lable at ScienceDirect

Quaternary International 236 (2011) 57e64

Contents lists avai

Quaternary International

journal homepage: www.elsevier .com/locate/quaint

Lake level variations of Qinghai Lake in northeastern Qinghai-Tibetan Plateausince 3.7 ka based on OSL dating

XiangJun Liu a,b, ZhongPing Lai a,*, David Madsen c, LuPeng Yu a,b, Kai Liu a,b, JingRan Zhang a,b

a Luminescence Dating Group, Key Laboratory of Salt Lake Resources and Chemistry, Qinghai Institute of Salt Lakes, Chinese Academy of Sciences, 18 Xinning Rd., Xining, Qinghai810008, ChinabGraduate School of Chinese Academy of Sciences, Beijing 100049, Chinac Texas Archeological Research Laboratory, University of Texas, 1 University Station R7500, Austin, TX 78712, USA

a r t i c l e i n f o

Article history:Available online 27 August 2010

* Corresponding author.E-mail addresses: zplai@isl.ac.cn, zplai@yahoo.com

1040-6182/$ e see front matter � 2010 Elsevier Ltd adoi:10.1016/j.quaint.2010.08.009

a b s t r a c t

Qinghai Lake is the largest internally drained lake in China and its unique location makes it sensitive toclimate changes. Late glacial climate changes associated with variation in Qinghai Lake levels have beenintensively investigated for the past 40 years, with particular attention paid to lake level fluctuationhistories between the last interglacial and the Holocene. However, the details of lake level fluctuationsduring the Holocene are still unclear. Using both optically stimulated luminescence (OSL) dating ofquartz (for 22 samples) and infra-red stimulated luminescence (IRSL) of feldspars (only for sample HYW1whose IRSL age is 37 � 15 years), a total of 23 samples are dated from paleoshoreline deposits, fluvialsediments and aeolian sands (a total of 13 sections) near the modern lake shore, with ages from 37 � 15to 3710 � 350 years. These ages are used to reconstruct the lake level fluctuation history spanning thelast 3700 years. The results indicate that: (1) the youngest IRSL age of 37 � 15 years is in agreement withthe independent age of 39e29 years, suggesting that luminescence dating is able to date decadal samplesfor sediments from the study area; (2) the lake experienced several oscillations imposed on an overallregressive trend during the past 3700 years; (3) the dated paleoshoreline deposits are generally relatedto warm and wet periods, suggesting that those shoreline deposits formed during cold and dry periods, ifany, may have been modified by later transgressions; (4) lake level fluctuations during the period of 3700e240 years ago are generally consistent with the climate conditions identified in other proxies, with thehighest lake level occurring about 1770 years ago; and (5) after 240 years ago the lake level droppedmore rapidly, which is inconsistent with the proxy records (showing a warmer and wetter phase).

� 2010 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction

Qinghai Lake lies in the northeastern part of the Qinghai-TibetPlateau (QTP), and is the largest saline lake in China. Its uniquegeographical location near the junction of three climate systems (theEast Asianmonsoon, the Indianmonsoon, and theWesterlies) makesit one of the most sensitive regions to climate change in the world(Fig. 1). Lake level changes spanningw130e20 ka have been investi-gated recently (Madsenet al., 2008; Liuet al., 2010;Rhodeet al., 2010),and Holocene lake level fluctuations have also been investigatedextensively (Yuan et al., 1990; Chen et al., 1990; Lister et al., 1991).

However, different views of lake level variation still exist (Zhanget al.,1988,1994; Chenet al.,1990,1991;Yuanet al.,1990; Listeret al.,

.cn (Z.P. Lai).

nd INQUA. All rights reserved.

1991; Wang and Shi, 1992; Rhode et al., 2010). Zhang et al. (1988)pointed out that there was no high lake level during the Holocene,and that the lake terraces around the lake were formed in thePleistocene. Yuan et al. (1990) reported that the highest lake levelduring the Holocene was 10e15 m above the lake level of 1990 AD(w3194.2 m above sea level [asl]). Chen et al. (1990) and Lister et al.(1991) concluded that the lake level was lower than 3208 m aslduring the Holocene, and that the highest Holocene lake leveloccurred between 7.4 ka an 6 ka. Thiswas based, in part, on a sectionof aeolian sediments containing several soils at 3207 m asl. Chenet al. (1990) and Lister et al. (1991) also reported that the lake leveldropped12.9m(from3206.6mto3193.7masl) between1884e1990AD. Lister et al. (1991) pointed out that a shoreline terrace atþ12mprobably represents the highest Holocene lake level dating to about7e6 ka. Chen et al. (1991) investigated loess deposits and theunderlying lake deposits at the southernmargin of Qinghai Lake andconcluded that the lake level elevations fluctuated within a 30 m

Fig. 1. (a) Location of Qinghai Lake, and (b) satellite image of Qinghai Lake. In (b), the rectangles marked A and B represent east Haiyanwan and Erlangjian, where OSL samples werecollected. Their enlarged images are shown in Figs. 2 and 3, respectively.

XiangJun Liu et al. / Quaternary International 236 (2011) 57e6458

range during the Holocene, and within 20 m during the past 4 ka.Using radiocarbon dating, Wang and Shi (1992) dated organicmaterials or charcoal from four lake terraces at different elevationsaround the southmargin of the lake, and concluded that (1) QinghaiLake had very high lake levels during the Holocene, and a lake leveldatingw6 kawas about 60m above the lake level in 1992 and about27 m higher at w4.5 ka, and (2) the lake level has dropped w10 mduring the last 1 ka. The chronologies for these studies were allestablished by radiocarbon dating. Zhang et al. (1994) reconstructedlake level fluctuations during the Holocene based on the linearrelationshipbetween the ratioof Sr/Ca and lake salinity. Their resultsindicate that the lake level dropped 17.5mduring theHolocene, andthat the lake level at w3 ka was 13 m higher than the lake levelelevation in 1994 (3193.7 m asl). However, Colman et al. (2007)pointed out that salinity reconstructions, based on the traceelement geochemistry of biogenic carbonate, is extremely prob-lematic for Qinghai Lake due to aragonite overgrowth and possiblediagenetic alteration, and that the relationships between Sr/Ca andlake salinity is complicated. Yu (2005) investigated the calcareoussediments within two drill cores from Qinghai Lake (Q14B andQ16C), and concluded that the lake level varied between neardesiccation and a depth of 30 m in the past 14 ka. That is, themaximum Holocene depth was only 4.5 m more than the depth of25.5 m in 2008 AD. Henderson and Holmes (2009) reviewed thestate of knowledge for the last millennium in Qinghai Lake, andconcluded that a detailed picture of climate change cannot beestablishedwhenusing radiocarbondatesdue topoor chronologicalconstraints caused by a hard-water reservoir effect. However, esti-mates of lake level changes during theHolocene in virtually all thesestudies were either obtained indirectly from proxy records or werebased on radiocarbon dating of organic samples collected fromsediment layers below the paleoshoreline deposits.

As remnant paleoshoreline sediments represent past periods ofstable lake levels, the age of their formation and their three-dimen-sional coordinates can be used to retrieve several past hydrologicalparameters, such as lake level elevation, lake area and lake watervolume, that cannot be obtained fromother proxy records. Due to thelack of suitable organicmaterials inmost beach sediments, the directdating of paleoshorelines by radiocarbon dating is difficult. However,OSL dating is particularly appropriate for aeolian deposits (Lai et al.,2007a, 2009; Long et al., in press) and shoreline deposits (Madsenet al., 2008; Rhode et al., 2010; Liu et al., 2010; Sun et al., 2010;Zhang et al., submitted for publication) that are abundant aroundQinghai Lake. Madsen and Murray (2009) reviewed recent applica-tions of theOSL dating to young sediments, and concluded thatOSL isan accurate and reliable tool for determining the timeof deposition ofyoungwater-laid sediments from coastal zones, and aeolian depositsfrom both coastal and inland environments.

Most of the easily identified low elevation paleoshorelinesaround Qinghai Lake were formed during the late Holocene (Liet al., 1995). Older paleoshorelines have mostly been eitherdestroyed by later erosion or covered by later sediments and cannotbe readily identified. The purpose of this paper was to date the lowelevation paleoshorelines along Qinghai Lake with remainingsignificant landforms, but which have limited chronologicalcontrols. OSL dating was used to: (1) explore if OSL dating could beused to date young beach deposits in this arid plateau environ-ment; and (2) determine the lake level fluctuation history duringthe late Holocene by directly dating paleoshoreline deposits.

2. Study area

Qinghai Lake lies on the northeastern QTP (36�320e37�150 N;99�360e100�460 E), has a lake surface area of 4473 km2 and a watervolume of 850 � 108 m3 (Fig. 1, Ma, 1998). The lake level of QinghaiLake was 3193.4 m asl on March, 2010. Its catchment is prismatic inshape, with a westeeast length of about 106 km, a northesouthwidth of about 63 km and a perimeter of about 360 km. Observa-tional data from the Gangcha meteorological station on thenorthwestern margin of the lake (Fig. 1b) indicate that the annualmean temperature is �0.3 �C, the highest monthly mean temper-ature is 10.9 �C (July), and the lowest monthly mean temperature is�13.5 �C (January) (Ma, 1998). The annual mean precipitation is300e400 mm. The annual mean evaporation in the Qinghai Lake isabout 1300e2000mm. Themain tributaries of Qinghai Lake are theBuha River, which contributes about half of the annual water inputto Qinghai Lake (Li et al., 2007), the Shaliu River, the Haergai River,the Quanji River and the Heima River.

3. Methods

3.1. OSL sample collection

Although paleoshorelines provide direct evidence of past lakelevel elevations, it is often difficult to obtain organic materials for14C dating; and the particle sizes of their sediments are often toocoarse for OSL dating. After extensive field investigations, an areaaround Haiyanwan on the northern shore of the lake was selectedto collect OSL samples because of the presence of a number of wellpreserved and easily recognized paleoshorelines with limitedvegetation cover (Fig. 1b and Fig. 2). Another site with distinctshoreline features at Erlangjianwas also sampled (Figs.1b and 3). Atboth localities the particle size of the sediments is appropriate forOSL dating. The elevations of shorelines were surveyed usinga Differential Global Positioning System (DGPS) with elevationsaccurate to within 30 mm.

Fig. 2. The enlarged region of east Haiyanwan (A in Fig. 1b). The full circles are the siteswhere OSL samples were collected: HYW1, HYW2, HYW3, HYW4, HYW5, HYW6,HYW7, HYW8, HYW9, HYW10, HYW11 and HYW12. White lines are the 3200 m and3210 m asl contours.

XiangJun Liu et al. / Quaternary International 236 (2011) 57e64 59

OSL sample HYW1 was collected from beach sediments at3195 m asl (Figs. 2 and 4). Analysis of remote sensing images takenat different periods indicates that this shoreline lies between the1970 and 2000 AD lake levels. This suggests that sediments ofsample HYW1 were deposited after 1970 and prior to 2000 AD.Observed lake level elevation records show that this shoreline was

Fig. 3. The enlarged region of Erlangjian (B in Fig. 1b). The full circle is the site whereOSL samples were collected. White dashed lines are paleoshorelines.

formed between 1970 and 1980 AD (lake level elevations at 1970and 1980 ADwere 3195.6m and 3194.1m asl). As a result, the age ofthis sample should be between 39 and 29 years old relative to theyear 2009, when the OSL samples were analyzed. This sample wasused to test the OSL dating of the youngest sediments in this area.

The top 0.1 m of the HYW2 section is aeolian sand, with beachgravels continuing downwards to an unknown depth (Fig. 4). SampleHYW2was collected from a ripple laminated sand layer of near shoredepositswith a thickness of 0.1m. The elevation of HYW2 is 3197.4masl. The shoreline deposits exposed in HYW3 section are just 0.2 mthickandarecomposedof small gravelsandverycoarsesands thataretoo coarse for OSL dating. Fortunately, below the shoreline depositsare ripple laminated near shore sands deposited when the lakemargin was not far from this site (Fig. 4). Samples HYW3-A andHYW3-B were collected from the near shore sands at elevations of3198.2 m and 3198 m, respectively. Sample HYW4 was taken froma thin layer of near shore sands that interfacewith shoreline depositsat 3198.4 m asl. Samples HYW5 and HYW6were collected from twoshorelines in which some thin sand lenses are interbedded withcoarser shoreline deposits (Fig. 4). At the top of the HYW7 sectiona 0.2 m thick unit of aeolian sand containing small gravel overliescoarse sand and gravel shoreline deposits of unknown depth (Fig. 4).Sample HYW7-A was collected from the lower part of the aeoliansands, and HYW7-B was collected from a 0.1 m thick fine sand lenswithin the shoreline deposits at an elevation of 3200.1 m asl.

SectionHYW8has a similar stratigraphic context to the sequenceat HYW7 (Fig. 4). One OSL sample was taken from a fine sand layerwithin the coarser shoreline deposits at an elevation of 3201.5m asl.The HYW9 section is located in an excavated gravel pit exposing1.3 m thick coarse shoreline deposits containing two horizontallyorientedmedium sand layers (Fig. 4). Samples HWY9-A and HYW9-Bwerecollected fromthose twosand layers at elevationsof 3201.6mand 3201.3 m asl, respectively. The sections at HYW10, HYW11 andHYW12 have a similar stratigraphic context in that they all containthree distinct layers from top to bottom. The top unit is composed ofaeolian sands with small gravel inclusions. This overlies thinshoreline deposits that, in turn, overlie fluvial stream deposits ofsilty-clay sediments without evident bedding (Fig. 4). The shorelinedeposits lie unconformably above the underlying fluvial sediments.Samples HYW10-A and HYW10-B were collected from the lowerpartof theaeolian sandsanda ripple laminated sand layerwithin theshoreline deposits at elevations of 3201.7 m and 3201.4 m asl,respectively. Samples HYW11-A and HYW11-B were collected fromthe middle and lower portions of the upper aeolian sands, whilesample HYW11-C was taken from silt deposits 0.25 m below theoverlying shoreline deposits. The shoreline deposits at HYW11werenot sampled because the particles are too coarse. SamplesHYW12-Aand HYW12-B were collected from the upper aeolian sands, sampleHYW12-C was taken from fine sands located within the shorelinedeposits, and sample HYW12-Dwas collected from the silt deposits0.15 m below the overlying shoreline deposits.

Several low Holocene shorelines in Erlangjian (ERLJ) can bereadily identified in both field surveys and from Google Earthimages (Fig. 3). As this region is protected for tourism, only threeOSL samples could be collected from a garbage pit along oneshoreline at 3201.7 m asl. The top 0.3 m of ERLJ section is aeoliansand with fine gravel. This overlies a 0.35 m thick shoreline depositthat, in turn, overlies fine sands that continue downwards to anunknown depth (Fig. 4). Samples ERLJ-A, ERLJ-B were collectedfrom aeolian sands and shoreline deposits, respectively.

3.2. OSL dating

The samples for OSL dating were collected in 25 cm long, 5 cmdiameter, iron tubes. In the laboratory, the sediments at each end of

Fig. 4. Graphic sedimentary logs showing the locations of OSL samples and the corresponding OSL ages.

XiangJun Liu et al. / Quaternary International 236 (2011) 57e6460

thecylinderswerescrapedandused forwatercontentmeasurements.The sediment samples used to determine dose rates were collectedfrom within 30 cm in of the OSL sample locations. The light-unex-posed sediment from the middle part of the tube was used for OSLsample preparation. Raw samples were first treated with 38% H2O2and 10%HCl to remove organicmaterials and carbonates. Particle sizebetween38and63mmwere selected bydry sieving.Different particlesizes (38e63, 88e125, and 125e200 mm, see Table 1)were chosen fordating for different samples according to the particle size distributionof a sample. The 38e63 mm particle fractions were treated with 35%H2SiF6 acid for about two weeks to remove feldspars. Quartz grainswere then washed with 10% HCl and rinsed with water. For thesamples without enough 38e63 mm particle fractions, the coarsergrains (88e125 or 125e200 mm) were used; and they were treatedwith 40% HF for 45min to remove feldspars and the alpha-irradiatedouter layer (w10 mm). After infra-red light checking, samples thatshowed obvious IRSL signals were retreated with H2SiF6 or HF, and

checked with IRSL again to avoid age underestimation (Lai andBrückner, 2008). The pure quartz grains were then mounted in thecenter part (diameter of w0.5 mm) of stainless steel discs (10 mmdiameter) using silicone oil.

For the very recent sample HYW1, the quartz OSL signal was tooweak to be separated from the background noise. Consequently,IRSL of feldspars was used for De determination. For all the othersamples, their initial quartz OSL signal of natural aliquots wassufficiently high for De measurements.

OSLmeasurementswere carried out in the Luminescence DatingLaboratory of theQinghai Institute of Salt Lakes, Chinese AcademyofSciences, using an automated Risø TL/OSL DA-20 reader. The OSLsignal was detected through a 7.5 mmHoya U-340 filter. Laboratoryirradiation used a 90Sr/90Y beta source. OSL stimulationwas carriedout for 40 s at 130 �C. Signals of the initial 0.64 s of stimulationwereintegrated for growth curve construction after backgroundsubtraction. The stimulation used blue diodes (l¼ 470� 20 nm) for

Table 1Environmental radiation measurements and OSL ages. Sample HYW1 was dated using feldspars IRSL and all the other samples were dated using quartz OSL. The elevationrefers to that of the sample location. For each sample, 24 aliquots were measured for De determination.

Sample ID Elevation(m asl)

Grain Size(mm)

K (%) Th (ppm) U (ppm) WaterContent (%)

Dose Rate(Gy/ka)

De (Gy) OSL Age (a)

HYW1 3195 38e63 1.25 � 0.04 4.19 � 0.13 2.63 � 0.13 10 � 5 2.45 � 0.17 0.09 � 0.03 37 � 15HYW2 3197.4 125e200 1.73 � 0.09 4.49 � 0.13 1.39 � 0.11 10 � 5 2.40 � 0.20 0.34 � 0.03 140 � 20HYW3-A 3198.2 38e63 1.16 � 0.04 3.68 � 0.99 1.33 � 0.09 10 � 5 1.99 � 0.16 0.48 � 0.05 240 � 30HYW3-B 3198 38e63 1.26 � 0.04 3.45 � 0.11 1.27 � 0.09 10 � 5 2.03 � 0.14 0.91 � 0.04 450 � 40HYW4 3198.4 38e63 1.22 � 0.05 1.48 � 0.16 1.44 � 0.09 10 � 5 1.91 � 0.14 0.55 � 0.08 290 � 50HYW5 3198.6 38e63 1.05 � 0.04 3.44 � 0.09 1.63 � 0.09 10 � 5 1.94 � 0.13 0.63 � 0.08 320 � 50HYW6 3200.4 38e63 1.25 � 0.04 5.19 � 0.13 3.03 � 0.14 10 � 5 2.62 � 0.17 2.29 � 0.23 870 � 110HYW7-A 3200.5 38e63 1.73 � 0.09 5.34 � 0.15 1.23 � 0.12 5 � 5 2.75 � 0.21 0.42 � 0.02 150 � 10HYW7-B 3200.1 38e63 1.49 � 0.08 5.29 � 0.14 1.30 � 0.11 10 � 5 2.39 � 0.18 2.16 � 0.05 900 � 70HYW8 3201.5 38e63 1.44 � 0.19 2.96 � 0.18 1.18 � 0.05 10 � 5 2.14 � 0.23 4.6 � 0.59 2150 � 360HYW9-A 3201.6 88e125 1.50 � 0.09 3.97 � 0.12 1.37 � 0.11 10 � 5 2.19 � 0.17 2.34 � 0.24 1070 � 140HYW9-B 3201.3 88e125 1.49 � 0.09 4.47 � 0.13 1.09 � 0.10 10 � 5 2.13 � 0.17 5.04 � 0.63 2370 � 350HYW10-A 3201.7 38e63 1.55 � 0.06 5.32 � 0.23 1.51 � 0.19 5 � 5 2.66 � 0.20 2.20 � 0.46 830 � 180HYW10-B 3201.4 88e125 1.34 � 0.05 4.51 � 0.13 1.52 � 0.21 10 � 5 2.12 � 0.16 2.30 � 0.21 1090 � 130HYW11-A 3202.7 38e63 1.30 � 0.06 4.40 � 0.22 1.17 � 0.19 5 � 5 2.25 � 0.17 2.2 � 0.34 980 � 170HYW11-B 3202.4 38e63 1.32 � 0.06 6.46 � 0.20 1.89 � 0.19 5 � 5 2.62 � 0.18 4.20 � 0.77 1600 � 310HYW11-C 3201.9 38e63 1.70 � 0.05 8.50 � 0.21 1.77 � 0.15 15 � 5 2.74 � 0.19 10.17 � 0.65 3710 � 350HYW12-A 3202.9 88e125 0.96 � 0.04 3.09 � 0.18 1.07 � 0.16 5 � 5 1.68 � 0.12 1.47 � 0.24 880 � 160HYW12-B 3202.6 88e125 1.16 � 0.06 4.42 � 0.22 1.19 � 0.21 5 � 5 1.96 � 0.15 2.37 � 0.37 1210 � 210HYW12-C 3202.3 88e125 1.39 � 0.06 3.78 � 0.19 1.29 � 0.17 10 � 5 2.02 � 0.15 3.59 � 0.35 1770 � 220HYW12-D 3202.1 38e63 1.74 � 0.07 11.8 � 0.44 2.76 � 0.27 15 � 5 3.20 � 0.23 7.80 � 0.55 2440 � 240ELJ-A 3201.5 38e63 2.23 � 0.11 7.26 � 0.19 1.28 � 0.13 5 � 5 3.39 � 0.27 0.91 � 0.11 270 � 40ELJ-B 3201.2 38e63 1.44 � 0.08 8.19 � 0.21 1.93 � 0.15 10 � 5 2.70 � 0.19 6.40 � 0.87 2370 � 360

XiangJun Liu et al. / Quaternary International 236 (2011) 57e64 61

quartzOSL, and infra-reddiodes (l¼830�10nm) for feldspar infra-red stimulated luminescence (IRSL). A a value of 0.035 � 0.003 forquartz (Lai et al., 2007b), and 0.1 � 0.01 for feldspars was used. Thesingle aliquot regenerative-dose (SAR) protocol was used for Dedetermination (Murray and Wintle, 2000).

Lithogenic radionuclide activity concentrations were deter-mined from measurements of U, Th and K concentrations usingneutron activation analysis (NAA) of dried and ground bulk

b

0

50

100

150

200

250

0 2 4 6 8 10

OSL

inte

nsity

(c/0

.16s

)

a

c d

HYW1 (IRSL)

N

0 Gy

0

100

200

300

400

0 2 4 6 8 10Stimulation Time (s)

)s61.0/c(ytisnetni

LSO

HYW2 (OSL)

N

0 Gy

Fig. 5. OSL characteristics for samples HYW1 (feldspar IRSL) and HYW2 (quartz OSL). Dosenatural (red line with open square) and zero Gy regeneration dose (black line with open tr

samples. The NAA data were measured in the Chinese AtomicEnergy Institute in Beijing. Based on measured modern shorelinedeposits and the water content of the collected samples, a long-termwater content of 5� 5%, 10� 5% and 15� 5% has been used toadjust a dry dose rate for aeolian deposits, paleoshoreline depositsand water-laid fine sand and silt. The cosmic ray dose was esti-mated for each sample as a function of depth, altitude andgeomagnetic latitude (Prescott and Hutton, 1994).

0

0.04

0.08

0.12

0.16

0.2

0 0.1 0.2 0.3 0.4 0.5

)xT/xL(LS

Odetcerro

C

HYW1

0.0

0.1

0.2

0.3

0.4

0 1 2Laboratory Dose (Gy)

)xT/xL(LS

Odetcerro

C

HYW2

response curves are shown in (b) and (d). OSL decay curves shown in (a) and (c) foriangle) of HYW1 and HYW2.

XiangJun Liu et al. / Quaternary International 236 (2011) 57e6462

It is important to investigate the influence of preheating on thecharge transfer from light-insensitive traps to light-sensitive ones,a process called thermal transfer (Aitken, 1998). A plateau test forsample HYW3-A showed that De is independent of preheating (for10 s) temperatures from200 �C to 260 �C (20 �C intervals with a cut-heatof 220 �C for 10 s for testdoses). ForHYW1,220 �Cwas chosenasthepreheating temperature; forother samples, 260 �Cwas chosenasthe preheating temperature. A dose recovery test was performed onsample HYW3-A. Ten natural aliquots were stimulated twice byblue-light stimulationat130 �C for40 s. Thealiquotswere thengivena laboratory dose of 0.49Gy, approximate to the naturalDe (0.48Gy).The measured De was 0.47� 0.04 Gy. Thus, the ratio of measured tothe given De was 0.96 � 0.08, suggesting that the SAR conditionswere appropriate for De determination.

For De determination, twenty-four aliquots were measured foreach sample to calculate afinalDe. Fig. 5A and C showgrowth curvesfor samples HYW1 and HYW2. The growth curve for these twosampleswasfittedby linearequation. Fig. 5a shows IRSL shine-downcurves for sample HYW1, and 5c shows quartz OSL shine-downcurves for sample HYW2. The zero dose in Fig. 5a and c gave an OSLsignal similar to the background, implying negligible recuperationcontribution to theDe. The natural decay curves of HYW1andHYW2show that the signals are strong enough to be separated frombackground noises.

4. Dating results

The values for equivalent doses, dose rates and OSL ages of allsamples are presented in Table 1. The OSL age determinations for 23

Fig. 6. The Qinghai Lake shoreline history for the past 3700 years based on OSL dating, in cocores (Ji et al., 2005). (b and c) Salinity and temperature proxies (Liu et al., 2006) of Qinghaages. The red empty diamonds and green empty triangles in (d) represent fluvial and near shthe black empty circles represent aeolian sands.

samples analyzed from the 13 sites and the elevations of samplingsites are shown in Fig. 4. The OSL ages of the paleoshorelinedeposits was within a range of 37 � 15 to 2370 � 350 years, withthe error associated with individual ages ranging from 8.8% to 17.2%(except HYW1 where the error reached 40.5%). The age of sampleHYW1, collected from the youngest shoreline deposits, is 37 � 15years (Table 1 and Fig. 4). This age is in agreement with the inde-pendently determined age of 39e29 years based on the analyses oftopographic maps and remote sensing data. This shows that theshoreline deposits were well-bleached prior to deposition, and thatrecuperation is negligible for this sample. All OSL ages of paleo-shoreline deposits are in stratigraphic order (Fig. 4). This suggeststhat OSL dating is a reliablemethod for determining a chronology oflate Holocene paleoshoreline deposits at Qinghai Lake.

Fig. 4 shows the stratigraphic profile for all sampled sections,sampling sites and their elevations. HYW4 and HYW5, HYW6 andHYW7-B, HYW9-A and HYW10-B, ERLJ-B and HYW9-B, constitutefour groups of samples collected from shoreline deposits that arevery close in elevation (within 3 cm) within each group. The ages ofthe samples within each group are essentially the same within therange of errors (Fig. 4 and Table 1). Hence, each of these four groupsimply a period of lake level stability. Weighted averages of the OSLdates within each group were used to obtain the ages of paleo-shoreline deposits with corresponding elevations. HYW11 andHYW12 are on the highest paleoshorelines (Fig. 2), and they havesimilar stratigraphic compositions. The OSL ages of the aeoliansands and fluvial silts bracketing the shoreline deposits at thesetwo sections are consistent with each other. These indicate theshoreline sediments in these two sections were deposited about

mparison with other climate proxies. (a) Redness curve of lacustrine sediments in lakei Lake. (d) A reconstructed lake level change curve based on OSL dating results. (e) OSLore sands, respectively, the black empty squares represent paleoshoreline deposits, and

XiangJun Liu et al. / Quaternary International 236 (2011) 57e64 63

1770 years ago, when the lake level was at 3202.3 m asl. This is8.9 m above the modern lake elevation of 3193.4 in 2010 AD, andwas the highest lake level during the past 3700 years. SamplesHYW3-A and HYW3-B were collected from near shore sands belowthe shoreline deposits. Although the ages of these two samples donot exactly define the age of the overlying paleoshoreline, they doshow the lake level was higher than 3198.2 m and 3198 m asl whentheywere deposited. The ages on the surface aeolian sands at all thesections are indicative of times that lake levels were lower than theelevations of the aeolian deposits. The ages on the fluvial siltsimmediately underlying the shoreline deposits also represent timeswhen lake levels were lower than these elevations. Samplescollected directly from shoreline deposits can exactly define lakelevel changes. The dated aeolian sands, fluvial silts, and processedpaleoshoreline deposit ages were used to construct a lake levelhistory for the past 3700 years (Fig. 6d).

5. Discussion

Holocene climate changes reflected in Qinghai Lake proxyrecords have been investigated bymany researchers during the late20th and early 21st centuries (Chen et al.,1990; Lister et al.,1991; Liuet al., 2003, 2006, 2007; Shen et al., 2005; Ji et al., 2005, 2009; Yu,2005, 2008). Some of these investigators have proposed that Qing-hai Lake experienced a warm and wet climate during the early andmiddleHolocene, but afterw4500 cal years BP the climate graduallybecame colder and drier (Chen et al., 1990; Lister et al., 1991; Shenet al., 2005). Yu (2005) concluded that at about 8 and 4 ka 14C BPthe climateofQinghai Lakebecamewarmerandwetter,with climateconditions following 4 ka 14C BP remaining similar to today. Otherinvestigators have found the climate of Qinghai Lake underwenta transition from dry to wet during the late Holocene (Ji et al., 2005;Liu et al., 2006, 2008). Given these different views of climateconditions at Qinghai Lake during the late Holocene, it is obviousthat the lake level history during the late Holocene is not yet clear.This study provides preliminary results for the late Holoceneshoreline history of the lake, and also provides a check on inter-pretations of late Holocene climate changes based on other proxies.

Fig. 6 shows the Qinghai Lake shoreline history for the past 3700years based on OSL dating of shoreline sediments, together witha redness curve of lacustrine sediments in a lake core (Ji et al., 2005)and salinity and temperature proxies (Liu et al., 2006) of QinghaiLake. The redness of the lake sediments is primarily related to ironoxide mineral concentrations brought to the lake by rivers, and isan indicator of dry/wet conditions in the lake area (Ji et al., 2005).High redness values indicate wetter conditions. U37

k0 and %C37:4 ofthe alkenone proxies of lacustrine sediments reflect the tempera-ture and salinity changes of Qinghai Lake (Liu et al., 2006). Highvalues imply higher temperatures and lower salinity.

Fig. 6d shows that the lake level declined 5e6 m since 2370years ago, but the lake experienced several periods of transgressionand regression within the overall regression. The lake level was ata low stand between 3710 and 2370 years ago, as inferred from twostream fluvial sediments at 3201.9 m and 3202.1 m, and confirmedby proxy records from Ji et al. (2005) and Liu et al. (2006). Ji et al.(2005) reported that a two-millennium long dry period occurredin Qinghai Lake from w4200 through w2300 years ago, as can beseen in Fig. 6a. Liu et al. (2006, 2008) proposed that the climate inQinghai Lakewas relatively dry beforew2000 years BP, as shown inFig. 6b and c. The lake began to rise afterw2400 years ago, reaching3201.3 m asl at 2370 years ago, and its highest point (3202.3 m asl)at 1770 years ago. This lake level interval may correspond to theRomanWarm Period (RWP,150 BC-270 AD), as has been reported inresearch from Qinghai Lake (Ji et al., 2005; Liu et al., 2006, 2008),and western China (Yang et al., 2004). Fig. 6a,b and c all show that

during the period between w1500 and 2100 years ago, the areaaround Qinghai Lake experienced a warm and wet climate. Theseveral hundred years offset that is evident on Fig. 6a, b and c maybe due to the uncertainties associatedwith the dating. From 1770 to320 years ago, lake levels declined gradually, from 3202.3 m asl to3198.6 m asl. The three core-based proxies (Fig. 6a, b and c) implya cold dry stage during the Dark Ages Cold Dry Period (DACDP,900e1400 AD). However, paleoshoreline deposits formed duringthis period have not yet been identified, suggesting shorelinesformed during this time interval may have been destroyed by a lakelevel rise at the onset of the following Medieval Warm Period(MWP, 1100e600 AD).

During the MWP, lake levels gradually declined, in contrast tothe other proxy records (Fig. 6a, b and c). However, these proxyrecords indicate numerous fluctuations during the MWP. Liu et al.(2008) also reported that the moisture source for Qinghai Lakemay have changed at about 1500e1250 years ago. Changed mois-ture sources might mean that climate conditions during the MWPwere different than during the RWP, and that effective moistureduring the MWP was not as high as during the RWP, causing thelake level to continue to decline during the MWP.

During the Little Ice Age (LIA, 1430e1850 AD), the lake leveldeclined continuously, in agreement with the proxy recordsreflecting a cold, dry period. However, about 240 years ago, the lakelevel began to drop more rapidly, which is inconsistent with theproxy records shown in Fig. 6(a, b and c). Liu et al. (2008) proposedthat the isotopic enrichment in the Qinghai Lake region mightreflect the combined effects of warmer temperatures and lessmonsoonal moisture during the past 100 years, which would havecaused the lake to decline rapidly after about 1870 AD. Li et al.(2007) also concluded that the lake level decline during the past50 years was mainly caused by climate factors, and that the humaninfluence on water depletion is negligible. The reason for theinconsistency between this rapid lake level drop after 240 BP andproxy records showing a warm and wetter phase is not known yet.

6. Conclusion

Changes in the levels of Qinghai Lake during the late Holocenewere investigated using OSL dating of 23 samples from paleo-shoreline deposits, fluvial sediments and aeolian sands near themodern lake shore. Twenty-two samples were dated using quartzOSL and one sample using IRSL due to the low signal level of quartzOSL. The IRSL age of 37 � 15 years ago for young shoreline deposits(sample HYW1) is in agreement with an independent age of 39e29years, which was based on analyses of topographic maps andremote sensing data. This shows that the shoreline deposits werewell-bleached prior to deposition, and that IRSL recuperation wasnegligible for this sample. Quartz OSL recuperation was alsonegligible for all the other samples (see Fig. 5). All OSL ages ofpaleoshoreline sediments are in stratigraphic order (Fig. 4). Asa result, OSL dating can be used to date the young shorelinedeposits in Qinghai Lake.

A reconstructed lake level change curve indicates that the lakeexperienced several periods of transgression and regression overthe past 3700 years. Dated paleoshoreline deposits almost allcorrespond to warm and wet periods, indicating that shorelinedeposits formed during cold and dry periods were possibly alteredor destroyed by following transgressive lake level phases.

Lake level fluctuations during the past 3700 years recorded inthe shoreline history are generally consistent with the climateconditions retrieved from other proxies. The highest lake level forthe past 3700 years was at 1770 years ago at an elevation of3202.3 m asl, which is 8.9 m above the modern lake level (at3193.4 m asl on March, 2010). Lake levels were quite stable during

XiangJun Liu et al. / Quaternary International 236 (2011) 57e6464

the RomanWarm Period, but declined steadily during the MedievalWarm Period. This decline was possibly caused by a change inmoisture source (Liu et al., 2008).

After 240 years ago, the lake level dropped more rapidly, whichmay be due to the combined effects of warmer temperature andless monsoonal moisture (Liu et al., 2008). This rapid drop is not inagreement with the proxy records shown in Fig. 6 (curves a, b,and c) and this requires further study.

Acknowledgements

This study was financially supported by a One-Hundred TalentProject of CAS granted to ZPL and China NSF grants (40872119,40761010). We thank JunFeng Ji for providing the redness data, andZhongHui Liu for the temperature and salinity reconstruction data.Thanks are also given to Lewis Owen, Steffen Mischke and ananonymous reviewer for their constructive suggestions andlanguage corrections.

References

Aitken, M.J., 1998. An Introduction to Optical Dating [M]. Oxford University Press,Oxford.

Chen, F.H., Wang, S.L., Zhang, W.X., Pan, B.T., 1991. The loess profile at south bank,climatic information and lake-level fluctuations of Qinghai Lake during the Holo-cene. Scientia Geographica Sinica 11, 76e85 (in Chinese with English abstract).

Chen, K.Z., Bowler, J.M., Kelts, K., 1990. Palaeoclimatic evolution within the Qinghai-Xizang (Tibet) plateau in the last 40,000 years. Quaternary Sciences 1, 21e32 (inChinese with English abstract).

Colman, S.M., Yu, S.Y., An, Z.S., Shen, J., Henderson, A.C.G., 2007. Late Cenozoicclimate changes in China’s western interior: a review of research on LakeQinghai and comparison with other records. Quaternary Science Reviews 26,2281e2300.

Henderson, A.C.G., Holmes, J.A., 2009. Palaeolimnological evidence for environ-mental change over the past millennium from Lake Qinghai sediments:a review and future research perspective. Quaternary International 194,134e147.

Ji, J.F., Shen, J., Balsam, W., Chen, J., Liu, L.W., Liu, X.Q., 2005. Asian monsoonoscillations in the northeastern Qinghai-Tibet Plateau since the late glacial asinterpreted from visible reflectance of Qinghai Lake sediments. Earth andPlanetary Science Letters 233, 61e70.

Ji, J.F., Balsam, W., Shen, J., Wang, M., Wang, H.T., Chen, J., 2009. Centennialblooming of anoxygenic phototrophic bacteria in Qinghai Lake linked to solarand monsoon activities during the last 18000 years. Quaternary ScienceReviews 28, 1304e1308.

Lai, Z.P., Wintle, A.G., Thomas, D.S.G., 2007a. Rates of dust deposition between 50 kaand 20 ka revealed by OSL dating at Yuanbao on the Chinese Loess Plateau.Palaeogeography, Palaeoclimatology, Palaeoecology 248, 431e439.

Lai, Z.P., Zoller, L., Fuchs, M., Bruckner, H., 2007b. Alpha efficiency determination forOSL of quartz extracted from Chinese loess. Radiation Measurements 43,767e770.

Lai, Z.P., Brückner, H., 2008. Effects of feldspar contamination on equivalent doseand the shape of growth curve for OSL of silt-sized quartz extracted fromChinese loess. Geochronometria 30, 49e53.

Lai, Z.P., Kaiser, K., Brückner, H., 2009. Luminescence-dated aeolian deposits of lateQuaternary age in the southern Tibetan Plateau and their implications forlandscape history. Quaternary Research 72, 421e430.

Li, X.Y., Xu, H.Y., Sun, Y.L., Zhang, D.S., Yang, Z.P., 2007. Lake-level change and waterbalance analysis at Lake Qinghai, West China during recent decades. WaterResource and Management 21, 1505e1516.

Li, Y.C., Zhang, L.Y., Zhou, S.Z., 1995. Study evolution Of Holocene lake level on <katime scale in Qinghai Lake. Journal of Qinghai Normal University (NaturalScience) 3, 39e45 (In Chinese with English abstract).

Lister, G.S., Kelts, K., Chen, K.Z., Yu, J.Q., Niessen, F., 1991. Lake Qinghai, China:closed-basin lake levels and the oxygen isotope record for Ostracoda since thelatest Pleistocene. Palaeogeography, Palaeoclimatology, Palaeoecology 84,141e162.

Liu, X.J., Lai, Z.P., Fan, Q.S., Long, H., Sun, Y.J., 2010. Timing for high lake levels ofQinghai Lake in the Qinghai-Tibetan Plateau since the last interglaciation basedon quartz OSL dating. Quaternary Geochronology 5, 218e222.

Liu, X.Q., Shen, J., Wang, S.M., Zhang, E.L., Cai, Y.F., 2003. A 16000-year paleoclimaticrecord derived from authigenetic carbonate of lacustrine sediment in QinghaiLake. Geological Journal of China Universities 9, 38e46 (in Chinese with Englishabstract).

Liu, X.Q., Shen, J., Wang, S.M., Wang, Y.B., Liu, W.G., 2007. Southwest monsoonchanges indicated by oxygen isotope of ostracode shells from sediments inQinghai Lake since the late Glacial. Chinese Science Bulletin 52, 539e544.

Liu, Z.H., Henderson, A.C.G., Huang, Y.S., 2006. Alkenone-based reconstruction oflate Holocene surface temperature and salinity changes in Lake Qinghai, China.Geophysical Research Letters 33, L09707.

Liu, Z.H., Henderson, A.C.G., Huang, Y.S., 2008. Regional moisture source changesinferred from late Holocene stable isotope records. Advances in AtmosphericSciences 25, 1021e1028.

Long, H., Lai, Z.P., Fuchs, M., Zhang, J.R., Yang, L.H. Palaeodunes intercalated in loessstratum from the western Chinese Loess Plateau: timing and palaeoclimaticimplications. Quaternary International, Submitted for publication.

Ma, W.L., 1998. The History of Qinghai Province and Qinghai Lake. Qinghai People’sPress, Xining (in Chinese).

Madsen, D.B., Ma, H.Z., Rhode, D., Brantingham, P.J., Forman, S.L., 2008. Ageconstraints on the late Quaternary evolution of Qinghai Lake, Tibetan plateau.Quaternary Research 69, 316e325.

Madsen, A.T., Murray, A.S., 2009. Optically stimulated luminescence dating of youngsediments: a review. Geomorphology 109, 3e16.

Murray, A.S., Wintle, A.G., 2000. Luminescence dating of quartz using an improvedsingle-aliquot regenerative-dose protocol. Radiation Measurements 32, 57e73.

Prescott, J.R., Hutton, J.T., 1994. Cosmic ray contribution to dose rates for lumines-cence and ESR dating: large depths and long-term time variations. RadiationMeasurements 23, 497e500.

Rhode, D., Ma, H.Z., Madsen, D.B., Brantingham, P.J., Forman, S.L., Olsen, J.W., 2010.Paleoenvironmental and archaeological investigation at Qinghai Lake, westernChina: geomorphic and chronometric evidence of lake level history. QuaternaryInternational 218, 29e44.

Shen, J., Liu, X.Q., Wang, S.M., Matsumoto, R., 2005. Palaeoclimatic changes in theQinghai Lake area during the last 18,000 years. Quaternary International 136,131e140.

Sun, Y.J., Lai, Z.P., Long, H., Liu, X.J., Fan, Q.S., 2010. Quartz OSL dating of archaeo-logical sites in Xiao Qaidam Lake of the NE Qinghai-Tibetan Plateau and itsimplication for palaeoenvironmental changes. Quaternary Geochronology 5,360e364.

Wang, S.M., Shi, Y.F., 1992. Review and discussion on the late Quaternary evolutionof Qinghai Lake. Journal of Lake Science 4, 1e9 (in Chinese with Englishabstract).

Yang, B., Braeuning, A., Shi, Y.F., Chen, F.H., 2004. Evidence for a late Holocene warmand humid climate period and environmental characteristics in the arid zonesof northwest China during 2.2w1.8 kyr B.P. Journal of Geopgysical Research 109,D02105.

Yu, J.Q., 2005. Lake Qinghai, China: a multi-proxy investigation on sediment coresfor the reconstructions of paleoclimate and paleoenvironment since the Marineisotope stage 3. Dissertation, Faculty of Materials and Geoscience, TechnicalUniversity of Darmstadt.

Yuan, B.Y., Chen, K.Z., Bowler, J.M., Ye, S.J., 1990. The formation and evolution of theQinghai Lake. Quaternary Science 3, 233e243 (in Chinese with Englishabstract).

Zhang, J.R., Jia, Y.L., Lai, Z.P., Long, H., Yang, L.H. Holocene lake evolution ofHuangqihai Lake in semi-arid northern China based on sedimentary sequenceand luminescence dating. The Holocene, submitted for publication.

Zhang, P.X., Zhang, B.Z., Yang, W.B., 1988. The evolution of the water body envi-ronment in Qinghai Lake since the postglacial age. Acta Sedimentologica Sinica6, 1e14 (in Chinese with English abstract).

Zhang, P.X., Zhang, B.Z., Qian, G.M., Li, H.J., Xu, L.M., 1994. The study of paleoclimaticparameter of Qinghai Lake since Holocene. Quaternary Sciences 3, 225e238 (inChinese with English abstract).