lable at ScienceDirect

Quaternary Science Reviews 177 (2017) 78e87

Contents lists avai

Quaternary Science Reviews

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

Loess deposits in Beijing and their paleoclimatic implications duringthe last interglacial-glacial cycle

Shengchen Tian a, b, Jimin Sun a, b, *, Zhijun Gong c

a Key Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, Chinab University of Chinese Academy of Science, Beijing, Chinac East China University of Science and Technology, Nanchang, China

a r t i c l e i n f o

Article history:Received 17 April 2017Received in revised form13 October 2017Accepted 17 October 2017

Keywords:LoessOSL datingPaleoclimateThe last interglacial-glacial cycleBeijing

* Corresponding author. Key Laboratory of CenozoInstitute of Geology and Geophysics, Chinese Academ

E-mail address: jmsun@mail.igcas.ac.cn (J. Sun).

https://doi.org/10.1016/j.quascirev.2017.10.0230277-3791/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

Loess-paleosol sequences are important terrestrial paleoclimatic archives in the semi-arid region ofnorth-central China. Compared with the numerous studies on the loess of the Chinese Loess Plateau, theeolian deposits, near Beijing, have not been well studied. A new loess section in the northeast suburb ofBeijing provides an opportunity for reconstructing paleoenvironmental changes in this region. Anoptically stimulated luminescence (OSL) chronology yields ages of 145.1 to 20.5 ka, demonstrating thatthe loess deposits accumulated during the last interglacial-glacial cycle. High-resolution climatic proxies,including color-index, particle size and magnetic parameters, reveal orbital-scale climatic cycles, cor-responding to marine oxygen isotope stages (MIS) 6 to MIS 2. In contrast to the loess deposits of thecentral Loess Plateau, loess near Beijing is a mixture of distal dust materials from gobi and sand deserts inthe arid part of northwestern China and proximal, local alluvial sediments. Climatic change in Beijingduring the last interglacial-glacial cycle was controlled primarily by the changing strength of the EastAsian monsoon. Paleosols developed during the last interglacial complex (between 144.0 and 73.0 ka)and the interstadial of the last glaciation (between 44.6 and 36.2 ka), being associated with an enhancedsummer monsoon in response to increased low-latitude insolation and a weakened Siberia High. Loessaccumulation occurred during cold-dry stages of the last glaciation, in response to the intensified wintermonsoon driven by the strengthened Siberia High and its longer residence time.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Loess covers about 10% of the land surface of the Earth (Tayloret al., 1983; Liu, 1985; Tsoar and Pye, 1987). Chinese loess de-posits consist of alternations of loess layers and paleosols. The loesslayers are indicative of cold and dry climate, while paleosolsdeveloped during warm and humid climatic conditions (Heller andLiu, 1982, 1984; Liu, 1985; Kukla et al., 1988). Continuous loess-paleosol sequences associated with glacial-interglacial cycles dur-ing the Quaternary are among the most detailed terrestrial recordsof climate changes since ~2.6 Ma (Heller and Liu, 1984; Kukla andAn, 1989). Various paleoclimatic proxies extracted from loess-paleosol sequences, such as magnetic susceptibility, particle size,stable isotopes, and pollen, are used to infer Quaternary climatic

ic Geology and Environment,y of Sciences, Beijing, China.

changes (e.g., Heller and Liu, 1982, 1984; Muhs, 2013).In China, loess accumulates not only on the Loess Plateau, but

also occurs in other arid and semi-arid areas, especially on thewindward sides of mountain ranges. These piedmont loess depositsare also useful for inferring regional climatic changes. In an earlystudy loess was reported in the Zhaitang Basin, in a northwesternsuburb of Beijing, where the last-glacial-aged Malan loess wasnamed (Liu, 1985). A few studies including chronology, composi-tion, and texture have been performed on this section (e.g., Li, 1990;Wang et al., 1998). However, compared with the detailed work onthe loess-paleosol sequences from the Loess Plateau, loess depositsin Beijing have not been well studied.

In recent years, a well-preserved loess section in a suburb ofBeijing was found and is the subject of the present study. Inexamining this new loess exposure, we have three objectives: (1) togenerate a chronology of this section using the optically stimulatedluminescence (OSL) dating method; (2) to reconstruct paleoenvir-onmental changes using multiple climatic proxies; and (3) todiscuss the provenance of the loess deposits and the forcing

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e87 79

mechanism for the paleoclimatic change.

2. Geological setting

Beijing is situated in North China (Fig. 1a). The geomorphologyof the Beijing region is characterized by a series of mountains in thenorth and the northwest, whereas alluvial plains are found in thesouth. The elevations of the mountains vary between 800 and1500 m above sea level (asl), while the average elevation of thealluvial plain is 20e60 m asl (Bureau of Geology and MineralResources of Beijing Municipality, 1991).

Beijing has a typical temperate monsoon climate, which ischaracterized by distinctive seasonal changes in temperature,precipitation and wind direction. The mean annual temperature is11e12 �C in the alluvial plain area and gradually deceases towardthe mountain areas. The mean annual precipitation is about630 mm, but most rainfall (more than 70%), transported by thesoutheast summer monsoon from the Pacific Ocean, occurs insummer season (from June to August) (Editorial Committee of LocalChronicles of Beijing, 1999). In winter, the climate is controlled bythe Siberia High and the prevailing wind is northwesterly, acomponent of the winter monsoon regime.

The loess section studied here (40�20.160N, 116�51.570E) islocated about 60 km to the northeast of Beijing, lying in a small,east-west, extension-derived intermontane basin (Fig. 1b). Thisbasin is bounded by the Yanshan Mountains to the north with el-evations ranging from 600 to 1500 m asl, and a relatively lowerNanshan Mountains to the south. The section is located on thewindward (northwest) slope of the southern mountain range,where airborne dust accumulated as loess cover on the mountainslopes and accumulated on top of fluvial terraces, due to blocking ofthe prevailing northwesterly winter-monsoon winds by themountain range (Fig. 2a).

3. Materials and methods

The loess section studied is about 12 m thick, consisting of loess

Fig. 1. (a) Map showing the mountain, gobi, sand desert and loess distributions in China as wred arrows indicate directions of the prevailing near-surface winds. (b) Digital elevation msampling section. (For interpretation of the references to colour in this figure legend, the r

and interbedded paleosols, and six units can be recognized(Fig. 2b). Detailed lithological features of each unit are shown inFig. 3. The uppermost unit is a light-yellow reworked loess layer,with a thickness of about 0.5 m, the existence of intercalatedangular gravels indicates a fluvial origin. This layer is unconform-ably underlain by a dull orange (7.5YR 6/4) loess bed which is about1.5 m thick. Underlying this loess bed is a reddish brown (5YR 4/6)weakly developed paleosol with a thickness of 2.1 m. Another dullorange (7.5YR 6/4) loess bed conformably underlies this weakpaleosol, which is about 4.9 m thick. Near the bottom of this sectionthere is a well-developed bright reddish brown (5YR 5/6) paleosolwith a thickness of 2.3 m. This paleosol is conformably underlain bya light-yellow loess bed whose base is not exposed.

Eight samples were collected for OSL dating (Fig. 3) byhammering stainless steel tubes into cleaned vertical sections. Thetubes were sealed at both ends using black plastic bags, to preventthe samples from exposure to sunlight and to preserve their naturalwater contents. In the laboratory, the outer layers at both ends ofthe tube were extracted for water content and dose rate mea-surements. The remaining material for each sample was treatedwith 10% HCl and 10% H2O2 to remove carbonates and organics,respectively. Quartz and K-feldspar separates were prepared, usinga combination of sieving, heavy liquid separation and HF etchingprocedures in subdued, safe red light conditions (Li et al., 2007).Particles between 63 and 90 mm were obtained by dry sieving.Quartz grains (density between 2.62 and 2.75 g/cm3) and K-feld-spar grains (density between 2.53 and 2.58 g/cm3) were separatedusing sodium polytungstate heavy liquids. The quartz grains werethen treated with 40% HF to remove remaining feldspars and outerlayers irradiated by alpha particles, while the K-feldspar grainswere etched by 10% HF to remove the alpha irradiated outer layers.HCl (10%) was used again to dissolve residual fluorides. Finally, thequartz and K-feldspar grains were mounted on 10-mm-diameteraluminum discs with silicon oil. For OSL dating of quartz grains,contaminationwith incompletely dissolved feldspars can affect theequivalent doses and the shape of the growth curves (Lai andBrückner, 2007), so infrared-stimulated luminescence (IRSL)

ell as the locations of Beijing and Luochuan loess sections (after Sun and Zhu, 2010). Theodel of the studied region in the northeast suburb of Beijing and the location of theeader is referred to the web version of this article.)

Fig. 2. Photos showing that loess accumulated on the windward slope of the southern mountain forming loess deposits in the studied region (a), and the stratigraphic units of thestudied loess section (b).

Fig. 3. Lithology of the studied loess section and the positions of optical dating samples.

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e8780

(Duller, 2003) and the 110 �C TL peak (Li et al., 2002) weremeasured to ascertain the purity of the quartz grains.

OSL dating measurements were performed by using an auto-mated Risø TL/OSL-DA-15 reader (Markey et al., 1997). The equiv-alent doses (De) for quartz were determined by using the single-aliquot regeneration (SAR) protocol (Murray and Wintle, 2000). Inorder to obtain a suitable preheat temperature, a preheat plateautest was carried out on sample MY-2 before measurements. The

result indicates that the De values were relatively stable at preheattemperatures from 220 to 280 �C (Fig. 4a). Recuperation ratio andrecycling ratio tests were also completed (Fig. 4b and c) by applyinga zero dose and a repeated dose, respectively (Murray and Wintle,2000; Lai and Wintle, 2006) and the results indicate that thesensitivity change is small at 220 �C. According to the tests above,we choose 220 �C as the preheat temperature. The detailed pro-cedure of De measurement is shown in Table 1. The equivalent

Fig. 4. Equivalent dose (De) (a), Recuperation ratio (b) and recycling ratio (c) as a function of the preheat temperature for sample MY-2. Each point represents the mean value of sixaliquots at each temperature. The horizontal dashed line in (a) denotes the plateau obtained for De values from preheat temperatures of 220 �C, 240 �C, 260 �C and 280 �C.

Table 1SAR protocol used for this study.

Step Treatment Observed

1 Apply regeneration dose Di (Di ¼ 0 Gy if natural) e

2 Preheat at 220 �C for 10s e

3 Stimulate with blue light for 40 s at 125 �C Li4 Apply test dose e

5 Heat to 200 �C e

6 Stimulate with blue light for 40 s at 125 �C Ti7 Repeat steps 1e6 to apply a range of regeneration dose

(including a zero dose and a repeated dose)

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e87 81

doses for K-feldspar aliquots were measured using the multi-elevated-temperature post-IR IRSL (MET-pIRIR) protocol (Li andLi, 2011). A preheat at 280 �C for 10 s was used for both regenera-tive dose and test dose measurements. The IRSL signals weremeasured by progressively increasing the stimulation temperaturefrom 50 to 250 �C in steps of 50 �C and the De values from 200 to250 �C were used for age calculations.

The environmental dose rate was determined by using varioustechniques and the results are shown in Table 2. The concentrationsof U and Th were determined by inductively coupled plasma massspectrometry (ICP-MS) and the K content was measured by atomicabsorption spectroscopy (AAS). Water contents were calculatedfrom the sample weights before and after drying at 100 �C. Thecosmic ray contribution to the dose rate was calculated from thegeomagnetic latitude, the burial depth, and the elevation of thesample locality (Prescott and Hutton, 1994). The internal dose ratefor K-feldspar was calculated using a K concentration of 13 ± 1% anda Rb content of 400 ± 100 ppm (Zhao and Li, 2005; Li et al., 2008).

Bulk samples (a total of 108) were collected at 10 cm intervalsfor analyses of color, particle size and magnetic parameters.

Color of loess and paleosols was expressed by the color index,which was determined using an automated YT-ACM colorimeter.

Table 2Dose rate, equivalent dose and OSL ages.

Samples U(ppm)

Th(ppm)

K(%)

Water content(%)

MY-1 1.65 ± 0.17 10.4 ± 1.0 2.21 ± 0.22 15MY-2 1.75 ± 0.18 10.6 ± 1.1 2.14 ± 0.21 10MY-3 1.72 ± 0.17 10.5 ± 1.1 2.09 ± 0.21 10MY-4 1.92 ± 0.19 11.2 ± 1.1 2.14 ± 0.21 14MY-5 1.90 ± 0.19 12.5 ± 1.3 1.98 ± 0.20 8MY-6 1.57 ± 0.16 10.8 ± 1.1 2.08 ± 0.21 10MY-7 2.11 ± 0.21 11.8 ± 1.2 2.20 ± 0.22 10MY-8 1.82 ± 0.18 11.9 ± 1.2 2.31 ± 0.23 12

a The error for the water content is estimated at ± 20%.b The error for the cosmic rays dose rate is estimated at ± 10%.

Dried samples were ground to a size fraction of less than 200 mesh(<75 mm) and then were pressed uniformly before measurements.

Particle size measurement was carried out by using a computer-operated Malvern Mastersizer 3000 laser particle size analyzer,with a measuring range from 0.01 to 3500 mm. The samples weretreated with H2O2 and HCl to remove organic matter and carbon-ates. Ultrasonic pretreatment involving the addition of 0.05 mol/L(NaPO3)6 solution was also used to disperse the samples beforeparticle measurement.

Air-dried samples for magnetic analysis were weighed and thenpacked in non-magnetic plastic boxes. A Batington MS3 instrumentwas used to measure mass magnetic susceptibility. Low-frequencymagnetic susceptibility (clf) was measured at 0.47 kHz, while high-frequency magnetic susceptibility (chf) was obtained at 4.7 kHz.Frequency-dependent magnetic susceptibility (cfd) was calculatedwith the formula: cfd (%) ¼ (clf - chf)/clf � 100. Saturationisothermal remanent magnetization (SIRM) was acquired in the 1 Tfield at room temperature by using an American 2G-760 super-conducting rock magnetometer.

4. Results and interpretations

4.1. Luminescence dating results

The chronology of the loess section was obtained using a com-bination of data from quartz and K-feldspar grains (Table 2). The Devalues of samples MY-1 to MY-5 were obtained by measuringquartz grains, while samples MY-6 to MY-8 were measured with K-feldspar grains because the signals of quartz grains were saturated.De distributions of all samples are displayed using histograms(Fig. 5) and radial plots (Fig. 6), respectively. For each sample, the Devalues of all discs are normally distributed (Fig. 5) and most of thedata points fall within the ± 2s standard error band (Fig. 6),meaning that the mineral grains were uniformly bleached prior to

a Cosmic rayb

(Gy/ka)De rate(Gy/ka)

De(Gy)

Optical age(ka)

0.18 3.01 ± 0.17 61.81 ± 2.67 20.5 ± 1.50.15 3.10 ± 0.17 112.13 ± 3.25 36.2 ± 2.20.12 3.00 ± 0.17 133.91 ± 3.97 44.6 ± 2.90.11 3.03 ± 0.17 151.48 ± 5.54 50.0 ± 3.30.08 3.12 ± 0.16 170.21 ± 5.41 54.5 ± 3.30.07 3.43 ± 0.18 250.27 ± 5.57 73.0 ± 4.00.05 3.70 ± 0.18 532.77 ± 17.87 144.0 ± 8.50.06 3.53 ± 0.18 511.45 ± 5.39 145.1 ± 7.4

Fig. 5. Histograms of De distributions for all luminescence dating samples. For each sample, the De values are normally distributed.

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e8782

deposition. Therefore, the average of all measured discs of eachsample was used as its equivalent dose for age calculation.

The luminascence dating results show that most of the lastinterglacial-glacial cycle is represented in this loess section (Fig. 7).The basal age of the section is 145.1 ± 7.4 ka, dating the uppermostpart of the oldest loess to the latest part of marine oxygen isotopestage (MIS) 6, the penultimate glacial period, according to theSPECMAP time-scale (Imbrie et al., 1984). IRSL ages from the bottomand the top of the well-developed paleosol are 144.0 ± 8.5 ka and73.0 ± 4.0 ka, respectively. Compared with the time-scale ofSPECMAP (Imbrie et al., 1984), this paleosol developed during thewhole of the last interglacial complex, corresponding to MIS 5. OSLdating from the top of the loess bed overlying this paleosol yieldsan age of 44.6 ± 2.9 ka, suggesting that this loess unit accumulatedfrom 73.0 to 44.6 ka, a period corresponding to MIS 4. The weakpaleosol developed between 44.6 and 36.2 ka, an interstadialperiod corresponding to MIS 3. The OSL age from the top of theloess unit overlying this weak paleosol is about 20.5 ka, suggestingthat it was deposited during the period corresponding to MIS 2, themost recent glacial period.

4.2. Color parameters

The color results are expressed by the color-coordination systemCIE L*a*b* (Commission Internationale de L’�Eclairage (CIE), 1978).For all samples, color indices of a* (redness relative to greenness)and L* (lightness) were used for analysis.

Redness of loess and paleosols is controlled primarily by typesand concentrations of iron oxides (Yang et al., 2001; Sun et al.,2011), including magnetite, hematite, maghemite and goethite (Jiet al., 2001; Chen et al., 2002). But among these minerals, rednessis most significantly influenced by the amount of hematite (Yanget al., 2001; Yang and Ding, 2003; Sun et al., 2011). Hematiteforms from chemical weathering of primary, Fe-bearing minerals(Walker, 1967), and is thus climate-dependent (Sun et al., 2015).Higher temperature and more precipitation associated with thesummer monsoon are beneficial to chemical weathering, so astrengthened summer monsoonwill lead to increased formation of

hematite and an increase in the redness of a soil or paleosol. Thevariation of redness is shown in Fig. 7a. Two peaks can be recog-nized in MIS 3 and MIS 5, implying an enhanced summer monsoonduring these two periods. Redness values are lower in MIS 2 andMIS 4, suggesting that the summer monsoon was weaker duringglacial times, and/or loess accumulation rates exceeded rates of soilformation and chemical weathering.

Lightness is generally controlled by the amount of carbonatesand organic matter (Yang et al., 2001). It is positively correlatedwith carbonate content, but negatively correlated with organicmatter content. The carbonate contents of loess and paleosols areinfluenced by pedogenic leaching (Liu, 1985; Chen et al., 2002),while the organic matter content is mainly affected by vegetationcover and biomass (Gasse et al., 1991; Zhou et al., 1996; Zaady et al.,2001). The vertical variation of lightness is opposite to that ofredness (Fig. 7b). Two troughs correspond to MIS 3 and MIS 5. Thelower values can be attributed to stronger carbonate leaching andhigher organic matter content during soil development when thesummer monsoonwas strengthened. Lightness values are higher inMIS 2 and MIS 4, which is associated with less carbonate leachingand lower organic matter content in winter monsoon dominatedglacial periods.

4.3. Median particle size

Particle size variations of loess deposits in northern China areassociated with the variable intensity of the East Asian wintermonsoon, in turn driven by the strength of the Siberia High (Liu,1985; An et al., 1991; Lu and An, 1998). The median particle sizeis a commonly used proxy for winter monsoon strength (Xiao et al.,1995; Zhou et al., 1996).

Significantly lower median particle size values are found in thepaleosols corresponding to MIS 3 and MIS 5 (Fig. 7c). The finermedian particle sizes suggest that the winter monsoon was weak-ened during these two stages. MIS 2 andMIS 4 have coarser medianparticle sizes, implying strengthened winter monsoon duringglacial periods. It is important to note here that loess accretion inChina continues, albeit at a much lower rate, even when soils form

Fig. 6. Radial plots of De distributions for all luminescence dating samples. The dashed area indicates the ± 2s standard error band.

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e87 83

during interglacial and interstadial periods. During these periods,however, loess particle size and flux are greatly reduced due to theweaker winter winds of the East Asian monsoon. Thus, the finerparticle size in paleosols in China is due both to continued low ratesof eolian accretion and production of fine particles by chemicalweathering.

4.4. Magnetic parameters

Magnetic susceptibility (c) depends on types, contents andparticle diameters of magnetic minerals (Kukla et al., 1988; Kuklaand An, 1989; Maher and Thompson, 1991; Maher, 2016). Themain types of magnetic minerals of eolian deposits are magnetite,maghemite and hematite (Heller and Liu, 1982; Liu, 1985; Maher

Fig. 7. Vertical variations of color index (a and b), median particle size (c) and magnetic parameters (deg), and their correlations with the marine oxygen isotope stages. (Forinterpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e8784

and Thompson, 1991; Liu et al., 1992a,b, 1995).Magnetic susceptibility is generally higher in soils than in loess.

The enhancement of susceptibility in paleosols can be explained bya variety of mechanisms. Heller and Liu (1982, 1984) proposed thatsoil compaction and carbonate leaching would cause a relativeenrichment of magnetite. Kukla et al. (1988) and Kukla and An(1989) suggested that the stable influx of ultrafine magnetitewould be diluted when the accumulation of dust increased duringglacial periods. Zhou et al. (1990) argued that the enhancement ofsusceptibility in paleosols could be partly ascribed to the new for-mation of magnetic minerals during pedogenic processes. Sun andLiu (2000) suggested that magnetic susceptibility may have mul-tiple origins, and all of the mechanisms above as well as the effectof source materials should be taken into consideration.

Two peaks correspond to MIS 3 and MIS 5 (Fig. 7d). Theenhancement of clf can be attributed to the newmagnetic mineralsformed during soil formation, which implies a strengthened sum-mer monsoon. clf values are lower in MIS 2 and MIS 4, suggestingless new formed pedogenic magnetic minerals due to the weak-ened summer monsoon during glacial times.

Compared to clf, frequency-dependent magnetic susceptibility(cfd) is more sensitive to paleoclimate changes (Liu et al., 1992a,b).cfd reflects the relative proportion of the susceptibility contributedby superparamagnetic minerals (<0.03 mm) that are mainly formedduring pedogenic processes (Maher, 1986, 2016; Liu et al., 1992a,b);warmer and wetter climate is more favorable for the formation ofultrafine magnetic minerals and the enhancement of cfd. As shownin Fig. 7f, the values of cfd are higher in MIS 3 and MIS 5, implying astronger summer monsoon during these two stages. The lowervalues of cfd correspond toMIS 2 andMIS 4, suggesting aweakenedsummer monsoon during glacial periods and/or rates of loess ac-cretion that exceed the rates of pedogenesis.

Saturation isothermal remanent magnetization (SIRM) is mainlyassociated with the content of magnetite (Liu et al., 1992a,b). Theenhancement of SIRM in paleosols is partly ascribed to the newformation of magnetite during pedogenic processes. Similar to thevariation trend of clf, two prominent peaks occur in MIS 3 and MIS5 (Fig. 7g), suggesting an enhancement of pedogenesis caused by astrengthened summer monsoon. MIS 2 and MIS 4 have lower

values of SIRM. The weaker pedogenesis associated with weakenedsummer monsoon can account for the decrease of SIRM.

4.5. Particle size distribution

Log-probability plots were used to display particle-size cumu-lative distributions (Fig. 8). The results indicate that the curves ofboth loess and paleosol can be divided into two log-normal seg-ments. Each segment is related to a different mode of sedimenttransport (Visher, 1969). Based on experiments inwind tunnels andstudies of modern dust storms, Pye (1987) summarized the modesof eolian sediment transport for different size fractions (Fig. 8). Theupper segment (Fig. 8) represents particles finer than 70 mm (>3.8Ф), which can be transported by short-term and/or long-termsuspension over relatively long distances. This segment makes upabout 55% of the total distribution in loess, while it is up to 90% inthe paleosols. The lower segment (Fig. 8), with a size range from 70to 500 mm (3.8-1Ф), comprises about 45% of the total distribution inloess and 10% in the paleosols. This size fraction can be onlytransported by saltation and for short distances.

Additionally, the particle size distributions of loess and paleo-sols from Beijing were compared with those from the Luochuansection in the central Loess Plateau (see location in Fig. 1a). Theresults indicate two major differences between them (Fig. 9). First,samples from Beijing aremuch coarser than samples from the LoessPlateau (Fig. 9a and b). Second, samples from Beijing and Luochuanfall into two distinct fields (Fig. 9c and d). The eolian deposits fromBeijing contain more particles of clay-size and sand-size comparedwith those of the Luochuan samples.

5. Discussion

5.1. Provenance of the loess deposits in Beijing

Provenance of dust materials is the basis for understanding dustaccumulation processes and for interpreting its paleoclimaticimplication. Although the provenance of loess on the Chinese LoessPlateau has been studied extensively (e.g., Liu, 1985; Derbyshireet al., 1998; Sun et al., 2001; Sun, 2002), the source area of loess

Fig. 8. Cumulative probability curves of loess and paleosol from Beijing and their correlations with the transport modes of eolian sediments (after Pye, 1987).

Fig. 9. Particle size distributions (a and b) and ternary diagrams comparing relative contents of clay-, silt- and sand-size particles (c and d) of loess and paleosols from Beijing andLuochuan.

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e87 85

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e8786

in Beijing has not been well studied. Although we have no newmineralogical, geochemical or isotopic data that bear on the issue ofsource areas, particle size data can be used to make some broadinferences about the provenance of loess deposits in the Beijingarea.

The particle-size cumulative distributions shown by the log-probability plots (Fig. 8) suggest that except for the long-rangetransportation of dust from the upwind gobi-desert regions, thereare proximal coarse particles from local sources for eolian depositsin Beijing, especially for loess. Comparisons have been made be-tween samples from Beijing and Luochuan, and the results shownin Fig. 9 suggest that the eolian deposits from Beijing contain moreparticles of clay-size and sand-size than those from Luochuan.

These results can be attributed to different transport distancesand source materials. Compared with the Luochuan Section, Beijinghas longer distances from the northwest source areas of gobi andsand deserts (see Fig. 1a). This can account for the relatively higherclay-size fraction in the loess deposits near Beijing. The highercontent of sand-sized particles in the eolian deposits in Beijing isattributed to local sources. Because the studied region is located inan intermontane basin, alluvial sediments can be transported to thepiedmont forming alluvial fan or river channel deposits that areloess sources. Wind sorting on these proximal sediments can leadto the transport of coarse particles to the nearby loess site in Beijing.In contrast, the Luochuan Section lies in the interior of the LoessPlateau and there are no local mountains for providing coarsematerials. Moreover, although there are river valleys, they are allcut to a depth of over 100 m. In such cases, the coarse particles inthe proximal river valleys are not likely to be transported to thehigher-elevation surface where the Loess Plateau is situated.

5.2. Climatic changes driven by the East Asian monsoon during thelast interglacial-glacial cycle

The temporal variations of multiple climatic proxies of the loesssection studied indicate that paleoclimatic change in Beijing duringthe last interglacial-glacial cycle was associated with the waxingand waning of the East Asian monsoon circulations. The strengthand timing of the East Asian monsoon are affected mainly by thevariations in solar radiation and global ice volume (Ding et al., 1994;Ding and Liu, 1998; Porter, 2001).

During the last interglacial (between ~144 ka and ~73 ka) andthe interstadial of the last glaciation (~45 ka to ~36 ka), the EastAsian summer monsoon was strengthened considerably due toincreased summer insolation in the latitude (Berger and Loutre,1991). First, increased low-latitude solar radiation caused highersea surface temperatures and an expandedWest PacificWarm Pool,along with more moisture evaporation from the western PacificOcean. Second, the melting of ice sheets and sea ice in high lati-tudes induced by the increased high -latitude insolation gave rise tohigher sea level. The enhanced summer monsoon would bringmore warm and humid air from the low-latitude oceans to theinterior of East Asia, leading to a warmer and wetter climate inBeijing. Moreover, the decay of ice sheets in high latitudes resultedin lower land surface albedo and fewer cold air outbreaks from thepolar region to middle latitudes, decreasing the intensity andduration of the Siberia High. Therefore, the winter monsoon wasgreatly weakened; the predominant warm and humid climate wasfavorable for soil development in Beijing.

During MIS 4 (from 73.0 to 44.6 ka) and the early stage of MIS 2(from 36.2 to 20.5 ka), global cooling caused by the reduced inso-lation led to the expansion of ice sheets and sea ice in high lati-tudes. Increased ice volume could intensify the winter winds bystrengthening the Siberia High (Ding et al., 1995; Liu et al., 1999).The enlarged Eurasian ice sheets increased the albedo of the

continental surface and then cooled the air above the Siberian re-gion, caused cold air to subside and then thermally enhanced thestrength of the high pressure cell. Additionally, the presence ofextensive ice sheets and sea ice in high latitudes would prolong theduration of the Siberia High (Ding et al., 1995; Liu and Ding, 1998),leading to longer-lasting winter monsoon and impeding thenorthward movement of summer winds. Moreover, the decreasedsea surface temperature, the exposure of large shelf areas of themarginal seas in the Western Pacific region and the reduced size ofthe West Pacific Warm Pool would also result in weakened EastAsian summermonsoon (Wang,1998,1999). All these paleoclimaticconditions can help explain loess accumulation during the twocold-dry stages of the last glacial period in Beijing.

6. Conclusions

(1) Luminescence dating of the eolian deposits in Beijing yieldsan age range of 145.1 to 20.5 ka, suggesting that the loess-paleosol sequence accumulated during the last interglacial-glacial cycle.

(2) High-resolution climatic proxies, including color-index, par-ticle size, and magnetic parameters, reveal orbital-scale cli-matic cycles, corresponding to marine oxygen isotope stagesof MIS 6 to MIS 2.

(3) Compared with loess and paleosols of the Luochuan Sectionon the central Loess Plateau, loess near Beijing contains morecoarse particles, suggesting that the they are derived fromproximal local alluvial sediments; the higher content of clay-size particles can be attributed to longer distances from thenorthwest source areas of gobi and sand deserts.

(4) Paleoclimatic change in Beijing during the last interglacial-glacial cycle was a response to the variations of the EastAsian monsoon. The intensified winter monsoon duringglacial stages was controlled mainly by the prolonged dura-tion and the increased strength of the Siberian High, drivenby both decreased insolation and enlarged high-latitude icesheets in the Northern Hemisphere. In contrast, theenhanced summer monsoons during the last interglacialperiod and the interstadial of the last glaciation werecontrolled largely by the increased low-latitude solar radia-tion in summer together with a decline in the strength of theSiberia High.

Acknowledgements

This work was supported by the National Nature ScienceFoundation of China (41290251 and 41672168). We thank DaniesMuhs and an anonymous reviewer for their valuable comments,language improvements and suggestions that have substantiallyimproved our manuscript.

References

An, Z., Kukla, G., Porter, S.C., Xiao, J., 1991. Late quaternary dust flow on the Chineseloess plateau. Catena 18, 125e132.

Berger, A., Loutre, M.F., 1991. Insolation values for the climate of the last 10 millionyears. Quat. Sci. Rev. 10, 297e317.

Bureau of Geology and Mineral Resources of Beijing Municipality, 1991. RegionalGeology of Beijing Municipality. Geological Publishing House, Beijing.

Chen, J., Ji, J., Balsam, W., Chen, Y., Liu, L., An, Z., 2002. Characterization of theChinese loessepaleosol stratigraphy by whiteness measurement. Palaeogeogr.Palaeoclimatol. Palaeoecol. 183, 287e297.

Commission Internationale de L’�Eclairage (CIE), 1978. Recommendations on Uni-form Color Spaces, Color-difference Equations, Psychometric Color Terms.Supplement No. 2 of CIE Publication No. 15. Bureau Central de la CIE, Paris.

Derbyshire, E., Meng, X., Kemp, R.A., 1998. Provenance, transport and characteristicsof modern aeolian dust in western Gansu Province, China, and interpretation ofthe Quaternary loess record. J. Arid Environ. 39, 497e516.

S. Tian et al. / Quaternary Science Reviews 177 (2017) 78e87 87

Ding, Z., Liu, T., 1998. Forcing mechanisms for East-Asia monsoonal variationsduring the late pleistocene. Chin. Sci. Bull. 43, 1497e1510.

Ding, Z., Liu, T., Rutter, N., Yu, Z., Guo, Z., Zhu, R., 1995. Ice-volume forcing of EastAsian winter monsoon variations in the past 800,000 years. Quat. Res. 44,149e159.

Ding, Z., Yu, Z., Rutter, N., Liu, T., 1994. Towards an orbital time scale for Chineseloess deposits. Quat. Sci. Rev. 13, 39e70.

Duller, G.A.T., 2003. Distinguishing quartz and feldspar in single grain luminescencemeasurements. Radiat. Meas. 37, 161e165.

Editorial Committee of Local Chronicles of Beijing, 1999. Meteorology Annals. Bei-jing Publishing House, Beijing, pp. 9e25 (in Chinese).

Gasse, F., Arnold, M., Fontes, J.C., Fort, M., Gibert, E., Huc, A., Li, B., Li, Y., Liu, Q.,Meli�er�es, F., Campo, E.V., Wang, F., Zhang, Q., 1991. A 13,000-year climate recordfrom western Tibet. Nature 353, 742e745.

Heller, F., Liu, T., 1982. Magnetostratigraphical dating of loess deposits in China.Nature 300, 431e433.

Heller, F., Liu, T., 1984. Magnetism of Chinese loess deposits. Geophys. J. R. Astron.Soc. 77, 125e141.

Imbrie, J., Hays, J.D., Martinson, D.B., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G.,Prell, W.L., Shackleton, N.J., 1984. The orbital theory of Pleistocene climate:support from a revised chronology of the marine delta 18O record. In: Berger, A.,Imbrie, J., Hays, J., Kakla, G., Saltzman, B. (Eds.), Milankovitch and Climate (PartI). Reidel, Dordrecht, pp. 269e305.

Ji, J., Balsam, W., Chen, J., 2001. Mineralogic and climatic interpretations of theLuochuan Loess section (China) based on diffuse reflectance spectrophotom-etry. Quat. Res. 56, 23e30.

Kukla, G., An, Z., 1989. Loess stratigraphy in central China. Palaeogeogr. Palae-oclimatol. Palaeoecol. 72, 203e225.

Kukla, G., Heller, F., Liu, X., Xu, T., Liu, T., An, Z., 1988. Pleistocene climates in Chinadated by magnetic susceptibility. Geology 16, 811e814.

Lai, Z., Brückner, H., 2007. Effects of feldspar contamination on equivalent doses andthe shape of growth curves for OSL of silt-sized quartz extracted from Chineseloess. Geochronometria 30, 49e53.

Lai, Z., Wintle, A.G., 2006. Locating the boundary between the Pleistocene and theHolocene in Chinese loess using luminescence. Holocene 16, 893e899.

Li, B., Li, S., 2011. Luminescence dating of k-feldspar from sediments: a protocolwithout anomalous fading correction. Quat. Geochronol. 6, 468e479.

Li, B., Li, S., Wintle, A.G., Zhao, H., 2008. Isochron dating of sediments using lumi-nescence of K-feldspar grains. J. Geophys. Res. 113, F02026.

Li, H., 1990. Thermoluminescence (TL) age of malan loess (IV) -an investigation ofthe Zhaitang profile. Geochimica 264e269 (in Chinese with English abstract).

Li, S., Chen, Y., Li, B., Sun, J., Yang, L., 2007. OSL dating of sediments from deserts innorthern China. Quat. Geochronol. 2, 23e28.

Li, S., Sun, J., Zhao, H., 2002. Optical dating of dune sands in the northeastern desertsof China. Palaeogeogr. Palaeoclimatol. Palaeoecol. 181, 419e429.

Liu, T., 1985. Loess and the Environment. China Ocean Press, Beijing, 251 pp.Liu, T., Ding, Z., 1998. Chinese loess and paleomonsoon. Annu. Rev. Earth Planet. Sci.

26, 111e145.Liu, T., Ding, Z., Rutter, N., 1999. Comparison of Milankovitch periods between

continental loess and deep sea records over the last 2.5 Ma. Quat. Sci. Rev. 18,1205e1212.

Liu, X., Heller, F., Liu, T., Xu, T., 1992a. Magnetic susceptibility of Chinese loess as apaleoclimatic indicator. Sci. China (Series B) 35, 612e620.

Liu, X., Liu, T., Heller, F., Xu, T., 1992b. Study on magnetic susceptibility of loess andQuaternary climate in China. Sci. Geol. Sin. 279e285 (in Chinese with Englishabstract).

Liu, X., Rolph, T.C., Bloemendal, J., Shaw, J., Liu, T., 1995. Quantitative estimates ofpalaeoprecipitation at Xifeng, in the loess plateau of China. Palaeogeogr.Palaeoclimatol. Palaeoecol. 113, 243e248.

Lu, H., An, Z., 1998. Paleoclimatic significance of grain size of loess-palaeosol depositin Chinese Loess Plateau. Sci. China (Series D) 41, 626e631.

Maher, B.A., 1986. Characterisation of soils by mineral magnetic measurements.Phys. Earth Planet. In. 42, 76e92.

Maher, B.A., 2016. Palaeoclimatic records of the loess/palaeosol sequences of theChinese Loess Plateau. Quat. Sci. Rev. 154, 23e84.

Maher, B.A., Thompson, R., 1991. Mineral magnetic record of the Chinese loess andpaleosols. Geology 19, 3e6.

Markey, B.G., Bøtter-Jensen, L., Duller, G.A.T., 1997. A new flexible system formeasuring thermally and optically stimulated luminescence. Radiat. Meas. 27,83e89.

Muhs, D.R., 2013. The geologic records of dust in the Quaternary. Aeolian Res. 9,3e48.

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

Porter, S.C., 2001. Chinese loess record of monsoon climate during the last gla-cialeinterglacial cycle. Sci. Rev. 54, 115e128.

Prescott, J.R., Hutton, J.T., 1994. Cosmic ray contributions to dose rates for lumi-nescence and ESR dating: large depths and long-term time variations. Radiat.Meas. 23, 497e500.

Pye, K., 1987. Eolian Dust and Dust Deposits. Academic Press, London, 334 pp.Sun, J., 2002. Provenance of loess material and formation of loess deposits on the

Chinese Loess Plateau. Earth Planet. Sci. Lett. 203, 845e859.Sun, J., Gong, Z., Tian, Z., Jia, Y., Windley, B.F., 2015. Late Miocene stepwise aridifi-

cation in the Asian interior and the interplay between tectonics and climate.Palaeogeogr. Palaeoclimatol. Palaeoecol. 421, 48e59.

Sun, J., Liu, T., 2000. Multiple origins and interpretations of the magnetic suscep-tibility signal in Chinese wind-blown sediments. Earth Planet. Sci. Lett. 180,287e296.

Sun, J., Zhang, M., Liu, T., 2001. Spatial and temporal characteristics of dust storms inChina and its surrounding regions, 1960-1999: relations to source area andclimate. J. Geophys. Res. 106, 10325e10333.

Sun, Y., He, L., Liang, L., An, Z., 2011. Changing color of Chinese loess: geochemicalconstraint and paleoclimatic significance. J. Asian Earth Sci. 40, 1131e1138.

Sun, J., Zhu, X., 2010. Temporal variations in Pb isotopes and trace element con-centrations within Chinese eolian deposits during the past 8 Ma: implicationsfor provenance change. Earth Planet. Sci. Lett. 290, 438e447.

Taylor, S.R., McLennan, S.M., McCulloch, M.T., 1983. Geochemistry of loess, conti-nental crustal composition and crustal model ages. Geochem. Cosmochim. Acta47, 1897e1905.

Tsoar, H., Pye, K., 1987. Dust transport and the question of desert loess formation.Sedimentology 34, 139e153.

Visher, G.S., 1969. Grain size distributions and depositional processes. J. Sediment.Petrol. 39, 1074e1106.

Walker, T.R., 1967. Formation of red beds in modern and ancient deserts. Geol. Soc.Am. Bull. 78, 353e368.

Wang, K., Hu, B., Yan, F., Guo, X., 1998. Texture and structure and paleoenvironmentof loess in Zhaitang Basin of Beijing. Beijing Geol. 1e10 (in Chinese with Englishabstract).

Wang, P., 1998. Western pacific in glacial cycles: seasonality in marginal seas andvariabilities of warm pool. Sci. China (Series D) 41, 35e41.

Wang, P., 1999. Response of Western Pacific marginal seas to glacial cycles: paleo-ceanographic and sedimentological features. Mar. Geol. 156, 5e39.

Xiao, J., Porter, S.C., An, Z., Kumai, H., Yoshikawa, S., 1995. Grain size of Quartz as anindicator of winter monsoon strength on the Loess Plateau of central Chinaduring the last 130,000 yr. Quat. Res. 43, 22e29.

Yang, S., Ding, Z., 2003. Color reflectance of Chinese loess and its implications forclimate gradient changes during the last two glacial-interglacial cycles. Geo-phys. Res. Lett. 30, 158e161.

Yang, S., Fang, X., Li, J., An, Z., Chen, S., Fukusawa, H., 2001. Transformation functionsof soil color and climate. Sci. China (Series D) 44, 218e226.

Zaady, E., Offer, Z.Y., Shachak, M., 2001. The content and contributions of depositedaeolian organic matter in a dry land ecosystem of the Negev Desert, Israel.Atmos. Environ. 35, 769e776.

Zhao, H., Li, S., 2005. Internal dose rate to K-feldspar grains from radioactive ele-ments other than potassium. Radiat. Meas. 40, 84e93.

Zhou, L., Oldfield, F., Wintle, A.G., Robinson, S.G., Wang, J., 1990. Partly pedogenicorigin of magnetic variations in Chinese loess. Nature 346, 737e739.

Zhou, W., Donahue, D.J., Porter, S.C., Jull, T.A., Li, X., Stuiver, M., An, Z., Matsumoto, E.,Dong, G., 1996. Variability of monsoon climate in East Asia at the End of the lastglaciation. Quat. Res. 46, 219e229.