EPSL ELSEVIER
Earth and Planetary Science Letters 146 ( 1997) 59 l-605
Modem and last glacial maximum sea surface S “0 derived from an Atmospheric General Circulation Model A. Juillet-Leclerc
a3*, J. Jouzel b, L. Labeyrie a, S. Joussaume b,c
a Centre de.7 Faibles Radioactiuitk, Laboratoire Mixte CNRS-CEA, Avenue de lu Terrusse, 91198 Gif sur Yvette cedex. France b Laboratoire de Mode’lisation du Chat et de I’Ennrironnement, CEA/ DSM CE Saclay. 91191 Gifsur Yr)ette cedex. France ’ Lnboratoire d’Oc&mographie Dynamiyue et de Climatologie. CNRS/ ORSTOM/ (/nkersit& Pierre et Marie Curie. 4 place Jussieu. Paris. France
Received 19 January 1996; accepted 18 October 1996
Abstract The past isotopic contents of sea surface waters (6,) cannot be measured directly, although they would be a good indicator of ocean circulation through their relationship with sea surface salinities. A method of 6, reconstruction is proposed based on the use of outputs from an Atmospheric General Circulation Model including a full isotopic model. Using the outputs of the NASA/GISS isotopic GCM and a simple box model, we have established that there is a strong correlation (r’ = 0.75) between atmospheric fluxes and S, measured in the frame of GEOSECS program, indicating that 6, is largely governed by atmospheric fluxes. This justifies the use of the present-day statistical relationship reflecting essentially the strong atmospheric forcing on sea surface water, for different conditions to those prevailing during the Last Glacial Maximum. For this period, over subtropical areas, lower isotopic compositions are obtained in the Atlantic Ocean whereas
higher values are obtained in the Indian and Pacific oceans, thus reducing the isotopic contrast between the Atlantic and Pacific oceans. The Pacific and Indian oceans show a similar isotopic pattern, with the tropical S, values accentuated by a marked decrease in the equatorial zone. However, whereas the predictions from the Atlantic and Indian oceans exhibit good agreement with proxy data derived from foraminifera, important discrepancies exist in the Pacific Ocean. Keywords: ocean circulation:
models: O-18/0-16;
last glacial maximum
1. Introduction Evaporation (E) and precipitation (P> significantly affect the present distribution of salinity in surface ocean waters and thus the general circulation of the oceans. As stressed by Peixoto [l], the distribution of
’ Corresponding
author.
E-mail:
gif.fr 0012-82 I X/97/$17.00 Copyright PII SO01 2-82 1X(96300237-3
[email protected]
sea surface salinity (SSS) is highly correlated with E-P: salinity is low over the equator, where convergence of water vapour prevails, whereas it is high in the subtropics, where divergence dominates. The excess salt resulting from evaporation in the subtropical Atlantic Ocean partly drives thermohaline circulation and the imbalance of evaporation and precipitation over the Pacific and Atlantic basins leads to an interoceanic salinity difference which maintains the global ocean circulation [2.3].
0 1997 Elsevier Science B.V. All rights reserved.
592
A. Juillet-Leclerc
et al. /Earth
and Planetary Science Letters 146 (1997) 591-605
In a somewhat parallel manner to salt, there is a link between the 6 ‘*O of sea surface waters and E-P, which, in turn, results in a relationship between the SSS and al80 sw, the oxygen-18 content of surface ocean waters ‘. This relationship, well documented and analysed for the modem ocean by Craig and Gordon [4], has been further examined by Broecker [5] using the more detailed GEOSECS dataset [6]. Craig and Gordon [4] noted that one of the difficulties of directly relating ai80s,,, with atmospheric parameters E, P, S E and 6 P (respectively evaporation, precipitation, isotopic composition of evaporation and isotopic composition of precipitation) is due to the fact that 6E cannot be estimated even if the isotopic data of the vapour over the sea were known. The situation has changed largely because of the incorporation of the water isotope atmospheric cycles in General Circulation Models (GCM) of the atmosphere. As those models account for the isotopic fractionation processes that occur at the air-water interface and for the complexity of the water cycle itself, they constitute a way of simulating the worldwide distribution of oxygen-18 in evaporated water and in precipitation. This approach, pioneered by Joussaume et al. [7,8], has now been followed using two other GCM [9,10]. After a description of these isotopic GCM, we use the outputs of one of them, the NASA/GISS model [9], to derive a relationship between 6 ‘*0 sw and E, P, 6 E and 6P for the present-day climate. One interest of this GCM approach is that it may, in principle, be applied to any climate, if the boundary conditions are sufficiently well known. As stressed by Broecker and Denton [ll], interaction between atmospheric fluxes and the ocean circulation may have been different in the past. This is the case for the Last Glacial Maximum (LGM; 18 kyr
’ The same reasoning applies to the deuterium ratios but we will discuss here only oxygen-18 ratios because only oxygen-18 measurements are of interest for paleoceanographic studies. The ‘so/ I60 ratios of waters are commonly expressed in 6 notation with respect to V-SMOW (Vienna Standard Mean Ocean Water), with isotopic ratio R,,,, of ‘*O,J”~O = 2005.2. lOme’. Foraminifeml isotopic composition is reported as a”0 vs PDB. 6 values are given by: S = (Rsample /RStandard)- 1, expressed in permil.
B.P. in traditional 14Cage scale), a period for which we have, for example, strong indications from distributions in foraminifera of either S 13C [12] or cadmium/calcium [13], that the circulation of North Atlantic Deep Water was weaker than now. NASA/GISS isotopic GCM outputs are also available for this period [ 141, and we focus on these in the second part of this article. Assuming that the calculated present-day relationship linking S 180sw and E, P, 6E and 6 P is relevant to the past, we deduce al80 sw for the LGM using simulated past atmospheric data.
2. The evaporation / precipitation isotopic model Incorporating the oxygen-l 8 cycle into an atmospheric GCM consists of following the behaviour of the isotopic species Hf 0 at every stage of the atmospheric water cycle. Since the saturation vapour pressure and molecular diffusivity in air of this molecule are different from those of H h”0, fractionation between these two molecules occurs at every change of phase: surface evaporation, atmospheric condensation and re-evaporation of precipitation. Other parts of the model, namely processes conceming water vapour transport and ground hydrology, are not affected by physical properties but still must be treated completely. All these various parts of the isotopic model are fully described in Jouzel et al. [9], whereas the GISS model II climate performances are extensively discussed in Hansen et al. [15,16]. These studies verified, in particular, that the major features of global climate can be realistically simulated with the coarse grid resolution model (8” by 10” in the horizontal) that we have run to get the isotopic evaporation and precipitation outputs used in the present article. Because of its direct relevance to the present study, we shortly describe here how surface evaporation over the ocean is treated [9]. The flux of water from the ocean is computed using a drag law coefficient [ 161. The calculated flux of water is: Fg = C4Vs(q, - q,>, where: C, is the humidity transfer coefficient; V, is the surface wind speed; qs is the saturation specific humidity at the ocean temperature; q, is the specific humidity at the top of the surface layer, defined by assuming that this surface layer is in equilibrium; that is, the flux of
A. Juillet-Leclerc et al. /Earth and Planetary Science Letters 146 (1997) 591-605
water from the ocean is equal to the flux of water from the surface layer into the upper air by diffusion. For calculating the fluxes of isotopic species, we followed the formulation adopted by Merlivat and Jouzel [17], which takes into account the kinetic effect which results from the differences between molecular diffusivities. The oxygen water flux is calculated as: Fgi = C,iV,(q,i - qsi), where the subscript i is used to denote the oxygen-18 water isotope. Csi is related to C, through Cqi = C,(l - k), where k is derived from experiments conducted in an air-sea simulating facility. Using a closure equation similar to the one used by Craig and Gordon [4]; that is, assuming a balance between the global mean of the isotopic content of 6,, the evaporation, 6,) and of the precipitation, Merlivat and Jouzel [ 171 were able to deduce local values of 6,. Such a closure equation is no longer needed in the GCM approach where qsi is calculated in the same way as q, (i.e., by equating the isotopic fluxes from the ground to those into the first layer). The isotopic properties of this first layer, which must be known for this calculation, are determined at each step of the simulation in adding (or subtracting) both the vapour fluxes coming from the ocean (Fsi) and from the adjacent horizontal and vertical grid boxes and those due to removal and formation of precipitation, if necessary. Finally, qgi is calculated using Henry’s law. Jouzel et al. [9] generated a full annual cycle of isotope fields; like the LMD [7,8] and the Hamburg model [lo], the NASA/GISS model successfully reproduces the important features of present-day global isotope distribution, including the spatial correlations between isotope and temperature at mid to high latitudes and between isotope concentration and precipitation at low latitudes. NASA/GISS results for the annual average 6 “0 in evaporation and in precipitation for a 3 year experiment are given in Jouzel et al. [9], along with the observed distribution of 6 I80 in precipitation established from IAEA and other data. We limit the presentation of these results to some aspects specific to ocean areas: (1) A data/model comparison based on oxygen- 18 data measured in precipitation collected on islands shows a generally satisfying agreement between data and model over ocean areas [9]. Note that, in the simulation, the S “0 of surface water is taken as
593
equal to C%Oall over the ocean, which introduces a slight bias in this comparison. A first-order correction could be done using the GEOSECS data and assuming that precipitation is of local or at least proximal origin, but this would have only a marginal influence on the quality of the comparison. (2) Other characteristics of isotopic data in ocean areas are well simulated. First, the lack of seasonal cycle in the precipitation S “0 in tropical and equatorial sites is correctly reproduced. Second, in the North Pacific, winter precipitation has a slightly higher 6 ‘*0 than summer precipitation with no concurrent inversion of the temperature cycle. This unexpected feature, more probably linked with precipitation patterns, is also seen in the data. Third, the oxygen-18 content of water vapour collected above the sea [4] compares well with the value predicted in the model first-layer vapour, both ranging between - 10 and - 15%~. (3) Broecker et al. [ 181 have examined how freshwater fluxes (evaporation, E, precipitation, P, and run-off), as simulated in this version of the GISS model, compare with the conventional climatology approach [19], which involves measurements of runoff and precipitation and calculations of evaporation from humidity, temperature and wind speed data. The authors noted that, given these approaches are based on totally different strategies, the two estimates statistically agree (within about &25%); in particular the atmospheric fluxes both across boundaries in the Atlantic (40”N and 40”s) and from the Atlantic to the Pacific and Indian oceans are correctly estimated in the atmospheric GCM. One anomaly has to do with run-off from the continents [18] but it is not of importance here as we will avoid discussion of the salinity-oxygen18 relationship in coastal grid boxes. Zaucker and Broecker [20] further examined the atmospheric water vapour fluxes produced by the GISS model through a comparison with the Oort [21] climatological dataset and they noted substantial differences, such as a much too low model export from the Atlantic basin. This is also observed with a version of the model different to that used here (resolution of 4 X 5” instead of 8 X 10”) and such model deficiencies have to be kept in mind in our present approach. The LGM simulation of the GISS isotopic model parallels the present-day one, differing only in the
594
A. Juillet-Leclerc
et al/Earth
and Planetal?, Science Letters 146 (1997) 591-605
imposed set of boundary conditions. Sea surface temperatures, sea-ice limits, land ice configuration and land albedo are changed to their glacial values according to CLIMAP data [22]. Orbital parameters and CO, concentrations are set to their glacial values and we have accounted for the average changes in ocean 6 180 due to the build-up of continental ice sheets (see Section 4). Except that, due to this 6 ‘*0 change, the distribution of 6 “0 in precipitation over ocean areas shows little modification with respect to that simulated for the current climate [14]. For most of the area between 50”N and 50”s this modification is within + 1%0. Despite this constancy of the precipitation isotopic content, the net isotopic flux (E6, - PS,) shows noticeable differences, largely due to changes in water fluxes. We also use below the outputs of a second LGM simulation in which ocean temperatures have been decreased by 2°C [14].
3. Relationship between the model outputs and the measured modern sea surface S’*O distribution 3.1. Measured sea su$ace al80 The measured 6 ‘*O values from the Atlantic, Pacific and Indian oceans are published in the GEOSECS Atlas [6]. Due to the sparsity of data in the Indian Ocean, measurements performed in our laboratory were added to the GEOSECS database
Fig. 2. Statistical relationship between atmospheric outputs of model and measured sea surface isotopic composition for the present. The general relationship calculated for all oceans is compared with specific equations derived for each ocean basin: all relationships are close, except a slight difference for the Indian Ocean.
after careful checking of the consistency between the two datasets. As 6 I80 measurements are not available for all model grid boxes, we calculated the average of 6 ‘*0 values for each model latitudinal belt (Appendix A). We note that similar values are obtained for a given latitude in the Atlantic and Pacific sides of the Southern Ocean and also around the equator for the Indian and Pacific oceans. This is expected because in both cases the latitudinal belts correspond to areas where oceans are connected but the above mentioned agreement can be considered as an indication of the general validity of the procedure followed to transfer GEOSECS data on the NASA/GISS grid. 3.2. Comparison between sured sea su$ace al80
Fig. 1. Schematic model describing water exchange between atmosphere and ocean. E, P, 6, and 6, are evaporation, precipitation, isotopic composition of evaporated water, and isotopic composition of precipitation, respectively. K, is the coefficient of diffusion between the upper and the deeper layer, h the depth of the surficial box, 6, the isotopic content of the deep ocean.
model outputs and mea-
As atmospheric data will be compared with S ‘*O sea water averaged on a latitudinal belt, we calculated the average of E, P, 8, and 6, for the same latitudinal belt. To avoid problems due to continent proximity and deep water formation in high latitudes, we restricted reconstruction to grid boxes containing more than 90% ocean, between 55”N and 55”s in all oceans. The isotopic composition of surface ocean waters, a,, can be estimated as a function of atmospheric fluxes and of the isotopic content of the deep ocean,
595
A. Juillet-LRclerc et al. /Earth and Planetary Science Letters 146 (1997) 591-605
a,, using a very schematic model (Fig. 1). Only diffusive vertical transports (characterized by diffusion coefficient) are considered between surface and deep waters and we neglect vertical water transports such as lateral exchanges. We express the steady state as: P6, - ES, = pK,/h
. (6, - 6,)
that is: 6, = h/pK,
. (PS, - ES,)
+ 6,
(1)
In this equation, K, is the coefficient of diffusion between the upper and the deeper layer, h the depth of the surficial box and p the sea water density. In order to establish a relationship between atmosphere and ocean in terms of oxygen isotopic ratio, an empirical relationship is obtained between measured S, and simulated values of ES, - PS, (Fig. 2) (the corresponding table is given in Appendix A). 6, = 0.22 - 6.55.
10-5(E6,
- Pa,)
(2)
with r’ = 0.75 (r being the correlation coefficient). The high value of r emphasises the strong atmospheric forcing on sea surface water. Interestingly, the value of 0.22%0 (deep water box) compares well with averaged GEOSECS data from both the Atlantic, Pacific and Indian oceans, assuming a 100 m depth surficial box. With this value for h, we derive an estimate of 4.8 X 10V3 m*/s for K,, whereas actual K, values are much lower. Note that, rigorously, Eq. (1) considers only transports related to vertical diffusion whereas advection terms are naturally included in the measured 6 I80. These specific circulation patterns are not strictly similar in each ocean basin and it is possible to take into account of the differences existing between the three oceans by comparing the statistical relationship calculated individually, which includes geographical features (Fig. 2). In the Atlantic
Ocean:
6, = 0.36 - 6.44.
lo-‘(ES,
- Pa,)
where
r2 = 0.82
(3)
In the Pacific Ocean: 6, = 0.15 - 6.04. r2 = 0.68
10e5(E6,
- Pa,)
where (4)
In the Indian Ocean: 6, = 0.20 - 4.29.
10-5(ES,
- PS,) where
r2 = 0.72
(5)
However, whatever the relationship used, the strong atmospheric forcing is a justification for the use of the GCM outputs to estimate ocean surface 6 “0 during the LGM. We propose to reconstitute a very general equation and to compare with other values, taking into account the properties of each basin.
4. Application to the Last Glacial Maximum
4.1. %U.X, reconstruction Ocean surface 6 “0 values were different during the Last Glacial Maximum for various reasons. First, 6, changed with the storage of large amounts of continental ice of low S “0. Second, local sea surface isotopic variations were probably affected by changes in atmospheric water fluxes to and from the ocean. The first contribution is quite precisely known: several authors [23-251 estimated at 1.2%0 the 6, effect due to ice storage during the Last Glacial Maximum. To estimate the second contribution, we assume that the statistical relationship established for modem conditions traducing essentially atmospheric forcing on sea surface 6”O is also valid for the LGM, taking into account the global change of 1.2%0 for 6,. We assume that the box ocean model with features similar to those previously defined, in particular the calculated modem diffusive coefficient K ,, , may be applied to the LGM. The following general equation thus applies for glacial conditions: S S(LGM)
=
-6.55
X 10-5(ES,
- PS,) + 1.42
(6)
We consider only grids completely free of ice (as defined in the GCM) and which lie between 50”N and 50”s. Values are plotted on Fig. 3 and given in Appendix B. While the Ss(LGMj range is comparable for all ocean basins south of 2O”S, the Atlantic Ocean does not present the minimum SS(LGM)around 1.2%0 shown in the Indian and Pacific Oceans between 20”s and 30”N (Fig. 3). Atlantic Ocean SSuGM) remains around 2%0 between 30”s and 30”N.
596
A. Juillet-Leclerc
Indian
Ocean
4 -60
and Planetary Science Letters 146 (1997) 591-605
Ocean Pacific
0.5
et al. /Earth
I -40 s
-20
0 Lalitude
20
40
60
N
Fig. 3. Reconstruction of the sea surface 6’*0 values during the Last Glacial Maximum: 6,(,,,, for latitudes between 50”N and 50”s in the Atlantic, Pacific and Indian oceans. The 6,(,,,, range is comparable for all oceanic basins south of 20’S, but the Atlantic Ocean does not present the minimum 8S(LGM) around 1.2%0 shown in the Indian and Pacific oceans between 20”s and 30”N.
To add precision of the changes between the LGM and modem ocean for each ocean region, we compare in Fig. 4 the 8S(LoMJ and the modem data from the GEOSECS Atlas corrected by + 1.2%0 to compensate for the build-up of global ice sheets. Fig. 4 shows that the observed present-day difference between the Atlantic and Pacific oceans was reduced during the LGM, especially between 20”N and 40”N where similar isotopic values are observed. Besides latitudinal 6, distribution being shifted towards the south, the main result for the Atlantic Ocean is an apparently lower aScLGMj excess in the tropical latitudes for the LGM for both hemispheres. The reverse is true both for the Indian and Pacific oceans. There, tropical waters had higher i&oMj (by about 0.25%0) in the tropical areas, and lower &s(LGMj(by about the same amount) along the equator. By comparing E/P distribution for the present and LGM, we conclude that the surface isotopic composition pattern obtained for the past is not only due to an excess of evaporation over precipitation. Based on specific relationships for each basin reconstitution show only reduced (Fig. 2) $_oMj modifications: values are higher by 0.15%0 in the Atlantic Ocean and overestimated by less that 0.1%0 in the Pacific Ocean; the only noticeable discrepancy
is showed in the Indian Ocean where QGMj are decreased by 0.2%0 over south tropical and subtropical latitudes. The simulation in which SSTs were decreased by 2°C [14] shows no change in the isotopic content of the ocean precipitation. This results in only small changes (less than 0.15%0) in the calculated 6,(,,,, values (Fig. 5, Appendix Cl. Consequently, the isotopic composition contrast between the Atlantic and Pacific oceans is still reduced in northern mid latitudes and the main features of each ocean basin are unaffected. This additional experiment thus suggests that the main conclusions derived above are not very sensitive to temperature changes and thus to inaccuracies in the estimate of LGM sea surface temperatures.
4
0.50
-60
2.50
0
I -40
-20
T
-60
2 50
0.50
40
20
40
60
20
40
60 N
60
Ocean
k
-40
1
-20
0
Indian
,
Ocean
LGM
J -60
*
20
J
50
5
Pacific
0
I -40
-20
dt”d*
and the modem data from the Fig. 4. Comparison of 6,(,,,, GEOSECS Atlas corrected by + 1.2%0 to compensate for the build-up of global ice sheets. The present-day difference between the Atlantic and Pacific oceans was reduced during the LGM, especially between 20% and 40”N.
A. Juillet-Leclerc et &/Earth
591
and Planetary Science Letters 146 (1997) 591-605
2.50
0.50
!
I -40
-60
-20
0
Pacific
20
40
60
Ocean
_._”
-60
0.5
-40
-20
0
20
40
60
-40
-20
0
20
40
60
4
I
-60 S
Latitude
N
calculated from outputs of a model using CLIMAP temperatures as boundary conditions and BscLoMJ Fig. 5. Comparison of S,,,, derived from a model using CLIMAP temperatures decreased by 2°C. The simulation in which SSTs were decreased by 2°C shows no change in the isotopic content of the oceanic precipitation, this results in only small changes for SS(LGMj.
4.2. Comparisons between the GCM outputs for LGM and available foraminiferal isotopic and micropaleontological data Sea surface 6 t80s may be derived from the oxygen isotopic ratio of fossil planktic foraminifera ( 6 I8Ocalcite) using the Shackleton [26] paleotemperature equation, if the changes in isotopic fractionation with temperature of growth CT,> are independently assessed: T, = 16.9 - 4.38(6’80,,,,it,
- 6’*0,)
+ O.lO( 8i80calcite - 6i80s)2
(7)
T, was estimated by CLIMAP [22] using transfer functions based upon the fossil foraminifera distribution. This method was proposed by Duplessy et al. [27] to reconstruct the distribution of salinities in the North Atlantic Ocean during LGM. 4.2.1. Comparison with the Duplessy et al. reconstruction Until 45”N there is good agreement between calfor the culated 8suGMj with specific relationship Atlantic Ocean and Duplessy et al.‘s [27] reconstruction (Fig. 6). Our estimates, deduced from the specific Atlantic Ocean relationship, show a progressive increase in SS(rGMj with decreasing latitude, while
A. Juiller-Leclerc
598
-80
-70
60
et al./ Earth and Planetary Science Letters 146 (1997) 591-605
40
-60
QO
-20
-10
0
10
20
65
65
60
60 4
55
50
P
55 1.20 . 50 1.3;
,
45
45
It* wticr
40
n2.05
40
35
30
I
I
I
I
I
I
I
I
I
-90
%O
-70
60
d0
40
-30
-20
-10
0
10
25 20
-90
40
-70
-60
-50
40
-30
-20
-10
0
10
20
25
65
65
60
60
55
55
50
50
45
45
40
40
35
35
30
30
25 -90
-80
-70
-60
-60
40
-30
-20
-10
0
10
25 20
Fig. 6. Map of c&,,,, (a) from Duplessy et al.‘s reconstruction [27] and (b) deduced from the specific Atlantic Ocean relationship. Until1 with decreasing latitude, while the 45% there is a good agreement but southwards our estimates show a progressive increase in 6soo,, estimates of Duplessy et al. exhibit a rapid increase.
A. Juillet-Leclerc Table 1 Comparison
of sea surface temperatures
Core
Latitude
Longitude
et al./Earth
and Planetary Science Letters 146 (1997) S-605
599
derived from this study and CLIMAP temperature Depth (ml
CLIMAP temp. (“C)
Foram. spec.
6’80 foram vs. PDB (%‘uu)
IS’s0 warrr vs. SMOW(%,)
1.18 1.24 1.04 I .04 1.24 1.24 1.83 2.29 2.29 2.11 1.78 1.84 2.03 2.06 2.06 2.17
4 5 4 4 5 7 26 28 26 26 26 24 26 26 24 24
6 6 7 6 6 9 27 27 26 26 27 25 27 25 25 25
Atlantic Ocean V27-116 52”5N v23-82 52”35N v27-19 52”06N 50”05N v27-I7 49”52N V23-83 45”18N V29-180 24”48N A179-15 V25-60 03”17N Ol”22N v25-59 V25- 182 OO”33S 03”33S v25-56 v22-177 07”45S v22-38 09”33S v22- 174 I O”O4S 1O”O4S VX- I74 24”048 RC8-18
30”2OW 21”56W 38”48W 37”18W 24” 15W 23”52W 75”56W 34”5OW 33”29W 17”16W 35”14W 14”37W 34”15w 12”49W 12”49W 15”07W
3202 3974 3466 4054 3871 3179 3109 3749 3824 3776 3512 3290 3797 2630 2630 3977
bull. bull. bull. bull. bull. bull. rub. sac. sac. sac. sac. sac. rub. sac. sac.
4.52 4.34 4.46 4.42 4.3 1 3.80 - 0.02 0.19 0.41 0.43 0.05 0.50 0.13 0.30 0.67 0.62
Pacific Ocean Kg-294 X8-239 RCll-210 V28-238 V28-203 RC13-I 13 RC 1 l-230 RClO-114 v9-55 RC13-81 RC8-94 RC9- 124
28”43N 03”25N Ol”82N 01”OlN OO”95N Ol”65S OS”8OS 1 l”18S l7”oos 19”OlS 27”283 28”743
139”58E 159”llE 1 lO”O3W 160”29E 179”25W lo3°38w I lO”48W 162”55W 114”l IW 124”13W 102”05W 172”35E
2308 3490 4420 3120 3243 2195 3259 279 I 3177 3751 3074 2540
sac. sac. sac. sac. sac. sac. sac. sac. sac. sac. sac. sac.
-0.50 -0.84 0.16 - 0.95 - 0.60 -0.10 -0.13 - 1.01 -0.12 - 0.22 0.8 1 0.44
2.77 1.34 0.9 1 1.34 1.09 1.63 2.04 I.13 1.65 1.39 1.89 2.42
30 26 19 26 24 24 26 26 24 23 21 25
27 29 27 29 28 26 26 30 38 28 24 22
Indian Ocean MD77-203 MD77- 179 v34-88 RC 12-343 RC 12-34 1 MD77-171 MD77-194 V14-102 A 15558 V14-10lB MD77-191 RC 14-37 RCl l-147
20”70N 18”35N l6”52N 15”16N 13”05N 1 l”75N I O”47N 1O”25N 08”98N 08”65N 07”50N 0 l”46N 19”05S
059”34E 091”OlE 059”32E 090”34E 089”35E 094”09E 075”14E 057”llE 05 l”44E 058”34E 076”43E 090” 1OE 1 l3”45E
2442 1986 2120 2666 2988 1760 1222 3215 3985 2849 1254 2226 1953
rub. rub. rub. rub. sac. rub. rub. rub. rub. rub. rub. sac. sac.
0.07 -0.99 -0.27 - 1.04 -0.78 - 0.96 - 0.60 -0.67 -0.25 -0.54 -0.28 -0.56 - 0.26
1.66 1.46 I .66 1.64 1.64 1.64 1.21 1.34 1.53 1.47 1.41 1.20 1.97
23 27 25 28 27 28 24 25 24 25 24 24 26
22 27 25 36 26 27 26 26 26 26 23 27 25
SX.
-
The sea surface 6’sO values during the Last Glacial Maximum: 6so,,, are calculated from the following relationships: In the Atlantic Ocean: S, = 1.56 - 6.44. 10mi (E6, - P&l. In the Pacific Ocean: 6, = 1.35 - 6.04. IO-’ (E6, - P&,1. In the Indian Ocean: 6, = 1.40 - 4.29 10e5 (ES, -P&I. E, P, S, and 6, are evaporation, precipitation, isotopic composition of evaporated water. and isotopic composition of precipitation, respectively. Calculated temperature from &t,,,, reconstruction and CLlMAP temperatures show good agreement in the Atlantic and Indian oceans. whatever latitudes, but great discrepancy in the Pacific Ocean,
600
A. Juillet-Leclerc et al. /Earth and Planetary Science Letters 146 (1997) 591-605
Duplessy et al. estimates exhibit a rapid increase (N 0.5%0) south of 43”N. As we discarded cores from the Bay of Biscay, where dilution by fresh water occurred [27], and cores along the west coast of Africa, located in an upwelling [28], this discrepancy may not be attributed to coastal influence or vertical transport of water masses. The high SS(roM) values calculated by Duplessy et al. from the isotopic composition of foraminifera and the inferred high QLGM) gradient (more than 0.5%0 for less than 10” latitude) may be explained by an overestimation of the foraminiferal growth temperature in the CLIMAP reconstruction (a difference of 1°C lead to a 0.25%0 isotopic error). On the other hand, our estimate may be biased by the simplification resulting from the previously described statistical model and/or by the fact that the isotopic composition of waters were, in this area, modified by the influence of a surface ocean current bringing waters enriched in “0. 4.2.2. Comparison with other isotopic data Such a direct comparison of this 6suGM) reconstruction with other estimates of 6s during the past is only possible for the North Atlantic Ocean. In order to check our data in other ocean basins, we adopt another approach. Sea surface temperatures may be independently estimated: (1) by introducing 8S(LoM) calculated from the simulation in the equation derived for each ocean basin; and (2) using the foraminiferal transfer function. Such an approach can be used because we have checked that the fluxes derived from the GCM model are not sensitive to the SSTs used as boundary conditions for the simulation (i.e., CLIMAP or CLIMAP -2°C). Table 1 gives the calculated temperatures, T,, and the values inferred by the foraminiferal transfer function [22]. We used 6 l8 Ocaco, from Duplessy et al. [27] and the data base from Prell [29]. We remark that almost all these cores are located at low latitudes. In the Atlantic Ocean, taking into account that SST - 1°C for G. buEloides [27] Tisotope = summer and Tisotope -> summer SST for G. ruber [30], temperature reconstructions are consistent by f l”C, except for two cores. For the assessment based on isotopic values from G. sacculifer, although Tisotope has not been calibrated, SST estimates are also in good agreement.
In the Indian Ocean isotopically calculated temperatures from G. ruber and G. sacculijizr are roughly similar to CLIMAP temperatures (within k l”C), except for some cores such as RC14-37 for which Tisotope is 3°C lower than the CLIMAP estimate [29]. No systematic trend appears between our temperature reconstruction and the CLIMAP one in the northern part of the Indian Ocean. In the Pacific Ocean temperatures calculated from the isotopic composition of G. sacculifer are systematically much lower than CLIMAP values except for subtropical latitudes. This discrepancy could not be attributed to the isotopic pattern obtained by our reconstruction in the Pacific Ocean. High isotopic compositions of surface waters are found in subtropical latitudes while low values marked the equatorial zone; too high 8SuoM) values could induce lower temperatures than CLIMAP values in the subtropics while too low SS(roM) values would give higher temperatures. This could be due to the dissolution affecting foraminifera isotopic composition or to CLIMAP reconstruction giving overestimated temperature. Note, however, that commonly it is admitted that dissolution is reduced for LGM sediments
[311. Summarizing this comparison, calculated temperand CLIMAP atures from SSuGM) reconstruction temperatures show good agreement in the Atlantic and Indian oceans, whatever latitudes, but important discrepancies in the Pacific Ocean, although there is a great coherency in the QLoM) reconstruction between the Indian and Pacific oceans.
5. Conclusion The correlation obtained for the statistical relationship linking the measured isotopic composition of water from GEOSECS and atmospheric fluxes derived from a present-day simulation of the GISS model emphasizes the major influence of these atmospheric fluxes on the isotopic composition of sea surface water. Although this relationship is based on a simple model neglecting meridional currents and vertical transfers, the high degree of correlation (r2 = 0.75) justifies its use for different boundary conditions such as those prevailing during the Last Glacial Maximum. A reduced isotopic contrast between the Atlantic and Pacific oceans over subtropical latitudes
A. Juillet-Leclerc et al./Earth and Planetary Science Letters 146 (19971591-605
as well as an important dilution along equator in the Pacific and Indian oceans are predicted. This weaker global isotopic composition difference between the Atlantic and Pacific oceans for the LGM could be attributed to a poor reconstitution of water fluxes between these two oceans given by the GISS model. 6,(,,,, assessment shows a general good agreement with geochemical proxies in the Atlantic and Indian oceans whereas significant discrepancies are exhibited in the Pacific Ocean. The approach that we have followed is promising but we are well aware of its limitations at the present stage. These stem first from the difficulty of a correct evaluation of the water fluxes at the air-ocean interface [l&20]. Unlike the fluxes themselves, estimates of the isotopic composition of the evaporated and precipitated water are probably sufficiently well predicted and do not introduce additional large uncertainty. A second limitation comes from the fact that we do not account for ocean circulation. Although the very simple approach used here gives satisfying results at the large scale on which we have focused the present study, this would not be the case if regions such as the North Atlantic were considered to establish the statistical relationship.
601
We are now aiming to improve our approach on both aspects, first in using the outputs of different atmospheric GCMs (beyond the three models we have mentioned, water isotopes are now being introduced in a version of the NCAR GCM) and second, in introducing the simulated fluxes in an OGCM (Oceanic General Circulation Model) in which ocean circulation is taken into account.
Acknowledgements Randy Koster, Gary Russell, Bob Suozzo, Reto Ruedy, David Rind and Jim Hansen were very helpful during all phases of the development of the isotopic version of the GISS model. We would like to thank Wally Broecker and Jean-Claude Duplessy for very stimulating discussions on the links existing between sea surface salinity, water isotope cycles and the hydrological cycle. This manuscript was greatly improved by comments from W.L. Prell and an anonymous reviewer. Financial support came from the French Programme National d’Etudes de la Dynamique du Climat. [MKI
602
Appendix
A. Juillet-Leclerc
et al. /Earth
and Planetary Science Letters 146 (1997) 591-605
A
Present atmospheric parameters derived from a GCM and average of GEOSECS for latitudinal belts in each ocean basin Latitude
S180 measurements
E
P
6,
8,
6, - 6,
6%0 vs. SMOW
908.00 1142.50 1974.67 1675.00 1681.25 1733.33 1457.00 1573.75 1825.25 1816.50 1840.80 1675.67 1033.38 528.83
1093.67 1109.00 1396.33 1129.50 1205.75 1345.00 863.67 641.00 974.25 991.25 1067.80 1157.17 1148.38 937.67
- 4.69 -5.51 - 6.78 -6.31 - 6.39 - 6.30 - 7.47 - 8.00 - 7.48 - 7.06 - 7.39 - 7.57 - 6.44 - 3.29
-6.16 - 4.48 - 2.89 - 2.43 - 2.20 - 2.64 -2.91 - 1.50 - 1.20 - 1.53 - 2.01 -3.11 -6.13 - 9.70
248 1.49 1326.94 9348.58 7815.22 8086.33 7363.42 8367.64 - 11626.90 - 12479.21 - 11301.27 - 11464.60 - 9092.66 382.88 7353.26
0.07 0.75 1.14 1.17 1.03 0.93 0.92 0.86 1.06 1.14 0.91 0.79 0.14 -0.17
Pacific Ocean 50.9 N 556.57 43 774.89 35.2 1498.78 27.4 1879.27 19.6 1720.73 11.7 1596.38 3.9 1565.21 3.9 1581.27 11.7 1838.83 19.6 1721.36 27.4 1679.3 1 35.2 1421.45 43 996.64 50.9 s 785.23
1034.00 1049.33 1218.33 1374.55 1438.82 1633.38 2142.43 1237.09 1944.33 1263.73 1071.69 1183.55 1214.00 1009.08
- 2.27 -5.18 - 6.08 - 6.27 -5.56 -4.41 - 2.56 - 5.08 -4.89 -5.80 - 6.83 - 6.45 - 4.95 -4.41
- 9.49 -7.14 - 3.97 -3.21 -3.10 - 3.35 -4.17 - 2.65 - 3.00 - 2.52 - 2.50 - 3.79 - 6.06 - 7.46
8546.40 3482.98 - 4272.45 - 7359.7 1 - 5093.85 - 1564.5 1 4927.57 -4761.14 - 3163.84 - 6799.42 - 8792.45 - 4683.39 2425.38 4065.10
- 0.67 - 0.39 0.14 0.51 0.29 0.42 0.31 0.52 0.49 0.64 0.75 0.38 0.14 0.02
Indian Ocean 11.7 N 2262.50 3.9 1416.00 3.9 1403.00 11.7 1736.00 19.6 1697.20 27.4 1716.20 35.2 1429.13 43 758.25 50.9 s 497.45
2973.00 1128.67 1394.60 1636.20 1087.60 1021.80 1095.25 959.33 83 1.OO
- 2.06 - 4.73 -4.16 -5.15 - 6.79 - 7.50 - 6.87 - 5.28 -3.14
- 3.78 -4.16 -4.11 - 3.09 - 2.20 - 2.20 - 3.65 - 5.86 - 8.04
Atlantic 50.9 N 43 35.2 27.4 19.6 11.7 3.9 3.9 11.7 19.6 27.4 35.2 43 50.9 s
Ocean -
6562.32 - 2007.15 - 113.08 - 3894.56 - 9129.09 - 10618.06 -5823.16 1615.44 5119.99
0.50 0.41 0.49 0.30 0.55 0.57 0.49 0.08 -0.21
given
A. Juillet-Leclerc
Appendix
et al. / Earth and Planetary Science Letters 146 (1997) 591-605
603
B
LGM atmospheric parameters derived from a GCM and estimates latitudinal belts in each ocean basin
of aso,,,
calculated
from Eq. (3) for
E
P
6,
6,
6, - 6,
6%~ vs SMOW
1067.50 1759.50 2065 .OO 2252.00 2185.00 1872.33 1445.00 1730.50 1743.00 1567.80 1502.17 807.13 467.00
820.00 1281.50 1167.20 1465.50 1262.67 1268.33 460.25 913.25 1026.50 927.20 1102.83 947.00 837.00
-5.10 - 4.86 - 4.75 -4.82 - 5.65 - 5.75 - 5.20 - 4.92 -4.58 - 4.82 - 5.39 -4.26 -0.11
- 3.04 - 1.49 - 0.03 - 0.23 -0.19 -0.15 0.31 0.28 0.05 - 0.59 - 2.26 -5.71 -8.31
- 295 1.45 - 6654.01 - 977 1.94 - 10511.95 - 12105.34 - 10575.67 -7651.91 - 8774.10 - 803 1.70 - 7010.32 - 5607.95 1973.23 6904.10
1.61 1.86 2.06 2.11 2.21 2.11 1.92 1.99 1.95 1.88 1.79 1.29 0.97
Pacific Ocean 432.80 50.9 N 830.00 43 1925.22 35.2 2372.00 27.4 2118.33 19.6 1844.14 11.7 1487.73 3.9 1390.83 3.9 1605.75 11.7 1919.00 19.6 1973.08 27.4 1729.00 35.2 1253.45 43 800.00 50.9 s
1021.20 1186.00 1408.56 1504.09 1845.75 2184.57 2172.33 799.00 1739.92 1724.38 1258.15 1334.09 1291.73 1103.75
1.25 - 2.74 - 4.74 - 4.29 - 3.45 - 1.30 -0.41 - 3.61 -3.11 - 3.64 - 4.92 - 5.00 - 4.63 - 3.08
- 7.90 - 4.86 - 1.80 - 1.02 - 1.55 -2.15 - 2.63 - 0.66 - 1.59 - 1.23 - 0.59 - 1.64 - 3.85 - 6.85
8606.13 3488.57 - 6589.16 - 8628.98 - 4457.47 2309.39 5094.50 - 4494.55 - 2225.97 - 4865.64 - 8963.16 - 6457.09 - 839.60 5093.10
0.86 1.19 1.85 1.99 1.71 1.27 1.09 1.71 1.57 1.74 2.01 1.84 1.47 1.09
Indian 11.7N 3.9 3.9 11.7 19.6 27.4 35.2 43 s
2000.00 1135.50 1550.20 2 124.20 1355.60 973.50 964.13 907.64
- 1.30 - 2.29 - 1.49 -3.11 - 4.77 - 5.54 - 5.09 -4.36
- 1.10 -2.71 - 3.05 - 1.15 -0.10 - 0.26 - 2.03 - 4.80
- 553.40 194.07 2813.51 - 3769.34 - 9525.90 - 9066.63 - 4226.11 99 1.44
1.46 1.41 1.24 1.67 2.04 2.01 1.70 I .36
Latitude Atlantic 43 N 35.2 27.4 19.6 II.7 3.9 3.9 11.7 19.6 27.4 35.2 43 50.9 s
Ocean
Ocean 2118.00 1260.25 1287.40 1997.40 2026.60 1680.67 1215.50 772.45
604
A. Juillet-Leclerc et al. /Earth and Planetary Science Letters 146 (1997) 591-605
Appendix C LGM atmospheric parameters derived from a GCM using CLIMAP temperatures as boundary conditions and estimates of Ss(LGMjcalculated from Eq. (3) for latitudinal belts in each ocean basin P
6,
6,
6, - 8,
6%0 vs SMOW
Atlantic Ocean 43 N 939.00 35.2 1543.17 27.4 1859.40 19.6 1997.25 11.7 1850.33 3.9 1660.67 3.9 1365.75 11.7 1562.00 19.6 1488.50 27.4 1403.20 35.2 1401.50 43 s 717.38
725.50 1249.00 1119.80 1329.75 1143.67 1061.33 459.50 826.50 926.00 896.00 1028.17 841.13
- 5.33 - 4.67 - 4.88 - 5.05 - 5.67 - 6.32 - 5.70 -5.25 - 4.73 - 4.79 - 5.82 - 4.44
- 3.34 - 1.60 0.13 0.03 - 0.03 -0.10 0.76 0.56 0.11 - 0.54 - 2.39 - 6.00
- 2585.33 -5205.13 -9231.36 - 10129.33 - 10447.10 - 10383.74 - 8137.41 - 8667.25 - 7 136.43 - 6231.88 - 5707.5 1 1863.24
1.59 1.76 2.02 2.08 2.10 2.10 1.95 1.99 1.89 1.83 1.79 1.30
Pacific Ocean 50.9 N 412.00 43 758.29 35.2 1763.67 27.4 2149.64 19.6 1914.92 11.7 1725.86 3.9 1386.00 3.9 1291.17 11.7 1489.50 19.6 1726.00 27.4 1839.54 35.2 1554.45 43 1147.73 50.9 s 701.75
812.67 1021.00 1326.33 1444.82 1657.75 1990.64 2023.87 642.08 1627.92 1497.69 1187.15 1283.91 1230.64 967.50
0.78 - 3.47 - 5.09 - 4.60 - 3.75 - 1.72 - 0.57 - 4.36 - 3.20 -4.15 -5.14 -5.10 - 4.78 - 3.61
- 7.87 - 5.32 - 2.25 - 1.10 - 1.90 - 2.47 - 3.02 - 0.62 - 1.84 - 1.40 -0.61 - 2.03 - 4.59 - 7.62
6712.96 2804.84 - 5990.87 - 8296.40 -4032.81 1941.87 5331.58 - 5234.10 - 1769.68 - 5061 .OO - 8730.97 - 5310.82 155.85 4846.57
0.98 1.24 1.81 1.96 1.68 1.29 1.07 1.76 1.54 1.75 1.99 1.77 1.41 1.10
Indian Ocean 11.7 N 1980.00 3.9 1199.50 3.9 1202.40 11.7 1838.20 19.6 1898.80 27.4 1620.33 35.2 1123.63 43 s 697.83
2103.33 1004.00 1367.40 1867.00 1282.00 1012.67 926.00 818.42
- 1.58 - 2.85 - 2.30 - 3.47 - 4.97 -5.58 - 5.36 - 4.55
- 1.69 - 2.88 - 2.99 - 1.36 - 0.39 - 0.63 - 2.66 -5.37
432.83 - 526.57 1325.74 - 3828.35 - 8929.46 - 8410.57 - 3558.73 1218.21
1.39 1.45 1.33 1.67 2.00 1.97 1.65 1.34
Latitude
E
A. Juillet-Leclerc et al. / Earth and Planetary Science Letters 146 (1997) 591-605
References [II J.P. Peixoto,
The atmospheric branch of the hydrological cycle and climate, in: Variations in the Global Water Budget. A. Street-Perrott et al.. eds., pp. 5-65, Reidel, Dordrecht. 1983. Dl G. Walin, The thermohaline circulation and the control of ice age, Paleogeogr. Paleoclimatol. Paleoecol. 50, 323-332, 1985. water S”O and [31 GE. Birchfield, Changes in deep-ocean temperature from the Last Glacial Maximum to the present. Paleoceanography 2, 431-442, 1987. [41 H. Craig and A. Gordon, Deuterium and oxygen 18 variations in the ocean and the marine atmosphere, in: Stable Isotopes in Oceanic Studies and Paleotemperatures, E. Tongiorgi, ed., pp. 9-130, Spoleto, 1965. 151W.S. Broecker, Oxygen isotope constraints on surface ocean temperatures, Quat. Res. 26. 121-134. 1986. [61H.G. Gstlund, GEOSECS Atlantic, Pacific and Indian Ocean expeditions, Shorebased data and graphics 7, IDOE Natl. Sci. Found., 1987. [71 S. Joussaume and J. Jouzel. Paleoclimatic tracers: an investigation using an Atmospheric General Circulation Model under Ice Age conditions 2. Water Isotopes. J. Geophys. Res. 98, 2807-2830, 1993. [81 S. Joussaume, R. Sadoumy, and J. Jouzel, A general circulation model of water isotope cycles in the atmosphere, Nature 31 I. 24-29, 1984. [91 J. Jouzel, G.L. Russell, R.J. Suozzo, R.D. Koster, J.W.C. White and W.S. Broecker, Simulations of the HDO and Hi80 atmospheric cycles using the NASA GISS General Circulation Model: the seasonal cycle for present-day conditions. J. Geophys. Res. 92, 14,739-14,760, 1987. [lOI G. Hoffmann and M. Heimann, Water tracers in the ECHAM general ciculation model, in: Isotope techniques in the study of past and current environmental changes in the hydrosphere and the atmosphere, Int. Atomic Energy Agency, 1993. [Ill W.S. Broecker and G.H. Denton, The role of ocean-atmosphere reorganizations in glacial cycles, Quat. Sci. Rev. 9. 305-341, 1990. R.G. Fairbanks, L.D. [I21 J.C. Duplessy, N.J. Shackleton, Labeyrie. D. Oppo and N. Kallel, Deepwater source variations during the last climatic cycle and their impact on the global deepwater circulation, Paleoceanography 3, 343-360, 1988. [I31 E.A. Boyle and L.D. Keigwin, North Atlantic thermohaline circulation during the past 20,000 years linked to high-latitude surface temperature, Nature 330. 35-40, 1987. [I41 J. Jouzel, R.D. Koster, R.J. Suozzo and G.L. Russell, Stable water isotope behaviour during the LGM: a GCM analysis, J. Geophys. Res. 99, 25,791-25,801, 1994. [I51 J. Hansen, G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy and L. Travis, Efficient three-dimensional global models for climate studies: Models I and II, Mon. Weather Rev. 111, 609-662, 1983. [I61J. Hansen, A. Lacis, D. Rind, G. Russel, P. Stone. I. Fung, R.
605
Ruedy and L. Lenner, Climate sensitivity analysis feedback effects, in: Climate process and climate sensitivity. J. Hansen and T. Takahashi, eds., Geophys. Monogr. 29, 1984. [I71 L. Merlivat and J. Jouzel, Global climatic interpretation of the deuterium-oxygen 18 relationship for precipitation, J. Geophys. Res. 84. 5029-5033. 1979. 1181W.S. Broecker, T.H. Peng, J. Jouzel and G. Russell, The magnitude of global fresh-water transports of importance to ocean circulation, Climate Dynam. 4, 73-79, 1990. [I91 A. Baumgartner and E. Reichel, Mean annual global. continental and marine precipitation, evaporation and run-off, in: The World Water Balance, Elsevier, Amsterdam, 1975. DO1 F. Zaucker and W.S. Broecker. The influence of atmospheric moisture transport on the fresh water balance of the Atlantic drainage basin: General Circulation Model simulations and observations. J. Geophys. Res. 97. 2765-2773, 1992. Dll A.H. Oort, Global Atmospheric Circulation Statistics, 180 pp., Rep. US Government print office, NOAA Prof. Pap. 14, 1983. D21 CLIMAP, Seasonal reconstructions of the Earth’s surface at the last glacial maximum. GSA Map and Chart Series. MC-36, GSA, Boulder, CO, 1981. b31 L.D. Labeyrie, J.C. Duplessy and P.L. Blanc, Variations in mode of formation and temperature of oceanic deep waters over the past 125,000 years, Nature 327. 477-482, 1987. t241 N.J. Shackleton, Oxygen isotopes, ice and sea level, Quat. Sci. Rev. 6, 183-190, 1987. sea level D51 R.G. Fairbanks, A 17,000-year glacio-eustatic record: influence of glacial melting rates on the younger Dryas event and deep-ocean circulation, Nature 342, 637642, 1989. L’61 N.J. Shackleton, Attainment of isotopic equilibrium between ocean water and the benthonic foraminifera genus Ucigerina: isotopic changes in the ocean during the last glacial, Colloq. CNRS 219. 203-210, 1974. 1271J.C. Duplessy, L. Labeyrie. A. Juillet-Leclerc, F. Ma?tre, J. Duprat and M. Sarnthein, Surface salinity reconstruction of the North Atlantic Ocean during the last glacial maximum, Oceanol. Acta 14, 311-323, 1991. D81 R. Zahn, Spatquartlre Entwicklung von Kiistenauftrieb und Tiefenwasserzirkulation in Nordost-Atlantik. Rekonstruktion anhand stabiler Isotope kalkschaliger Foraminiferen, Ph.D. Thesis, Univ. Kiel. Kiel, 1986. 1291W.L. Prell, The stability of low-latitude sea-surface temperatures: an evaluation of the CLIMAP reconstruction with emphasis on the positive SST anomalies, Rep. TR 025. US Dept. Energy, Washington, DC, 1985. [301 L. Wang, M. Sarnthein, J.C. Duplessy, H. Erlenkeuser, S. Jung and U. Pflaumann, Paleosea surface salinities in the low-latitude Atlantic: the 6 “0 record of Globigerinoides ruber (white), Paleoceanography 10. 749-761, 1995. [311 J. Farrell and W.L. Prell, Climatic change and CaCO, preservation: an 800.000-year bathymetric reconstruction from the central Equatorial Pacific Ocean. Paleocanography 4. 447-466, 1989.
