J. H., Physical Properties of Rocks: Fun- Fredericton, New Brunswick, E3B 5A3 Canada.
Voelker, and D. Blamart, Rapid climatic var- damentals and Principles of Petrophysics, 583 (mweber@unb.ca)
iations during marine isotopic stage 3: Mag- pp., Pergamon, New York, 1996.
netic analysis of sediments from Nordic Seas Shackleton, N. J., S. Crowhurst, T. Hagelberg, N. (Received June 30, 2000;
and North Atlantic, Earth Planet. Sci. Lett., G. Pisias, and D. A. Schneider, A new late revised April 9, 2001;
171, 489 – 502, 1999. Neogene time scale: Application to Leg 138 accepted May 10, 2001.)

View publication stats