HAL Id: hal-03334776
https://hal.archives-ouvertes.fr/hal-03334776
Submitted on 5 Sep 2021
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Surface water temperature changes in the high latitudes
of the southern hemisphere over the Last
Glacial-Interglacial Cycle
Jean-Jacques Pichon, Laurent D Labeyrie, Gilles Bareille, Monique
Labracherie, Josette Duprat, Jean Jouzel
To cite this version:
Jean-Jacques Pichon, Laurent D Labeyrie, Gilles Bareille, Monique Labracherie, Josette Duprat, et
al.. Surface water temperature changes in the high latitudes of the southern hemisphere over the Last
Glacial-Interglacial Cycle. Paleoceanography, American Geophysical Union, 1992, 7 (3), pp.289-318.
�10.1029/92PA00709�. �hal-03334776�
PALEOCEANOGRAPHY, VOL. 7, NO. 3, PAGES 289-318, JUNE 1992
SURFACE WATER TEMPERATURE CHANGES
IN THE HIGH LATITUDES OF THE SOUTHERN
HEMISPHERE OVER THE LAST GLACIAL-
INTERGLACIAL CYCLE
Jean-Jacques Pichon, 1 Laurent D. Labeyrie, 2 Gilles
Bareille, • Monique Labracherie, 1 Josette Duprat, 1
and Jean Jouze13,4
Abstract. A set of numerical equations is
developed to estimate past sea surface temperatures
(SST) from fossil Antarctic diatoms. These
equations take into account both the biogeographic
distribution and experimentally derived silica
dissolution. The data represent a revision and
expansion of a floral data base used previously and
includes samples resulting from progressive opal
dissolution experiments. Factor analysis of 166
samples (124 Holocene core top and 42 artificial
samples) resolved four factors. Three of these
factors depend on the water mass distribution (one
Subantarctic and two Antarctic assemblages); factor
4 corresponds to a "dissolution assemblage".
Inclusion of this factor in the data analysis minimizes
the effect of opal dissolution on the assemblages and
1D6partement de G6ologie et Oc•anographie,
CNRS URA. 197, Universit6 de Bordeaux 1,
Talence, France.
2Centre des Faibles Radioactivit•s, Laboratoire
mixte CNRS-CEA, Gif-sur-Yvette, France.
3Laboratoire de Mod•lisation du Climat et de
l'Environnement, Gif sur Yvette, France
4Also at Laboratoire de Glaciologie et de
Geophysique de l'Environnement, St Martin
d'H•res, France.
Copyright 1992
by the American Geophysical Union.
Paper number 92PA00709.
0883- 8305/92/92PA-00709510.00
gives accurate estimates of SST over a wide range of
biosiliceous dissolution. A transfer function (DTF
166/34/4) is derived from the distribution of these
factors versus summer SST. Its standard error is +
IøC in the-1 to +10 øC summer temperature range.
This transfer function is used to estimate SST
changes in two southern ocean cores (43øS and 55øS)
which cover the last climatic cycle. The time scale is
derived from the changes in foraminiferal oxygen
and carbon isotopic ratios. The reconstructed SST
records present strong analogies with the air
temperature record over Antarctica at the Vostok site,
derived from changes in the isotopic ratio of the ice.
This similarity may be used to compare the oceanic
isotope stratigraphy and the Vostok time scale
derived from ice flow model. The oceanic time scale,
if taken at face value, would indicate that large
changes in ice accumulation rates occurred between
warm and cold periods.
INTRODUCTION
Paleoclimatological studies in the southern
hemisphere high latitudes are important for
understanding some of the major phenomena
controlling climatic changes over the last 103 to 105
years. The interactions between surface ocean, ice,
and atmosphere should be easier to model in the
southern than in the northern henrisphere, because of
the near axial geometry of the Antarctic continent and
the surrounding southern ocean, permanent ice cover
over Antarctica and the relative stability of the
Antarctic Polar Front (APF) during glacial-
interglacial transitions [Hays et al., 1976a]. The
Vostok ice core, which covers the last climatic cycle
290 Pichon et al.: Surface Water Temperature Changes
[Lorius et al., 1985], provides a high-resolution
deuterium and oxygen isotopic record interpretable in
terms of temperature changes over Antarctica [Jouzel
et al., 1987]. The atmospheric CO2 content during
that period has been measured in the air trapped
during ice formation [Barnola et al., 1987]. Age
models for the Vostok core and its included gas,
derived from estimated ice accumulation and flow
rates [Lorius et al., 1985] differ significantly from
the time scale developed for the ocean sediments,
with a longer apparent duration for the last
interglacial. This is a major problem for the study of
climatic feedbacks between changes in insolation,
ocean circulation, and pCO2 in the atmosphere
[Lorius et al., 1990; Sowers et al., 1991]. Petit et al.
[ 1990] proposed to use the close similarity that exists
between the changes in atmospheric dust content of
the Vostok core and available dust records coveting
the last 150 ka [Kukla, 1987; Robinson, 1986; Kent,
1982] to construct a common chronostratigraphy
between the ice and ocean sediment records.
However, there are no indications that the different
dust records are synchronous. Another approach
may be to use the close climatological connections
between the Antarctic ice sheet and the surrounding
southern ocean. The sea surface temperature (SST)
changes at high latitudes influence the temperature
over Antarctica by transfer of heat and atmospheric
water. Lorius et al. [1990] have shown the good
covariance existing between the reconstructed
changes in air temperature at Vostok and the SST
record derived from a radiolarian transfer function in
sediment core RCll-120 [Hays et al., 1976b,
Martinson et al., 1987]. Limitations in the
interpretation of this relationship derive from the
location of this sediment core north of the Antarctic
Polar Front and thus still under the strong influence
of the subtropical/transitional waters. Time lags have
been observed between southern ocean temperature
records and the global climatic signal [Hays, 1978;
Morley and Robinson, 1980; CLIMAP Project
Members, 1984; Labracherie et al., 1989], as well as
between isotopic records of planktonic and benthic
foraminifera [Labeyrie et al. 1986].
We present here a reconstruction of SST changes
over the last 140 kyr in two cores from the subpolar
(MD 84 527) and polar (MD 84 551) waters in the
southern ocean using a transfer function based upon
floral assemblages derived from that developed by
Pichon et al. [1987], with an increased core top
reference data base and a correction of the effect of
dissolution as described in Pichon et al [1992]. The
time frame is derived from the changes in
foraminiferal isotopic ratio, and the results are
compared with the changes in air temperature over
the ice estimated at Vostok [Lorius et al 1985].
DIATOM TRANSFER FUNCTION
Materials and Methods
Pichon et al. [1987]. have shown that the
geographic distribution of Antarctic flora (diatoms
and silicoflagellates) in modern core top samples
from the southern ocean coincides with present-day
sea surface parameters (summer SST, phosphate
concentration, and annual sea ice cover). The
quantitative method used here to derive the transfer
function follows Imbrie and Kipp [1971]. The aim
of the present study is to increase the range of
applicability of the diatom transfer function (DTF)
and its accuracy, particularly in view of the bias
possibly introduced by differential dissolution of the
fossil diatom assemblages. The geographic sample
coverage was increased from 28 to 124 core top
samples, thus permiting construction of more
detailed descriptions of regional biogeography
patterns and temperature gradients (from +14øC to-
1.9øC). The 124 samples from the southern ocean
consist of 77 samples from box cores, 40 samples
from high-sedimentation-rate piston cores, and seven
samples from trigger-weight cores collected between
67øW and 140 ø E and 37øS and 78øS (Table 1 and
Figure 1). In all cores, the sediment-water interface
was retrieved with no apparent disturbance and,
based on micropaleontological, sedimentological and
statistical arguments [Pichon et al., 1987], represents
TABLE 1. Positions and Water Depths of Surface Samples Used in This Study
Core
MD24-KK02
MD24-KK32
MD24-KK35
MD24-KK37
MD24-KK63
MD73-026
MD82-424
MD82-425
MD82-430
MD82-434
Latitude (S) Longitude Depth, m
54o13,0 ' 03ø31,0' E 1522
54o30,0 ' 03ø48,4' E 2020
53o06,3 ' 19ø24,7' E 2725
52o58,4 ' 23ø46,3' E 2905
51o56,0 ' 42ø53,0' E 2550
44o59,0 ' 53ø17,0' E 3429
54o05,8 ' 00ø20,7'W 2350
55o34,7 ' 00ø43,0'W 1940
57o52,3 ' 10ø40,3'W 3863
58o51,8' 16ø39,0 'W 3640
Core Latitude (S) Longitude Depth, m
KR88-30 61 ø00,2' 93ø11,9' E
KR88-31 59o00,0 ' 89o24,3 ' E
V14-53 56o43,0 ' 24ø31,0' W
V14-58 57o37,0 ' 13ø36,0' W
V 16-60 49o59,6 ' 36o45,5 ' E
V 16-65 45o00,0 ' 45o46,0 ' E
RC8-46 55o20,0 ' 65ø28,0' E
RCll-71 49o08,0 ' 37025,0 ' W
RCll-77 53o03,0 ' 16027,0 ' W
RC11-78 50o52,0 ' 09ø52,0'W
43O0
4595
79O6
3546
4575
1618
2761
5537
4098
3115
Pichon et al.: Surface Water Temperature Changes
291
TABLE l. (continued)
Core Latitude (S) Longitude Depth. m Core
Latitude (S) Longitude
MD82-436 61013,6 ' 19ø28,8'W 3620 RCll-80 46045,0 ' 00003,0 ' W
MD84-529 48054,3 ' 61059,8 'E 2600 RCll-90 56038,0 ' 25043,0 ' E
MD84-530 66006,6 ' 73ø59,1'E 2412 RCl1-118 37048,0 ' 71ø32,0'E
MD84-531 66057,7 ' 75024,6 ' E 365 RC11-119 40ø18,0' 74034,0 ' E
MD84-532 66006,9 ' 76ø45,6'E 2700 RC12-292 39040,6 ' 15ø28,5'W
MD84-533 65ø09,1 ' 78ø21,2'E 3363 RC13-255 50034,0 ' 02ø53,7'E
MD84-540 60044,5 ' 86ø23,3'E 3964 RC13-257 55002,2 ' 03ø00,1'W
MD84-552 54055,4 ' 73ø50,0'E 1780 RC13-263 53048,3 ' 08ø13,0'W
MD84-557 53019,6 ' 75ø48,0'E 1080 RC13-268 57002,3 ' 00ø05,6'W
MD84-561 53005,3 ' 71ø36,4'E 1754 RC13-269 52037,6 ' 00ø07,5'E
MD84-562 51ø55,1' 68013,6 ' E 3553 RC13-270 55028,8 ' 04038,2 ' E
MD84-563 50042,7 ' 68ø09,1'E 1720 RC13-272 55ø05,1 ' 08ø00,0'E
MD84-569 47038,6 ' 73ø23,1'E 2385 RC13-273 55004,5 ' 11ø34,5'E
KR87-01 52032,2 ' 31ø11,0'W 3510 RC13-274 53009,0 ' 12ø25,6'E
KR87-02 56027,6 ' 33ø58,5'W 3350 RC15-91 49055,3 ' 15ø34,1'W
KR87-03 60013,3 ' 48ø56,9'W 1300 RC17-56 65ø24,0 ' 37ø43,0'E
KR87-05 64003,3 ' 67ø17,2 'W 2810 IOl176-55 53022,9 ' 06ø39,6'E
KR87-06 63003,6 ' 63ø03,7'W 630 IO1176-65 57012,5 ' 08ø12,4'E
KR87-07 62021,2 ' 57ø58,3'W 2810 101176-82 49031,2 ' 13ø11,5'E
KR87-08 60055,2 ' 56026,3 ' W 2150 IOl176-86 48002,3 ' 13ø49,0'E
KR87-10 59039,6 ' 51ø16,6'W 2820 IOl176-88 46057,8 ' 14ø18,2'E
KR87-12 54055,8 ' 39ø47,6'W 2790 IOl176-91 44056,7 ' 15ø02,9'E
KR87-13 52042,8 ' 36ø59,3'W 2020 IO1277-2 45ø02,1 ' 22ø28,2'E
KR87-14 51027,7 ' 34ø00,5'W 2600 IO1277-8 50032,5 ' 20ø53,0'E
KR88-01 46040,6 ' 79ø29,2'E 2925 IO1277-12 54000,6 ' 19ø47,5'E
KR88-02 45045,2 ' 82ø56,0'E 3480 IO1277-28 61028,0 ' 09ø11,0' E
KR88-03 46ø04,1' 90006,7 ' E 3400 IO1277-41 69059,9 ' 05004,6 ' W
KR88-04 49055,3 ' 100ø05,0'E 3350 IO1578-49 61005,6 ' 19ø51,9'W
KR88-05 52056,7 ' 109ø55,1'E 3510 101678-64 54000,5 ' 24ø11,7'W
KR88-06 49001,3 ' 128ø46,3'E 3850 101678-80 47057,0 ' 13ø01,4'W
KR88-07 47008,9 ' 145ø47,8'E 2890 IO1678-84 51057,5 ' 14ø25,2'W
KR88-08 49015,7 148ø48,2'E 3885 101678-89 57003,6 ' 18ø32,4'W
KR88-09 50035,6 ' 147ø09,3'E 4350 101678-96 60027,9 ' 21ø37,1'W
KR88-10 54011,2 ' 144ø47,9'E 2785 P1010 77020,0 ' 35ø00,0'W
KR88-11 54055,0 ' 144ø04,0'E 2880 Pl141 61030,0 ' 63ø30,0'W
KR88-12 56023,8 ' 145ø17,2'E 3020 Pl147 61040,0 ' 54045,0 ' W
KR88-13 57056,9 ' 144ø35,0'E 3740 Pl160 61000,0 ' 46ø00,0 ' W
KR88-14 61016,8 144ø26,5'E 4200 Pl178 61025,0 ' 47ø00,0'W
KR88-15 63018,3 ' 141ø55,5'E 3880 Pl184 62010,0 ' 57ø15,0'W
KR88-16 64046,2 ' 141ø13,4'E 3320 Pl192 77025,0 ' 39ø00,0'W
KR88-17 66ø12,1 ' 140ø30,2'E 180 P1212 75030,0 ' 57ø00,0'W
KR88-18 65045,0 ' 138ø12,0'E 615 P1223 76000,0 33030,0 ' W
KR88-19 64034,3 ' 135ø37,5'E 2930 PC82-35 62021,7 ' 57ø22,0'W
KR88-20 64ø56,1 ' 129000,3 ' E 1670 PC82-71 62038,4 ' 59032,0 ' W
KR88-21 64049,3 ' 126ø43,5'E 2250 PC82-136 64035,4 ' 62ø45,5'W
KR88-22 64ø40,1 ' 119ø30,2'E 3140 PC82-140 64049,6 ' 62ø37,8'W
KR88-23 63018,0 117ø15,8 ' E 3292 PC82-197 63043,0 ' 57013,6 ' W
KR88-24 63044,8 ' 116ø44,9'E 2600 DF83III-11C 78015,0 ' 172ø17,0'W
KR88-25 64017,9 ' 115ø42,1'E 2232 DF83 III-12 78016,0 ' 170ø08,0'W
KR88-27 63ø39,1 ' 101ø08,9'E 1240 DF83 III-19 77018,0 ' 158ø43,0'W
KR88-28 64009,5 ' 98ø09,5'E 1200 DF83 III-41 76040,0 ' 164ø01,0'W
KR88-29 62029,5 ' 95ø53,1 ' E 3790 DF83 III-42 76038,0 ' 166ø03,0 ' W
MD: kullenberg from Marion-Dufresne; KR:
box core from Marion-Dufresne; V: Verna;
Depth, m
3656
5334
4354
3709
3541
3332
2837
3389
4005
2591
3160
2538
4967
3372
3775
4794
2926
5483
4100
4338
5106
4649
4806
4492
3178
5322
1873
4718
4515
3102
3952
4285
4177
476
1417
2300
320
500
1882
842
431
772
1484
1350
452
392
750
430
541
677
516
420
RC: Robert Conrad; IO: piston core from Islas Orcadas (cruises 11,12,15, and 16); P: box core
from Polarstern; PC and DF: box core from U.S. Coast Guard Glacier.
292 Pichon et al.' Surface Water Temperature Changes
//// Antarctic Polar Front •
ß Surface samples used in transfer .
function D I$G/:54/4 +
' ß Down.core profiles used in this study
leo e I,,,•0 • l•O ß IrO ß I•0
ti 0 ß
IJO •'
Fig. 1. Geographic distribution of 124 surface sediment samples and three core profiles used in this study.
Location of the Antarctic Polar Front is taken from the reference data map of Gordon and Molinelli [ 1982].
non-reworked Holocene sediments. Forty-two
samples with assemblages altered by step wise opal
dissolution of five Holocene core tops [Pichon et
a1.,1992] were added to the natural floral data for the
Q-mode factor analysis. Before the calculations, the
floral counts were ranked in four abundance classes
defined for each of the 34 species (Table 2,
following Pichon et al., [1987]). This method
increases considerably the sensitivity of the DTF for
changes in SST. However, the separation between
abundance class 1 (absent) and class 2 (present at
low levels) has a heavy statistical weight for certain
low abundance species. This may introduce an error
exceeding the average standard error of the
estimates. [Pichon et al., 1987].
The data, once ranked, were treated by the
programs CABFAC, REGRESS and THREAD
kindly provided by N. Kipp and adapted to the IBM
9121 of the Bordeaux I University Computer Center.
Modern sea surface temperature estimates for
summer, autumn, winter, and spring are 3- month
averages (January-February-March .... ) determined
by interpolating values from reference data maps
[Gordon and Molinelli, 1982].
Results of Factor Analysis
A Q-mode factor analysis of the 166 samples (124
core tops + 42 partly dissolved samples) resolves
four varimax factors (which accounts for 34%,
30.2%, 12.6%, and 6.6% of the total variance,
respectively). Addition of a fifth factor accounts for
<2.4% of the variance and is unrelated to any
discernible biogeographic, physical, or chemical
influence. The varimax factor matrix gives the
composition of each sample in terms of the four
varimax factors (Table 3). We have plotted in
Figures 2 to 5 the distribution of the loadings for
each successive factor over our sample set; floral