QUATERNARY RESEARCH 37, 346-360 (1992)

Sea Level, Surface Salinity of the Japan Sea, and the Younger DryasEvent in the Northwestern Pacific Ocean

L. D. KEIGWIN

Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543

AND

S. A. GORBARENKO

Pacific Oceanological Institute, Far Eastern Branch, USSR Academy of Sciences, 7 Radio Street,Vladivostok, 690032 USSR

Received January 28, 1991

The Japan Sea was profoundly different during glacial times than today. Available &180evidenceindicates that sea surface salinity was lower by several per mil. This probably increased thestability of the water column and caused anoxic sedimentary conditions in the deep sea, as shownby the absence of benthic microfossils and the presence of laminated sediment. These changes arelikely related to the effects of late Quaternary sea-level change on the shallow sills (ca. 130 m)across which the Japan Sea exchanges with the open ocean. The Hwang He (Yellow River) haspreviously been implicated as the source of fresh water to the Japan Sea during glaciation, but thepossible roles of the Amur River and excess precipitation over evaporation must also be consid-ered. Ambiguous radiocarbon chronologies for the latest Quaternary of Japan Sea cores do notadequately constrain the timing of salinity lowering. Previous studies have suggested that lowestsea surface salinity was achieved 27,000 to 20,000 14C yr B.P. However, if global sea-level fallrestricted exchange with the open ocean circulation, then lowest salinity in the Japan Sea may haveoccurred as recently as 15,000 to 20,000 yr ago when sea level was lowest. If this alternative iscorrect, then as sea level abruptly rose about 12,000 yr ago, relatively fresh water must have beendischarged to the open Pacific. This might have affected the dynamics of outflow, local faunal andfloral expression of the polar front, and stable isotope ratios in foraminifera. These environmentalchanges could be misinterpreted as evidence for the cooling of Younger Dryas age, which has notbeen identified in nearby terrestrial records. 'iJ1992Universityof Washinglon.

INTRODUCTION

Study of nearshore marine paleoclimaticrecords is especially important becausethey reflect both terrestrial and marine pro-cesses, and through their study the two canbe linked. Where nearshore locations aresomewhat isolated from the open ocean, asin marginal seas, local climate influencesthe local hydrography, geology, and biol-ogy. Whether or not local hydrographicchanges in marginal seas become importanton the larger (ocean basin) scale dependson the particular geographical situation.For example, the modern outflow of salinewater from the Mediterranean Sea mayhelp precondition the North Atlantic for

0033-5894/92 $5.00Copyright 'iJ 1992 by Ihe University of Washington.All rights of reproduction in any form reserved.

deep-water production (Reid, 1979). Al-though no deep water is produced in theNorth Pacific Ocean at present, hydro-graphic conditions in the Japan Sea and theOkhotsk sea may be critical to the produc-tion of Pacific Intermediate Water (Reid,1973; Riser, 1990; Talley, 1991).

Recently, Younger Dryas-type climaticoscillations have been reported in marinesediments near Japan. Using foraminiferaland stable isotopic techniques, Chinzei etat. (1987) and Kallel et at. (1988) claim tohave identified a cooling episode that oc-curred 10,500 yr B.P., but in some of thevery same sediment cores Heusser andMorley (1990) failed to find pollen evidencefor a climatic reversal. It is important to

346

JAPAN SEA PALEOCEANOGRAPHY 347

identify and interpret deglacial events out-side of the North Atlantic region, wherethey are best known, in order to determinewhether they were regional or global phe-nomena (Rind et aI., 1986).

The Japan Sea experienced large envi-ronmental changes from glacial to intergla-cial time at least in part due to the silledstraits through which its waters exchangewith the open Pacific. Stable isotopic evi-dence for these changes was first publishedin the Japanese (Oba et al., 1980) and So-viet (Gorbarenko, 1983) literature, butthose (and subsequent) contributions havenot yet seen wide exposure in the Englishlanguage literature. In this paper we reviewsome of the latest Quaternary paleoenvi-ronmental changes in the Japan Sea, anddiscuss possible mechanisms linking sea-level change driven primarily by North At-lantic events and a Younger Dryas-type os-cillation in the northwest Pacific off thecoast of Japan.

PHYSICAL SETTING

Japan Sea water exchanges with waterfrom adjacent marginal seas and the openPacific Ocean through four straits (Fig. 1).A good review of the general circulationand water masses of this sea can be found inMoriyasu (1972). The broad TsushimaStrait, with a sill depth of about 130 m, sep-arates the Japanese islands from Korea andis the principal conduit for marine water en-tering the sea (Fig. 2A). The Tsugaru Strait(Sangarsky St. in Russian nomenclature),between Honshu and Hokkaido, has a silldepth of 130 m, and is the main outlet fromthe Japan Sea (Fig. 2B). The TsushimaWarm Current originates as a branch of theKuroshio Current and flows in through theTsushima Strait and out through the Tsu-garu Strait as the Tsugaru Warm Current.Geostrophic estimates suggest that each ofthese currents accounts for a net flow ofabout 1 or 2 Sv (1 Sv = 106 m3 sec -1, Yi,1966; Toba et al., 1982; Shuto, 1982). Flowthrough the other two straits is small bycomparison because of much shallower

sills. The Soya Strait (Laperuza St. in Rus-sian), between Sakhalin and Hokkaido, hasa sill depth of 55 m and the Tartarsky Strait(Mamiya St. in Japanese), between Sakha-lin and Asia, has a sill of only 12 m.

Minato and Kimura (1980) modeled thedynamic interaction between a marginal seaand a western boundary current to deter-mine why the net flux of surface waters inthe Japan Sea is from south to north. Theyfound that the flow is from the TsushimaStrait to the Tsugaru Strait due to the pres-sure difference between the two straitscaused by the wind-driven circulation.Thus, in essence, accumulation of water inthe East China Sea by the subtropical gyredrives the Tsushima Warm Current.

Since only near-surface waters exchangebetween the Japan Sea and other waterbodies, the sea develops its own peculiarhydrographic properties below the upperfew hundred meters (Yasui et al., 1967;Moriyasu, 1972). Waters below the surfacelayer display little variability, with low tem-perature (0-1°C), low salinity (34.05%0),andhigh oxygen content (5.0-5.5 ml/liter). Thewell-mixed cold, fresh, oxygenated deepwaters are a result of a wintertime convec-tion in the northern part of the basin and asouthward return flow along the coast ofSiberia (Vasiliev and Makashin, 1991).Only near-surface waters exhibit significantseasonal and spatial variability, mostly re-flecting conditions in their East China Seasource. As first noted by Niino et al. (1969;see also Ujiie and Ichikura, 1973), the un-usual hydrography of the Japan Sea is re-flected in the unusually low sedimentarycontent of organic carbon, and an unusuallyshallow calcite compensation depth.

METHODS

Even though the analytical results dis-cussed below have been reported else-where (aba, 1984; Oba et al., 1991; Gor-barenko, 1983, 1987), it is important to re-view the methods prior to discussing thedata. Stable isotope data from Japanesecores represent a group effort, having been

348 KEIGWIN AND GORBARENKO

55

OKHOTSKSEA

50

40

0L

500kmJ

. 30130 140 150

FIG. 1. Map of Japan Sea and surrounding regions showing core locations discussed in text. Thesolid contour is the 200-m isobath; the 50-m isobath is shown only in the region of the Tatarskiy(Mamiya) Strait to the north (dotted contour). Core locations marked as solid circles are from the workof Oba and colleagues, and cores marked as open circles have been studied by Gorbarenko andcolleagues. Numbers next to the Japan Sea core locations are the minimum 1)180 values at eachlocation which reflect glacial maximum conditions at the sea surface. Core locations marked withtriangles are from Chinzei et al. (1987), and the square between Honshu and Hokkaido marks theposition of core CH84-14 (KalleI"et al., 1988). A bar symbol in the lower left of the figure (East ChinaSea) marks a seismic line which shows evidence for a former position of the Hwang He channel(Milliman et al. 1989). A dashed circle between Honshu and Hokkaido illustrates the position of theTsugaru Warm Current when it is in the "gyre mode" (Conlon, 1982).

analyzed in three laboratories, using VGMicromass 903 and Finnegan MA T251mass spectrometers (Oba et ai., 1991).Sample dissolution was at 60°C, with no

pretreatment, on specimens which were ei-ther > 177 11mor >250 11m.Where possible,analyses were on individual species, but inintervals of very low abundance mixed

349JAPAN SEA PALEOCEANOGRAPHY

129' 130'

JAPAN SEA

A

140' 141'

35'

40'

41"

34'HONSHU

0L

B

FIG. 2. (A) Map of the Tsushima Strait (after Ujiie, 1973) showing the IIO-m isobath (solid line) andthe 120-m isobath (dashed line). With a sea-level fall of 120 m, flow through the Tsushima Strait wouldbe restricted to a narrow passage west of Tsushima Island. A 130-m drop in sea level would eliminatethe passage in the absence of any tectonic effects. (B) Map of the Tsugaru Strait (after Mogi, 1979 anda map published by the Japan Railway Construction Corp.) showing the 150-m isobath (solid contour)and the location, width, and depth of three "choke points." The dashed contour marks the 130-misobath. Sea-level fall of 130 m would result in a passage only 3 km wide at the northernmost sill.

planktonic specimens were used. Gor-barenko analyzed mixed specimens ofplanktonic foraminifera on a MU 1309 massspectrometer, after dissolution "off-line"at 50°C. His results are related to PDBthrough analysis of SMOW and a labora-tory carbonate standard. Radiocarbon de-terminations on both Soviet and Japanesesediment cores were made on carbonatecarbon in bulk sediment.

CORE RESULTS

Of the several Japan Sea sediment coresthat have been reported by Oba, Gor-barenko, and their colleagues (Fig. 1), wewill describe results of one Japanese coreand one Soviet core. These cores have thehighest resolution paleoenvironmentalrecords and are believed to be most repre-

sentative of the stable isotopic history ofthe late Quaternary of this region.

Oba's (1984) results from a site on OkiBank (37°04'N, 134°42.4'E, 935 m) are pre-sented in Figure 3. Cores KH79 L-3 andC-3 were taken from almost exactly thesame location. Because L-3 had the higherlatest Quaternary deposition rate and wasradiocarbon-dated, its record is patched tothe longer record from C-3 at 215 cm usingthe distinctive lithologic changes that markJapan Sea cores (henceforth, and in Fig. 1,the composite record of these two coreswill be referred to as C-3). These includethree layers of volcanic ash that are datedon land (Ah = 6300 yr; Oki = 9300 yr; andAT = 21,000 to 22,000 yr B.P.; Machidaand Arai, 1983), and the top of a thick in-terval of finely laminated sediment.

Maximum enrichment in 180 in plank-

350 KEIGWIN AND GORBARENKO

+ +++

~ 400e~..c:-Q.,II>

Q 600

.:+

+ ++

800

+e

+

+++

+ + +e

Mixed planktonics 0

N. pachyderm a (R) mG. umbilicata +

1000

FIG. 3. Oxygen isotope values for latest Quaternaryplanktonic foraminifera from core C-3 (Oba, 1984). Al-though this is a composite record made of two speciesand mixed assemblages, N. pachyderma (right coiling)records the signal from the maximum in 1)180to thecore top, and G. umbilicata records the full range of1)180below core top values.

tonic foraminifera (mostly N. pachydermaand G. umbilicata) in core C-3 is found atabout 100 cm and below about 300 cm (Fig.3). Lowest &t80 values for the entire coreare found at about 150 cm, and values about0.5%0heavier are found throughout the up-per 50 cm. Although in the whole core thereare only two pairs of stable isotope analy-ses that allow direct comparison of N.pachyderma and G. umbilicata, the latterspecies records the full oxygen isotopic os-cillation from heavy to light to heavy valuesbetween 100 and 300 cm (Fig. 3). For sim-plicity, G. umbilicata, N. pachyderma, andthe mixed planktonics are considered to-gether in Figure 4A.

Benthic foraminifera are commonthroughout most of core C-3, excepting theinterval 132 to 255 cm which includes thelaminated sediment (Fig. 4A). Above 132cm, only Cassidulina japonica were ana-lyzed, whereas deeper analyses were

mostly on species of Uvigerina and on Cas-sidulina norcrossi. Considering just the&180pattern of C. japonica, the signal am-plitude (about 1.6%0)is less than half that ofthe planktonic foraminifera from the sameinterval. Minimum values are found at 112cm, above the planktonic minimum, andthe maximum occurs at 76 cm (Fig. 4A).

In general, the pattern of &180changes inmixed planktonic foraminifera from core1670 (north slope ofYamato Rise; 39°48'N,133°27'E, 1105 m; Gorbarenko, 1987) is thesame as that found in Oba's cores (Fig. 4B).A few analyses of G. bulloides and N.pachyderma (sinistral) compare favorablywith the mixed planktonic data (Gor-barenko and Matunina, 1991). Maximum&180is found at 70 cm and below about 200cm, with minimum values found not at thecore top (as in core C-3) but at about 100 cm(Fig. 4B). Fine-scale lithological features,such as tephra Ah, and sediment laminationwere not recognized because the core wasnot split before sampling. Radiocarbonanalyses of core 1670 (Fig. 4B) were pub-lished first by Derkachev et al. (1985).

DISCUSSION

Stratigraphy and Chronology of JapanSea Paleoceanographic-Events

Results on cores C-3 and 1670 (and manyothers not discussed here) are so unlikethose found in typical deep sea cores fromthe open ocean that the conventional &180stratigraphy cannot be applied. For exam-ple, in open ocean cores the lowest &180inthe latest Quaternary is usually found at ornear the core top, but in Japan Sea cores itmay be as deep as 100 to 200 cm within thecore. Nevertheless, creating a local stratig-raphy for Japan Sea cores is simple usingboth the &180extrema and volcanic ash lay-ers identified as AT and Oki (Fig. 6).

A chronology is less certain because theradiocarbon age determinations performedon these cores do not agree very well. Incore 1670 the &180minimum has a 14Cageof about 15,000 yr (Fig. 4B), whereas the

48180

30 I . I

2 I. I0

+ 'ye .. 00 +

0+ +

200 -I ++0

+

0 .

JAPAN SEA PALEOCEANOGRAPHY 351

4

81802 03

1.9 :i~

~~.ec21

..~

$:~. I!I

.' . I!IEJ...

~~

11.314.215.4

'I!I

32.4

34.6

B

FIG. 4. Oxygen isotope results for cores C-3 (A) and 1670 (B). In each core the stippled patternalong the left axis shows the location of volcanic ash layers, the horizontal-ruled pattern shows theextent of laminated sediment (not noted in core 1670), and the numbers represent radiocarbon agedeterminations on bulk sediment (in l(}oo yr B.P.). In core C-3 planktonic species (Fig. 3) have beenlumped. Benthic species above 132 cm are Cassidulinajaponica; and below 250 cm analyses are basedon Uvigerina and Cassidulina spp. The cooccurrence of low 1)180 in planktonic foraminifera, theabsence of benthic foraminifera, and the presence of laminated sediment indicates that during glacialtimes sea-surface salinity was reduced, deep-sea ventilation was reduced, and the benthic macrofaunawas eliminated. Oxygen isotope results are for mixed planktonic assemblages in core 1670.

same event in core C-3 is dated at about20,000 yr B.P. (Fig. 4A). Identification ofvolcanic ash beds by Oba, Gorbarenko, andtheir co-workers indicates that these 0180minima must correlate (Fig. 5A). One pos-sible source of the age differences could bethe 14Canalysis of bulk sediment which isusually required in conventional radiocar-bon analysis of pelagic marine sediments. Itis well known that 14C age determinationson bulk sediment are frequently plagued bycontamination from detrital carbonate thatmay be infinitely old with respect to 14C(e.g., Keigwin et al., 1984; Jones et al.,1989). In core C-3, Oba (1983) reported a14Cage of 22,000 j: 200 yr B.P. for an in-terval centered on 169 cm, 64 cm above ashAT which has the same age, based on morethan 20 14Cdates which bracket the ash on

land (Machida and Arai, 1983). Oba et al.(1991) report that up to 30% of the cocco-liths in the upper laminated sediments arepre-Quaternary in age, supporting the pos-sibility of detrital carbonate contaminationof their 14Cages. In Oba's previously pub-lished work and in Oba et al. (1991) it isassumed that the 22,000 yr 14Cage for the169 cm level is accurate because Kato(1984) reported an age for ash AT of 26,000yr in a core from the East China Sea. Be-cause that age was determined by bulk ra-diocarbon dating of sediment above and be-low the ash layer, contamination by detritalcarbonate is possible at that location aswell. Although core 1670was not examinedfor detrital carbonate content, its basin cen-ter location and lower rate of sedimentationsuggest that the younger (15,000 yr B.P.)

8180

5 4 3 2 1 0

Or! DI!I

-I!Iki I!I

. 10.8100

--- -I%J

I I'"

BenthicPl/.; 22.2c.

200

j,300 LLlJ

A

352 KEIGWIN AND GORBARENKO

00

1670 depth (em)100 200

100

0180 minimum

e 200~---.c....c.~

"Cf") 300

U

AT ash

400

500A

300 4

0180

2 03

core C-3

core 1670

BFIG. 5. (A) Correlation of volcanic ash layers and features in the oxygen isotope records between

cores 1670 and C-3. Levels in core 1670 can be scaled to those in C-3 by multiplying by 1.44. (B)Composite 1)180record from cores C-3 and 1670, after scaling the latter. Results from the two coresare in excellent agreement at the oxygen isotopic maximum and minimum. Local hydrographic dif-ferences at the core sites and sampling differences probably account for differences between theisotopic records.

age for the minimum in 1)180may be accu-rate.

In this paper we assume that the upper-most laminated interval in Japan Sea cores,which contains the minimum 1)180in plank-tonic foraminifera, is equivalent to the levelof maximum latest Pleistocene glaciationmarked in open-ocean cores by maximum1)180. Thus, the greatest oceanographicchange in the Japan Sea may have coin-cided with the period of harshest terrestrialand atmospheric conditions, which per-sisted from about 20,000 to 15,000 yr B.P.This interpretation is consistent with the ra-diocarbon dating on core 1670, and is like-wise consistent with the terrestrial age forvolcanic layer AT in core C-3. Firm estab-lishment of a Japan Sea chronology for thelatest Quaternary must await acceleratormass spectrometry (AMS) 14C dating ofhand-picked foraminifera from these cores.

Our proposed correlation of "events" inFigure 5A allows scaling of depth in core1670 to that of C-3 by multiplying theformer by 1.44. It also indicates that theremay be about 20 em missing from the top ofcore 1670, although this is not supported bya young 14Cage (1900 yr) near the core top.Scaling core 1670 to C-3 and plotting thetwo records on a common depth axis showsexcellent agreement between the pro-nounced 1)180maximum and minimum withimportant differences above and below the1)180oscillation (Fig. 5B). Such differencesmight result from different sampling inten-sities of the two cores in the deeper inter-val, the choice and mixture of species, anddifferences in Holocene properties of near-surface waters at these two locations (Fig.1). For example, Holocene planktonic fora-minifera at the basin-center location of core1670 are subject to cooler waters outside

JAPAN SEA PALEOCEANOGRAPHY 353

the influence of the warm Tsushima Cur-rent. Summer sea-surface temperatures(SST) at the location of core 1670 are about3°C lower than over core C-3, and at 100 mthis difference increases to 10°C (Yasui etal. 1967).

Late Quaternary Environmental Changesin the Japan Sea

Late Quaternary changes in sea levelmust have had profound effects on the hy-drography of the Japan Sea since the silldepths of the Tsushima and Tsugaru Straits(about 130 m) are close to the 120-m fall ofeustatic sea level 15,000 14C yr ago docu-mented by Fairbanks (1989). Circulationthrough narrow straits no deeper than 10mwas probably restricted and may have beencompletely shut off if maximum sea-levellowering was greater than 120 m or if therewere any local tectonic effects at thestraits. As discussed previously by Oba etal. (1980, and in subsequent papers) andGorbarenko (1987), the 8180 of planktonicforaminifera during glacial times in the Ja-pan Sea was so low that salinity must havebeen lowered in near-surface waters. With-out the influence of salinity on 8180, unrea-sonably high SSTs would be indicated.Low salinity probably could have beenmaintained only if the basin was in re-stricted communication with saltier watersfrom the Pacific surface. Thus, we disagreewith the conclusion of Morley et al. (1986)that the Japan Sea must have been in opencommunication with the Pacific during fulllowering of glacial sea level because of thepresence of open sea radiolaria and forami-nifera. An alternate interpretation is thatthe Japan Sea was large enough to maintainits own planktonic fauna and that the faunawas able to live in waters of reduced salin-ity.

There are only two ways of lowering sur-face ocean salinity: (1) through increasedprecipitation relative to evaporation and (2)through dilution by iceberg or river dis-charge. Iceberg flux can probably be ruled

out as a cause of the 8180 minimum in theJapan Sea because of the lack of ice-raftedsediment in the laminae of these cores. Oba(1983) and Chinzei and Oba (1986) pro-posed that during glacial lowering of sealevel the mouth of the Hwang He (YellowRiver) moved seaward and discharged freshwater very close to or into the TsushimaStrait. However, Milliman et al. (1989) con-ducted a seismic survey of the East ChinaSea and found what is probably an ancientchannel of the Hwang He seaward of thedeepest entrance to the Tsushima Strait(Fig. 1). This indicates that the river mayhave bypassed the Tsushima Strait at sometime in the past, although it still could haveinfluenced the salinity of local waters in theEast China Sea. Oxygen isotope resultsfrom a core in the East China Sea (core1595; Fig. 1), however, show a typical 8180stratigraphy which Gorbarenko (in press)cites as evidence of normal sea-surface sa-linity.

Possible influence of the Amur River(Fig. 1) on Japan Sea salinity should be con-sidered. The Amur River drains a substan-tial region of the Asian continent adjacentto the Japan Sea and has an annual dis-charge of 325 km3 yr-l (Milliman andMeade, 1983). Today, the Amur River emp-ties into the middle of the Tartarsky Straitand its discharge flows toward the OkhotskSea, but this pattern might have changedduring glacial times with lowered sea level.If the Amur River influenced Japan Sea sa-linity during the late Quaternary, then wewould expect the interval of low 8180 tohave begun earlier and lasted longer be-cause for most oflatest Quaternary time theshallow sill at Tartarsky Strait (12 m) wouldhave been flooded. Seismic studies shouldbe conducted to trace the former channel ofthis river and locate its fan, and additionalcoring of the Japan Sea should be con-ducted in higher latitudes to test for gradi-ents in 8180.

Gorbarenko (in press) recently assertedthat lowered surface salinity in the JapanSea could have resulted from increased pre-

354 KEIGWIN AND GORBARENKO

cipitation relative to evaporation duringglacial times rather than transport of low-salinity surface waters through the Tsu-shima Strait. Making assumptions aboutthe mass balance, Gorbarenko (in press) es-timates that salinity may have been 6 to 7%0lower than today. In addition, he mappedminimum glacial 8180 to show its limitedgeographic variability, with no evidence oflowest values nearest the Tsushima Strait.Combining the available Japanese and So-viet data on the 8180 minimum indicates arange of only 0.7%0 with no evidence forlowest values closest to the Tsushima Strait(Fig. 1). Within the limits of the availabledata, this indicates that either there was noone point source (i.e., river) of fresh waterdischarge to the southern Japan Sea duringglacial maximum times, or that surface wa-ters were very well mixed.

Assuming restricted exchange betweenthe Japan Sea and the open ocean, a simplecalculation shows that the modern excessof precipitation over evaporation (P-E)over the Japan Sea (about 30 mm yr-1;Baumgartner and Reichel, 1975) is morethan sufficient to account for glacial lower-ing of sea-surface salinity. Multiplying P-Eby the area ofthe Japan Sea (0.9 x 106km2)gives a net flux of fresh water of 27 km3yr-1 which is comparable to the freshwaterdischarge of the Hwang He (40 km3 yr - 1;

Milliman and Meader, 1983). If we assumethat the upper 100 m of the sea was fresh-ened, then the volume under considerationwas 90,000 km3 (neglecting reduced surfacearea due to lowered sea level), which couldbe completely replaced by fresh water inabout 3000 yr. Unfortunately, available cli-mate model results do not resolve P-E foran area as small as the Japan Sea, but thereis some evidence for a slight increase overmodern values in the annual zonal meanduring glacial times (Kutzbach and Guetter,1986).

Local pollen data, however, indicate thateffective precipitation in the coastal zone ofthe Japan Sea at about 36°N was reducedby ~50% relative to today (Tsukada, 1986).

Even if precipitation reduction of thisamount was typical of the entire Japan Sea,there would still be sufficient excess P-E tohave substantially lowered the salinity ofsurface waters during maximum glaciation.Thus, it is not necessary to call on theHwang He as a source of fresh water.

Substantial changes in the salinity of theJapan Sea surface during glacial time musthave affected the convection which venti-lates that basin today. As first noted byUjiie and Ichikura (1973), at some depth be-low the seafloor Japan Sea sediments arelaminated. Both Gorbarenko and Oba pro-posed that low-salinity surface waters musthave stratified the water column, prevent-ing convection in the northern part of thesea and leading to deep anoxia. This hy-pothesis accounts for the laminated sedi-ments and the absence of benthic foraminif-era in the interval marked by anomalouslylow 8180 in core C-3 (Fig. 4A) and in manyother cores. Both Gorbarenko (1987) andOba et al. (1991) presented other sedimen-tary evidence for restricted convection dur-ing glaciation. Their evidence includeshigher carbonate and authigenic pyrite con-tents in the low 8180 interval, which theyattributed to better preservation during sul-fate reduction on the seafloor, and higheropal and organic C content during the de-glacial and Holocene intervals due to in-creased surface water productivity.

Oba (1984) showed that benthic forami-nifera reappear in C-3 at 132 cm, just abovethe laminated sediment and the planktonic8180 minimum (150 cm). Shortly after theyreappeared, the 8180 of C. japonica de-creased, at a level where 8180 of planktonicspecies was increasing (Fig. 4A). Oba(1984) interpreted this pattern of change toindicate that as sea level increased and ex-change with the open ocean resumed, deepconvection was renewed and water oflower salinity (but not as low as at full-glacial times) was carried to depth.

Marine conditions in the Japan Sea be-fore the deposition of the latest interval oflaminated sediments (equivalent to 8180

JAPAN SEA PALEOCEANOGRAPHY 355

stage 3) must have been unlike those of ei-ther the Holocene or the glacial maximum.Planktonic foraminifera had greater 0180than in the Holocene, suggesting communi-cation with a colder open ocean than todayor higher local salinity, but benthic forami-nifera had 0180 values similar to the Ho-locene (Fig. 4A). Although global sea levelis not known with the same precision be-fore 20,000 yr B.P. as it is since then, itwas probably not as low as that at glacialmaximum times. Deep convection in the Ja-pan Sea must have been reduced, judgingfrom the scarcity of benthic foraminiferaand the occasional presence of laminae. Itis likely that salinity was lower in the regionof deep convection because of the rela-tively low 0180 of benthic foraminifera.This scenario would require a significantlatitudinal gradient in surface salinity forpre-glacial maximum conditions or ventila-tion of the basin at lower latitudes.

We propose that the abrupt maximum inplanktonic 0180 values following the glacialminimum in 0180 reflects the reestablish-ment of normal marine conditions duringlatest glacial time. Previously, Oba (1983)proposed that maximum 0180 (with mini-mum ol3C) and cold faunal elements wereconsistent with flushing of the basin fromthe north with cold subpolar (Oyashio) wa-ters through the Tsugaru Strait. However,during the glacial maximum the sense of thepressure differential between the twostraits produced by the wind-driven oceancirculation was probably the same as today:the North Equatorial Current would stillhave caused higher pressure at the Tsu-shima Strait and southerly advance of thepolar front may have caused even lowerpressure at the Tsugaru Strait. Thus, build-ing on the model of Minato and Kimura(1980), we speculate that as soon as ex-change with the open ocean was renewed,the flow must have been from south tonorth as it is today. In this alternate sce-nario the maximum 0180 in the Japan Searesulted from renewed exchange with openocean waters still enriched in 180 by the

effect of excess continental ice volume. By11,000 yr ago, deglaciation was only halfcomplete and open marine waters were stillenriched in 180 (by about 0.6%0).Low ol3Cat that time was probably due to the sea-water-reservoir effect (currently thought tobe a few tenths per mil; Duplessy et al.,1988; Curry et al., 1988).

Possible Influence of the Japan Sea onSurrounding Regions

The following discussion assumes thatthe interval of minimum 0180 in the JapanSea was coincident with maximum sea-level lowering. Discussion of the sequenceof events in the "older" chronology (mini-mum 0180 from about 27,000 to 20,000 yrB.P.) can be found in Oba et al. (1991).

When the rate of sea-level rise increasedabruptly 12,000 yr ago (Fairbanks, 1989),low-salinity surface water must have beenflushed from the Japan Sea to the PacificOcean via the Tsugaru Strait. Sedimentcores close to the Tsugaru Strait outflowmight be expected to record some evidenceof that event. Recently, such a core (CH84-14) from the bight between Honshu andHokkaido was described by Kallel et al.(1988; Fig. 1). AMS 14Cdates on planktonicforaminifera from that core showed that adistinct event of low 0180 occurred about12,000 yr ago (Fig. 6). Kallel et al. (1988)found decreased abundance in the polarplanktonic foraminiferal species N. pachy-derma (sin.) at about the same time, andlinked that event indirectly with tempera-ture by suggesting that the polar front musthave shifted northward. Kallel et al. de-scribed a read vance of the polar front atYounger Dryas time based on increasedabundance of N. pachyderma and in-creased 0180.

Polar front movement was discussed pre-viously by Chinzei et al. (1986), based onthree sediment cores taken from the Pacificcoast of Honshu (Fig. 1). Those authors re-ported faunal and floral evidence for asouthern advance of the polar front at

356 KEIGWIN AND GORBARENKO

0180 G. bulloides (---8-)5 4 3 2

8

~ 10~.......

M0 12....

'"OJ)

< 14

16100 90 80 70

% N. pachyderma (right)total N. pachyderma

(--)

FIG. 6. Late-glacial to early Holocene oxygen iso-tope results for core CH84-14 in the bight betweenHonshu and Hokkaido (after Kallel et al., 1988). Theisotopic minimum and minimum in abundance of N.pachyderma (left coiling), which are AMS 14C-dated toabout 12,000 yr B.P., may be a response to abruptsea-level increase and discharge of low-salinity surfacewater from the Japan Sea at that time. If that hypoth-esis is correct, then the isotopic maximum and thefaunal maximum between 11,000 and 10,000 yr B.P. donot necessarily reflect cooling similar to the YoungerDryas event in the North Atlantic region.

Younger Dryas time in northern cores C-land C-6, but not in core C-4. A comparable0180 oscillation was found only in the mid-dle core (C-6), although it might have beenmissed in the northern core (C-1) which isonly about 12,000 yr old at the bottom.

There is reason to believe that a link ex-ists between faunal evidence for localmovement of the Kuroshio-Oyashio (polar)front, deglacial 0180 changes in planktonicforaminifera, and discharge from the JapanSea. Conlon (1982) pointed out that theTsugaru Warm Current exits the Japan Seain two modes. In the first mode, occurringduring summertime, the outflow forms ananticyclonic gyre about 100 km in diameter,that dominates the gulf between Honshuand Hokkaido (Fig. 1). In the second mode,the flow turns abruptly southward along thecoast of Honshu. Conlon (1982) proposed

that the prevalence of either mode is gov-erned by the Rossby radius of deformation,R = (g(tlp/p)D)I/:J 1, where g is the accel-eration of gravity, tlp is the difference indensity between the surface and deeper lay-ers, D is the thickness of the surface layer,andfis the Coriolis parameter. In addition,Conlon speculated that the presence of thegyre mode might effectively block the pen-etration of Oyashio waters into coastal Ja-pan south of Hokkaido, thus controlling thelocal expression of the polar front.

Since the important variables in Rare (1)the thickness of the surface layer and (2) thedensity contrast, and since these parame-ters must have changed seaward of theTsugaru Strait over the last 20,000 yr as sealevel and the Japan Sea salinity changed, itis important to estimate how R (and the lo-cal polar front) might have changed on de-glaciation. Although Ichiye (1982) ques-tioned the strict dependency of circulationmode on R because the modern pycnoclineis not very sharp seaward of the TsugaruStrait, the pycnocline must have been muchsharper 12,000 yr ago due to the salinitycontrast between the Tsugaru outflow andthe open Pacific waters.

For the following analysis we make twoimportant assumptions: (1) that sill depth inthe Tsugaru Strait approximates the sur-face-layer thickness (D) and (2) that tlp isaffected mostly by salinity (for our calcula-tions we assume a temperature of 10°C).Figure 7 illustrates the dependency of R onlayer thickness and the salinity contrast be-tween outflowing low-salinity water fromthe Japan Sea and the underlying Pacificwater. In the modern situation, Conlon(1982) found that the gyre mode dominatedfor conditions where R > 15 km, and thecoastal mode was found for R < 10 km.Thus, for example, with outflowing surfacewaters less dense due to 5%0lower salinity,the gyre mode would prevail for layer thick-nesses greater than about 60 m (Fig. 7).

Following the late Quaternary maximumlowering of sea level (120 m) and minimumsill depth in the Tsugaru Strait (10 m), sea

JAPAN SEA PALEOCEANOGRAPHY 357

00 50 100 150 200 250

Surface layer thickness (m)

FIG. 7. Rossby radius of deformation (R) as a func-tion of surface layer thickness and salinity (density)contrast in the Tsugaru Strait outflow for various timesin the past. For reference, the surface layer thicknesstoday (0 yr) is 200 m, the density contrast is small, andfor R > 15 Tsugaru Strait outflow is in the "gyremode" (Conlon, 1982). This situation is typical of sum-mer months, and the gyre blocks cold Oyashio watersfrom the bight between Honshu and Hokkaido (Fig. 1).In contrast, Conlon (1982) found the "coastal mode"of circulation prevailed in winter months (R < 10),with cold waters invading the bight. As sea level roseabruptly about 12,000 yr ago during deglaciation,Tsugaru outflow would have changed from the"coastal mode" to the "gyre mode," assuming a 5%0salinity contrast between Japan Sea surface watersand Pacific surface waters and a layer thickness thesame as sill depth. This model could account for theapparent Younger Dryas age temperature oscillationbetween about 12,000 and 10,000 yr B.P. found byChinzei et al. (1987) and Kallel et al. (1988).

level slowly increased by 20 m until 12,500yr ago (from Fairbanks' 1989 curve). As-suming for that time a 5%0salinity contrastand a layer thickness of only 30 m, theTsugaru outflow still would have been inthe coastal mode (Fig. 7). In the ensuingmillennium, however, when sea level rap-idly rose an additional 24 m, the circulationregime between Honshu and Hokkaidowould have clearly left the coastal mode,approaching the modern threshold for thegyre mode. R may have actually exceededthe modern threshold for the gyre mode,considering that our assumptions are con-servative. For example, by using only sa-linity in calculating Llp,we may have under-estimated the density contrast. Surfacetemperatures in a well-stratified Japan Sea

300

at 12,000 yr ago, close to the time of thedeglacial maximum in seasonality, mayhave been warmer than those of the openPacific, enhancing the density contrast inthe Tsugaru outflow. In addition, we haveprobably underestimated the thickness ofthe outflowing surface layer by substitutingthe sill depth. Today, the average layerthickness seaward of the Tsugaru Strait isabout 200 m, 70 m greater than the sill depth(Conlon, 1982). Finally, the glacial depres-sion of sea-surface salinity probably ex-ceeded 5%0(Gorbarenko, in press).

The northward movement of the polarfront along coastal Japan 12,000 yr ago(Chinzei et al., 1987; Kallel et al., 1988)possibly was a response to Tsugaru outflowwhich must have intensified at that time assea level rose abruptly. Thus, since 8180evidence for a Younger Dryas-aged oscilla-tion is as much a function of low ratios12,000 yr ago as high ratios between 11,000and 10,000 yr ago, the oxygen isotopic rec-ord in the northwestern Pacific may be sig-nificantly influenced by events in the JapanSea. Although we have provided an alter-nate interpretation for the 8180 record ofthe Younger Dryas in the northwestern Pa-cific, faunal and floral oscillations likethose presented by Chinzei et al. (1987) andKallel et al. (1988) are still important evi-dence for brief events of local cooling.Elsewhere in the Pacific Ocean, AMS ]4C-dated cores from the Sulu Sea (Linsley andThunell, 1990) and the Gulf of California(Keigwin and Jones, 1990)contain YoungerDryas-aged oscillations that most likely re-flect widespread climatic forcing. Our mainpoint in this paper is that caution should beexercised in interpreting parallel records offaunal and isotopic change as evidence forpaleotemperature change where there is thepossibility of significant salinity overprint-ing of a 8]80 record. Our argument is anal-ogous to that made by Fairbanks (1989)concerning interpretation of the YoungerDryas in the North Atlantic, although themechanism is different. In the North Atlan-tic, low 8]80 may have been a direct re-

80

E1 lacialmaximum / 35 %06 : 2,500yr (fresh)

60g'"::I 40-I: : :/ 10%0:e --...... V 5%0.0

20 : :: Gvre mode :0 yr'"'"0..::

358 KEIGWIN AND GORBARENKO

sponse to the stratification changes occur-ring during meltwater events, whereas inthe northwest Pacific off Japan low 8180may be a direct response to sea-level rise.

CONCLUSIONS

A review of Japanese and Soviet paleocceanographic results for cores from the Ja-pan Sea, which are not widely available inthe English language literature, has resultedin some new interpretations of paleoenvi-ronmental change occurring in that basinand surrounding waters during the latestQuaternary.

As recognized previously, lowered sealevel during glaciation isolated the JapanSea because of its shallow sills, leading tolowered sea-surface salinity. With sills inthe Tsushima and Tsugaru Straits at 130 mand sea-level lowering of 120 m, exchangewith the open ocean may have been muchless than the 1-2 Sv estimated for today.Low salinity of surface waters could havebeen maintained by a small excess of re-gional precipitation over evaporation, or bydischarge of the Amur River and/or theHwang He into the isolated basin. Either ofthese mechanisms could have quickly led tostable stratification of surface waters anddeep anoxia. These hypotheses should betested by further coring. Our chronologyfor latest Quaternary events is based pri-marily on knowledge of the timing of globalsea-level change (Fairbanks, 1989) and in-terpretation of how dated sea-level changesmight have affected the Japan Sea hydrog-raphy. This chronology is a testable alter-native to the interpretations presented byOba (1984) and Oba et al. (1991). Our as-sertions should be checked by AMS 14Cdating of foraminifera from the Japan Sea.

By 12,000 yr ago, when the rate of sea-level rise increased rapidly, the Japan Seawas probably flushed of its low-salinity lid.This process may account for low 8180 atthat time in cores from the Pacific side ofJapan, and dynamics of the Tsugaru out-flow may account for evidence for local

movement of the polar front at that time.Thus, Younger Dryas-type oscillations insediment cores off Japan may contain animportant component of local oceano-graphic change that is independent of globalcooling. Sometime after 12,000 yr B.P., butbefore melting of northern hemisphere icesheets was complete, the Japan Sea wasprobably in full exchange with the openocean, which was still partially enrichedin 180.

ACKNOWLEDGMENTS

We thank R. Dunbar, D. Oppo, and L. Peterson forcritical reviews of the manuscript and L. Pratt forhelpful discussions on the dynamics of flow throughstraits. We are especially indebted to T. Oba, whodespite disagreeing with some of our interpretations ofhis data, generously provided his unpublished data,helpful discussions, and comments on the text. S.Honjo and T. Tanaka helped with translating Japa-nese. This research was supported by NSF GrantOCE90-00434.

REFERENCES

Baumgartner, A., and Reichel, E. (1975). "The WorldWater Balance." Elsevier, Amsterdam.

Chinzei, K., Fujioka, K., Kitazato, H., Koizuma, I.,Oba, T., Oda, M., and Okada, H., et al. (1987).Postglacial environmental change of the PacificOcean off the coasts of central Japan. Marine Mi-cropaleontology 11, 273-291.

Chinzei, K., and Oba, T. (1986). Oxygen isotope stud-ies of the deep sea sediments around Japan. RecentProgress of Natural Sciences in Japan 11, 35-43. [inJapanese]

Conlon, D. M. (1982). On the outflow modes of theTsugaru Warm Current. La Mer 20, 60-64.

Curry, W. B., Duplessy, J. C., Labeyrie, L. D., andShackleton, N. J. (1988). Changes in the distributionof S13Cof deep water lCO2 between the last glaci-ation and the Holocene. Paleoceanography 3, 317-341.

Derkachev, N. N., Utkin, I. V., and Gorbarenko,S. A., et al. (1985). Correlation and accumulationrate of sediments of the Japan Sea during and afterlate-glacial time. Pacific Ocean Geology N4, 22-29.[in Russian]

Duplessy, J.-C., Shackleton, N. J., Fairbanks, R. G.,Labeyrie, L., Oppo, D., and Kallel, N. (1988).Deepwater source variations during the last climaticcycle and their impact on the global deepwater cir-culation. Paleoceanography 3, 343-360.

Fairbanks, R. G. (1989). A 17,OOO-yearglacio-eustaticsea level record: Influence of glacial melting rates on

JAPAN SEA PALEOCEANOGRAPHY 359

the Younger Dryas event and deep-ocean circula-tion. Nature 342, 637-642.

Gorbarenko, S. A. (1983). Paleogeographic conditionsin the central part of the Sea of Japan during Ho-locene and late Pleistocene from data on 180/160 in

foraminiferal tests. Oceanology 23, 300-303. [inRussian]

Gorbarenko, S. A. (1987). Paleogeography of the Ja-pan Sea in the upper Pleistocene and Holocene. Iz-vestia Geographical Series 6, 106-113. [in Russian]

Gorbarenko, S. A. (in press). Stratigraphy and lateQuaternary sediments in seas of Japan and Okhotskand their paleooceanological conditions. In "Geol-ogy and Geophysics of the Japan Sea," Japan-USSR Monograph Series I.

Gorbarenko, S. A., and Matunina, T. V. (1991). Pa-leoceanology of the Japan Sea during the late Wurmglaciation and Holocene. In "Startigraphy and Cor-relation of Quaternary Sediments of Eastern Asiaand Pacific Region. Nauka, Moscow. [in Russian]

Heusser, L. E., and Morley, J. J. (1990). Climaticchange at the end of the last glaciation in Japan in-ferred from pollen in three cores from the northwestPacific Ocean. Quaternary Research. 34, 101-110.

Ichiye, T. (1982). A commentary note on the paper"On the outflow modes of the Tsugaru Warm Cur-rent" by D. M. Conlon. La Mer 20, 125-128.

Jones, G. A., Jull, A. J. T., Linick, T. W., and Don-ahue, D. J. (1989). Radiocarbon dating of deepseasediments: A comparison of accelerator mass spec-trometer and beta-decay methods. Radiocarbon 31,105-116.

Kallel, N., Labeyrie, L. D., Arnold, M., Okada, H.,Dudley, W. C., and Dupplessy, J.-C. (1988). Evi-dence of cooling during the Younger Dryas in thewestern North Pacific. Oceanologica Acta 11, 369-375.

Kato, M. 1984. Sediment: Based on analyses of KH79-3, C-3. Symposium on the Japan Sea. Chikyu Earth(Monthly) 6, 520-522.

Keigwin, L. D., Jr., Corliss, B. H., Druffel, E. R. M.,and Laine, E. P. (1984). High resolution isotopestudy of the latest deglaciation based on BermudaRise cores. Quaternary Research 22, 383-386.

Keigwin, L. D., and Jones, G. A. (1990). Deglacial cli-matic oscillations in the Gulf of California. Pale-

oceanography 5, 1009-1023.Kutzbach, J., and Guetter, P. J. (1986). The influence

of changing orbital parameters and surface bound-ary conditions on climate simulations for the past18,000 years. Journal Atmospheric Sciences 16,1726-1759.

Linsley, B. K., and Thunell, R. C. (1990). The recordof deglaciation in the Sulu Sea: evidence for theYounger Dryas event in the tropical western Pacific.Paleoceanography 5, 1025-1039.

Machida, H., and Arai, F. (1983). Extensive ash fallsin and around the Sea of Japan from large late Qua-

ternary eruptions. Journal. Volcanology and Geo-thermal Research 18, 151-164.

Milliman, J. D., and Meade, R. H. (1983). World-widedelivery of river sediment to the Oceans. Journal ofGeology 91, 1-21.

Milliman, J. D., Qin, Y., and Park, Y. A. (1989). Sed-iments and sedimentary processes in the Yellow andEast China seas. In "Sedimentary Facies in the Ac-tive Plate Margin" (A. Taira and F. Masuda, Eds.),Terra Scientific Pub. Co., Tokyo.

Minato, S., and Kimura, R. (1980). Volume transportof the western boundarycurrent penetrating into amarginal sea. Journal of the Oceanological Societyof Japan 36, 185-195.

Mogi, A. (1979). "An Atlas of the Sea Floor aroundJapan," Univ. Tokyo Press, Tokyo.

Moriyasu, S. (1972). The Tsushima Current. In"Kuroshio, Its Physical Aspects" (H. Stommel andK. Yoshida, Eds.), Univ. Tokyo Press, Tokyo.

Morley, J. J., Heusser, L. E., and Sarro, T. (1986).Latest Pleistocene and Holocene palaeoenviron-ment of Japan and it marginal sea. Palaeogeogra-phy, Palaeoclimatology, Palaeoecology 53, 349-358.

Niino, H., Emery, K. 0., and Kim, C. M. (1969). Or-ganic carbon in sediments of Japan Sea. Journal ofSedimentary Petrology 39, 1390-1398.

Oba, T. (1983). Paleoenvironment of the Sea of Japansince the last glaciation. Chikyu (Monthly, TheEarth) 5, 37-46. [in Japanese]

Oba, T. (1984). Japan Sea since the last glacial age:Oxygen and carbon isotopes based mainly on anal-ysis of core KH-79-3, C-3. Chikyu (Monthly, TheEarth) 6, 558-566. [in Japanese]

Oba, T., Horibe, Y., and Kitazato, H. (1980). Analysisof the paleoenvironment since the last glacial agebased on two cores from the Japan Sea. Archaeol-ogy and Natural Science 13, 31-49. [in Japanese]

Oba, T., Katon, M., Kitazato, H., Koizumi, 1.,Omura, A., Sakai, T., and Takayama, T. (1991). Pa-leoenvironmental changes in the Japan Sea duringthe last 85,000 years. Paleoceanography 6,499-518.

Reid, J. L. (1979). On the contribution of the Mediter-ranean Sea outflow to the Norwegian-GreenlandSea. Deep-Sea Research 26A, 1199-1223.

Reid, J. L., Jr. (1973). "North Pacific Ocean Waters inWinter." The Johns Hopkins Oceanographic Stud-ies 5.

Rind, D., Peteet, D., Broecker, W., McIntyre, A., andRuddiman, W. (1986). The impact of cold North At-lantic sea surface temperatures on climate: Implica-tions for the Yormger Dryas cooling (l1-lOk). Cli-mate Dynamics 1, 3-33.

Riser, S. C. (1990). Sources of North Pacific Interme-diate Water from the Sea of Japan and Sea ofOkhotsk. EOS Transactions American GeophysicalUnion 71, 881.

360 KEIGWIN AND GORBARENKO

Shuto, K. (1982). A review of sea conditions in theJapan Sea. La Mer 20, 119-124.

Talley, L. D. (1991). An Okhotsk Sea water anomaly:implications for ventilation in the North Pacific.Deep-Sea Research 38, supp!. I, S17I-S190.

Toba, Y., Tomizawa, K., Kurasawa, Y., and Hanawa,K. (1982). Seasonal and year-to-year variability ofthe Tsushima-Tsugaru Warm Current system withits possible cause. La Mer 20, 41-51.

Tsukada, M. (1986). Altitudinal and latitudinal migra-tion of Cryptomeria japonica for the past 20,000years in Japan. Quaternary Research 26, 135-152.

Ujiie, H. (1973). Sedimentation of planktonic forami-niferal shells in the Tsushima and Korea Straits be-

tween Japan and Korea. Micropaleontology 19,444--460.

Ujiie, H., and Ichikura, M. (1973). Holocene to upper-most Pleistocene planktonic foraminifers in a pistoncore from off San'in District, Sea of Japan. Trans-actions Proceedings of the Palaeontology Society ofJapan 91, 137-150.

Vasiliev, A. S., and Makashin, V. P. (1991). Ventila-tion of the Japan Sea waters in winter. Meteorologyand Hydrology, no. 2, 71-79. [in Russian]

Yasui, M., Yasuoka, T., Tanioka, K., and Shiota, O.(1967). Oceanographic studies of the Japan Sea (1)-Water characteristics. The Oceanographical Maga-zine 19, 177-192.

Yi, S.-U. (1966). Seasonal and secular variations of thewater volume transport across the Korea Strait. TheJournal of Oceanological Society of Korea 1, 7-13.