The link between grain-size components and depositional ... · The link between grain-size components and depositional processes in a ... It is widely recognized that lake sediment

The link between grain-size components and depositionalprocesses in a modern clastic lake

JULE XIAO, ZHIGANG CHANG, JIAWEI FAN, LANG ZHOU, DAYOU ZHAI,RUILIN WEN and XIAOGUANG QINKey Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, ChineseAcademy of Sciences, Beijing 100029, China (E-mail: [email protected])

Associate Editor – Daniel Ariztegui

ABSTRACT

It is widely recognized that lake sediment grain-size distributions tend to be

polymodal and consist of two or more grain-size components. However, for

specific cases, the genesis of each component usually is poorly understood.

In this study, the grain-size components of the surface sediments of Hulun

Lake, Inner Mongolia, were partitioned using the log-normal distribution

function method and the relationship between the identity of each grain-size

component and the hydraulic condition of the lake was investigated in order to

relate the constituent components to specific depositional processes in the

lake. The data indicate that the modern clastic sediments of Hulun Lake

contain six distinct unimodal grain-size distributions representing six grain-

size components. Each of the components retains its identity including modal

size, manner of transportation and environment of deposition, although the

relative percentage varies with the hydraulic conditions throughout the lake.

These components are specified from fine to coarse modes as long-term

suspension clay, offshore-suspension fine silt and medium to coarse silt, and

nearshore-suspension fine sand, saltation medium sand and traction coarse

sand. The percentage contribution of several grain-size components

interpreted as being indicative of nearshore environments is shown to be

correlated negatively with water depth across the modern lake bed; this

suggests that the proportion of these components in core data might be useful

as a proxy for water depth. This possibility was tested using a sediment core

from Hulun Lake where high percentages of the nearshore grain-size

components were found to be correlated with low regional precipitation

reconstructed from the pollen profile of the same core. The coincidence of two

independent proxies does not only demonstrate the validity of log-normal

distribution function in partitioning polymodal sediments but reveals the

potential of lake sediment grain-size components for the research of lake-level

fluctuations during the geological past.

Keywords Grain-size component, Hulun Lake, lake-level status, offshoredistance, polymodal sediment, water depth.

INTRODUCTION

The grain-size distribution of lake sediments haslong been used to reconstruct the history ofchanges in the hydrology of lakes associated withregional climatic and environmental conditions

(Hakanson & Jansson, 1983). The grain-sizeparameters, however, can be regarded only asapproximate proxies because lake sediments arecharacterized by polymodal grain-size distribu-tions, and different components within an indi-vidual distribution might be formed by different

Sedimentology (2012) 59, 1050–1062 doi: 10.1111/j.1365-3091.2011.01294.x

1050 � 2011 The Authors. Journal compilation � 2011 International Association of Sedimentologists

depositional processes (Visher, 1969; Middleton,1976; Ashley, 1978). A reliable interpretation ofthe grain-size data obtained from lake sedimentsrequires an adequate understanding of the phys-ical process that may have governed the forma-tion of the grain-size components of polymodaldeposits in lakes.

With the aid of high-resolution grain-size datagenerated by laser grain-size analysers, mathe-matical methods have been applied to definegrain-size distributions and partition the constit-uent components of fluvial (Kranck et al., 1996;Passe, 1997), aeolian (Sun et al., 2002; Qin et al.,2005) and marine (Prins et al., 2000; Weltje &Prins, 2003) sediments. Until now, however, littleattention has been paid to the partitioning ofgrain-size components of lacustrine sedimentsdespite the fact that clastic materials deposited inlakes potentially provide an important source ofinformation about changes in regional hydrologyand global atmospheric circulation. In this study,the log-normal distribution function method wasused to numerically partition the grain-size com-ponents within individual grain-size distribu-tions of 40 surface–sediment samples takenalong two sections in Hulun Lake in north easternInner Mongolia, China. The purpose of this studyis to interpret the genesis of polymodal grain-sizedistributions of lake sediments and relate theconstituent grain-size components to specificsedimentary environments.

HULUN LAKE BASIN

Hulun Lake (48�30Æ667¢ to 49�20Æ667¢ N,117�0Æ167¢ to 117�41Æ667¢ E), the fifth largest lakein China, lies in an inland graben basin in InnerMongolia (Fig. 1). It has an area of 2339 km2 and amaximum water depth of 8 m when the elevationof the lake level is 545Æ3 m above sea-level (a.s.l.;measurements taken in August 1964; Xu et al.,1989). Low mountains and hills of Mesozoicvolcanic rocks border the lake on the north westand form a fault-scarp shoreline. Broad lacustrineand alluvial plains scattered with aeolian dunesare present along the southern and eastern shore.The lake has a catchment of 37 214 km2 withinthe borders of China. Two major rivers, the Herlunand Urshen Rivers, enter the lake from the south-west and south-east (Fig. 1). The Herlun Riverrises in the southern part of the Hentiy Mountainsin Mongolia and has a total channel length of1260 km. The Urshen River drains from Beir Lakethat has an area of 608 km2 and a maximum water

depth of >50 m, and flows northward throughdepressions, swamps and dunes into Hulun Lake;it has a total channel length of more than 200 km.Hydrological observations suggest that both riversremain at a slow and steady flow even during thehigh-rainfall years and no obvious river floodshave occurred since the beginning of observationsin 1957 (Xu et al., 1989). The Dalanolom River, anintermittent river, drains the lake when theelevation of the lake level exceeds 543Æ4 m a.s.l.,and enters the lake when the lake level is lowerand the high discharge of the Hailar River occa-sionally supplies water to it during spring floodsin May and June (Xu et al., 1989; Fig. 1).

Hulun Lake is located in a semi-arid area of themiddle temperate zone. The climate of the HulunLake region is under the influence of the EastAsian monsoon (Chinese Academy of Sciences,1984; Zhang & Lin, 1985). During the summers,the warm, moist southerly air masses interactwith cold air from the north west and producemost of the annual precipitation. During thewinters, the cold, dry north westerly airflowsprevail and bring strong winds and dust storms tothe region. In the lake region, mean annualtemperature is 0Æ3�C with a July average of20Æ3�C and a January average of )21Æ2�C. Annualprecipitation is 247 to 319 mm, and more than80% of the annual precipitation falls during Juneto September. Annual evaporation reaches 1400to 1900 mm, which is five to six times the annualprecipitation. The lake is covered with ca 1 m ofice from November to April.

SAMPLES AND METHODS

Sediment sampling

The surface sediments of Hulun Lake were sam-pled in August 2009 along the Transects A and B,yielding 21 samples for Transect A and 19samples for Transect B (Figs 1 and 2; Table 1).The transects start at A1 (49�9Æ901¢ N,117�25Æ364¢ E) and B1 (48�59Æ939¢ N, 11714Æ023¢ E)on the north western shore of the lake and end atA21 (49�1Æ478¢ N, 117�41Æ889¢ E) and B19(48�50Æ983¢ N, 117�31Æ003¢ E) on the south easternshore. At each site, a sediment core was retrievedin either a 40 cm or 60 cm long polyethylene tubeusing a gravity corer. The water–sediment inter-face and millimetre-scale thick whitish-greydeposits on the sediment top are clearly discern-ible in the core tube. The top 1 cm of each coresection was cut for a sample of the surface

Grain-size components of Hulun Lake sediments 1051

� 2011 The Authors. Journal compilation � 2011 International Association of Sedimentologists, Sedimentology, 59, 1050–1062

sediments after siphoning the water out of thecore tube with a plastic pipe. The water depthand offshore distance of each sampling site weremeasured at the time of extraction (Figs 1 and 2;Table 1).

Modern aeolian dust that was trapped on theice of Hulun Lake was collected in the centralpart of the lake (49�6Æ214¢ N, 117�30Æ012¢ E) in

March 2009. In addition, modern riverbed sandswere sampled close to the mouths of the Herlunand Urshen Rivers (Fig. 1) in August 2009.

Grain-size analysis

Grain-size distribution of all the samples wasdetermined with a Malvern Mastersizer 2000

MANCHURIA

Hulun Lake

49°20'China

A1Red Flag Fishery

49°00'B1

A21

48°40'

B19

NEW BARAGRIGHT BANNER

10 km0 117°30'117°00'

Fig. 1. Map of Hulun Lake (from http://maps.google.com) showing locations of the surface–sediment sampling sitesalong the Transects ‘A’ and ‘B’. The inset gives a sketch map of China showing the location of Hulun Lake.

1052 J. Xiao et al.


laser grain-size analyser (Malvern InstrumentsLimited, Malvern, UK). About 200 mg of sedi-ment from each air-dried, disaggregated samplewas pre-treated with 10 to 20 ml of 30% H2O2 toremove organic matter and then with 10 ml of10% HCl with the sample solution boiled toremove carbonates. About 2000 ml of deionizedwater was added and the sample solution waskept for 24 h to rinse acidic ions. The sampleresidue was dispersed with 10 ml of 0Æ05 m

(NaPO3)6 on an ultrasonic vibrator for 10 minbefore grain-size analysis.

The Mastersizer 2000 works on the principle ofthe Mie theory that predicts the way light isscattered by spherical particles and deals with theway light passes through, or is absorbed by, theparticle. Based on the Mie theory, assuming thatmeasured particles are perfect spheres, the Mas-tersizer uses the volume of a particle to measureits size and calculate the diameter of an imaginarysphere that is equivalent in volume to the tech-nique of ‘equivalent spheres’. The Mastersizer2000 has a measurement range of 0Æ02 to 2000 lmin diameter and a grain-size resolution of 0Æ166/in interval, thus yielding 100 grain-size fractions[/ = )log2(D), where D is the grain diameter inmillimetres]. The Mastersizer provides the per-

centage by volume of each grain-size fraction in asample. The relative error is <1% on the 50thpercentile and 2% on the outlying percentiles.

Fitting and partitioning of grain-sizecomponents

The grain-size distribution of clastic depositswith a single component should exhibit a uni-modal, symmetrical distribution on a logarithmicscale (Inman, 1949; Folk & Ward, 1957; Tanner,1964; Visher, 1969; Ashley, 1978). When theshape of a grain-size distribution is asymmetricalor skewed, the total distribution can be consid-ered to be a combination of several unimodaldistributions (Inman, 1949; Folk & Ward, 1957;Tanner, 1964; Visher, 1969; Ashley, 1978). Inother words, a polymodal grain-size distributioncan be thought to be composed of two or moreunimodal distributions that represent two ormore grain-size components (modes).

The grain-size distribution of unimodal clasticdeposits follows the log-normal distribution

Table 1. Water depth and offshore distance of 40surface–sediment sampling sites along the Transects‘A’ and ‘B’ in Hulun Lake (measurements taken inSeptember 2009).

Samplenumber

Waterdepth(m)

Offshoredistance(km)

Samplenumber

Waterdepth(m)

Offshoredistance(km)

A1 0Æ70 0Æ02 B1 1Æ00 0Æ06A2 1Æ00 0Æ05 B2 1Æ45 0Æ23A3 1Æ60 1Æ26 B3 1Æ95 0Æ82A4 2Æ10 1Æ00 B4 2Æ35 2Æ80A5 2Æ45 1Æ88 B5 2Æ75 5Æ00A6 2Æ85 3Æ45 B6 2Æ85 7Æ00A7 2Æ95 5Æ00 B7 2Æ85 9Æ00A8 3Æ00 7Æ00 B8 2Æ85 11Æ00A9 3Æ10 9Æ00 B9 2Æ85 13Æ00A10 3Æ15 11Æ00 B10 2Æ85 12Æ10A11 3Æ05 13Æ00 B11 2Æ85 10Æ10A12 3Æ00 11Æ30 B12 2Æ85 8Æ10A13 2Æ90 9Æ30 B13 2Æ80 6Æ10A14 2Æ90 7Æ30 B14 2Æ55 4Æ10A15 2Æ85 5Æ30 B15 2Æ25 2Æ10A16 2Æ55 3Æ30 B16 1Æ75 1Æ20A17 2Æ15 1Æ80 B17 1Æ25 0Æ90A18 1Æ60 0Æ80 B18 0Æ80 0Æ70A19 1Æ10 0Æ30 B19 0Æ40 0Æ50A20 0Æ65 0Æ10A21 0Æ40 0Æ02

Lake level0

A20A21

2

1

Wat

er d

epth

(m

)

A1

A2

A3

A4 A17

A18

A19

A20

3

021018620 246810124

A5

A6A7 A8 A9 A10 A11 A12 A13 A14 A15

A16

Lake level0

1 B1

B2B17

B19

B18

2

3W

ater

dep

th (

m)

B3

B4

B5B6 B7 B8 B9 B10 B11 B12 B13

B14B15

B16

0210186420 24681012

Offshore distance (km)

A

B

Fig. 2. Profile of the lake floor along the transectsshowing the sampling sites: (A) Transect A; (B) Tran-sect B. The vertical axis indicates water depth and thehorizontal axis indicates offshore distance to the closer(north western or south eastern) lakeshore.



(Krumbein, 1938). With respect to polymodalsediments that consist of several unimodal com-ponents, each of the components can be de-scribed with sufficient accuracy by the log-normaldistribution function (Ashley, 1978; Passe, 1997;Qin et al., 2005). In this study, the log-normaldistribution function method described by Qinet al. (2005) was applied to identify, fit andpartition the grain-size components of the surfacesediments of Hulun Lake. The formula of the log-normal distribution function is expressed asfollows:

FðxÞ ¼Xn

i¼1

ci

ri

ffiffiffiffiffiffi2pp

Z1

�1

expð� ðx � aiÞ2

2r2i

Þdx

24

35

where n is the number of modes, x = ln(d), d isthe grain size in lm, ci is the percentage of the ithmode, ci ‡0, and the sum of nci equals 100%. ri isthe standard deviation of the ith mode, ai is themean value of the ith mode’s logarithmized grainsize in lm, i.e., ai = ln(di), and ai >0.

The fitting residual is calculated as follows:

dF ¼ 1

m

Xmj¼1

ðFðxjÞ � GðxjÞÞ2

where m is the number of grain-size intervals,F(xj) is the fitted percentage of the jth grain-sizeinterval and G(xj) is the measured percentage ofthe jth grain-size interval. A lower value of dFindicates a better fitting result.

Fitting experiments begin by assuming thenumber of modes, n, of the measured polymodaldistribution; n is determined by identifying thenumber of peaks on the measured distributioncurve and the skewness of each peak. Here, it isnecessary to designate two modes for one skewedpeak. Each mode of a measured polymodal dis-tribution is described by three parameters (c, aand r), and a combination of the three parametersof n modes describes the measured polymodaldistribution. By using Visual Basic for applica-tions on a PC Office platform, a program waswritten to perform fitting calculations. The pro-cedure is as follows.

(i) Place the measured grain-size data of all thesamples into a worksheet of MS Excel. (ii) Startthe program, select the target sample, and theinitial values of the estimated parameters (c, aand r) of each mode in the sample are givenautomatically by the random number generatorwithin the program. The fitting calculations are

performed using the above estimated parameters(n, c, a and r) and the measured grain-size data toyield the initial fitting residual. (iii) The programstarts iterative-loop calculations of the fourparameters (n, c, a and r) and designates a groupof new parameters in each calculation to obtainthe minimum fitting residual value. Here,0 <ci <100, ai)1 < ai <ai+1, 0 <ri <r (r is the stan-dard deviation of the measured grain-size data ofthe sample). The increment of each parameter (Dc,Da and Dr) in each calculation is designated a lowvalue to ensure a good enough fitting (Dc = 0Æ01,Da = (ai ) ai)1)/100 and Dr = r/100). At the endof each calculation, the program automaticallyoutputs a fitting residual, compares it with theprevious lowest residual and chooses the lowerone as the minimum fitting residual of thesample. When the iterative-loop calculation isfinished, the combination of the three parametersof n modes with the minimum fitting residual ischosen to represent the measured polymodaldistribution.

Fitting experiments on a sample are accom-plished when the residual reaches its minimum.Numerical partitioning of the unimodal compo-nents of a measured polymodal distribution canbe achieved simultaneously through log-normaldistribution function fitting because the para-meters and distribution functions of each com-ponent are determined in the course of fitting.The curves of all the components of a sample andthe modal size and percentage of each componentare output automatically by the program whilethe fitting of the sample is accomplished. Themodal size of each component is equivalent invalue to the mean and median sizes because allthe components assume a normal distribution ona logarithmic scale of grain size in micrometres.The modal size of the ith mode is calculated asfollows:

�di ¼ expðaiÞ

To illustrate identifying, fitting and partitioningof grain-size components within a polymodaldistribution using the log-normal distributionfunction, six surface–sediment samples of HulunLake were selected from the Transects A and B(Fig. 3). These samples represent the modernsediments deposited in three different environ-ments of the nearshore zone (A2 and B1), offshorezone (A10 and B9) and the transitional zone (A16and B15) of the lake (Figs 1 and 2). The fittingand partitioning result and the minimum residualof each sample are shown (Fig. 3).

1054 J. Xiao et al.


RESULTS

Fittings of the log-normal distribution functionsuggest that each polymodal grain-size distribu-tion of samples from the surface sediments ofHulun Lake consists of three to five unimodaldistributions. These unimodal distributions canbe identified as six distinct modes representingsix grain-size components, respectively, accord-ing to the dominant range of modal sizes of eachunimodal distribution. The six components aredesignated C1 to C6 from fine to coarse modes inthis study (Fig. 3).

As shown in Fig. 3, it is easy to determinerelatively separated modes. To separate out twostrongly overlapped components (for instance, C2and C3 in sample B9 or C3 and C4 in sampleA16), however, requires technical skills throughtrial-and-error fittings after careful observation.Fortunately, in this case, the fitting residualincreases significantly if the minor component(C3 in samples B9 and A16) is ignored. Such

experiments demonstrate the validity and accu-racy of the log-normal distribution function infitting and partitioning the grain-size componentsof polymodal sediments.

Grain-size data suggest that the components C1,C2 and C3 to C6 identified from 40 individualgrain-size distributions of the surface–sedimentsamples from Hulun Lake have identifiable char-acteristics (Figs 4 and 5; Table 2). The modal sizesof C1, C2 and C3 to C5 vary primarily within rangesof 0Æ8 to 2Æ3 lm, 3Æ0 to 9Æ0 lm, 8Æ8 to 62Æ2 lm,65Æ7 lm to 160Æ7 lm and 176Æ2 to 650Æ3 lm, respec-tively. The coarsest component that appears insample A16 displays a modal size of 739Æ5 lm(Fig. 3). This component was designated an inde-pendent component, C6, because it has a modalsize that is much larger than the dominant modalsize of C5 (Fig. 3).

The percentages of each grain-size componentin a sample were output by the program. Thestatistics of all 40 of the surface–sediment sam-ples from Hulun Lake (Fig. 5; Table 2) show that

C48

9 A10 C1:C2:C3:

1·0 µm, 6·6 µm,

54·8 µm,

7·3%90·8%1·8%

A16 0·9 µm,4·3 µm,

31·5 µm,

C1:C2:C3:

2·8%19·5%20·8%

A2 C1:C2:C3:

0·8 µm, 4·7 µm,

16·0 µm,

1·8%16·8%6·5%

C2

C4

6

5

7 C5: 240·7 µm, 1·8%0·3%

Fitting residual: 2·4%139·5 µm,739·5 µm,

C4:C6:

44·0%13·0%

Fitting residual: 1·8%

C4: 117·9 µm, 75·0%Fitting residual: 5·5%

C1

C6

C4

C1

C2 C3C2 C3

2

4Vol

ume

(%)

1

3

B1 C1:C2:C4:

0·8 µm, 6·5 µm,

77·5 µm,

2·5%43·8%53·8%


B9 C1:C2:C3:

1·0 µm, 7·7 µm,

24·8 µm,

4·8%74·0%21·3%


B15 C1:C2:C3:

1·6 µm, 6·8 µm,

57·6 µm,

9·0%8·8%

82·3%Fitting residual: 7·6%

C3

8

9

7

C5C3C1C1

0

1

C2C4

4

6

Vol

ume

(%)

5

7

C1

C3

C1C2

C1

C22

1

3

Grain size (µm, log) Grain size (µm, log)Grain size (µm, log)0·1 1 10 100 1000 0·1 1 10 100 1000 0·1 1 10 100 1000

0

Fig. 3. Frequency distribution curves (black lines) of six representative samples from the surface sediments of HulunLake illustrating identifying, fitting and partitioning of grain-size components (blue lines) within a polymodal dis-tribution using the log-normal distribution function. Altogether, six grain-size components can be recognized on thepolymodal distributions, and are designated C1 to C6 from fine to coarse modes. The components of each sample arelabelled on each curve. The sample number, the modal size and percentage of each component and the fittingresidual of each sample are shown.



component C1 constitutes <10% in each sample.In contrast, the components C2, C3, C4 and C5make up ca 90% of some samples and ca 20% ofthe others. The C6 component accounts for 13%in sample A16.

Among all sampling sites, the water depthranges from 0Æ40 to 3Æ15 m and the offshoredistance varies from 0Æ02 km up to 13 km (Figs 1and 2; Table 1). The distribution of differentcomponents in the samples (Figs 6 and 7; Table 2)indicates that: (i) components C1 and C2 exist inall samples; (ii) component C3 is missing from sixsamples in which the C4 component clearly ispresent; (iii) component C4 occurs only in 19samples that are distributed in the zone of<3Æ050 m water depth and <5 km offshore dis-tance and, when the C4 component is present, theC3 component is either absent or assumes a modalsize of ca <32 lm; and (iv) the C5 and C6components occur only in six samples and onesample, respectively.

DISCUSSION

Interpretation of the grain-size components

Hulun Lake is located in a semi-arid area of themiddle temperate zone. There are two majorrivers entering the lake (Fig. 1). In terms of thegeographic location and the basin structure, clas-tic sediments of Hulun Lake are derived mainlyfrom fluvial materials transported by the riversduring the summer season and aeolian dusttrapped on the ice of the lake during the dryand windy spring season. In addition, shorelinematerials eroded by storm-driven waves couldalso contribute to the accumulation of the lakesediments. Clastic materials derived from thethree sources will experience reworking andsorting by the hydraulic dynamics of lake watersbefore being deposited on the lake floor (Hakan-son & Jansson, 1983; Talbot & Allen, 1996).

Previous studies suggested that the three typesof clastic sediments assume different combina-tions of specific grain-size components due to

Table 2. Characteristics of the six grain-size components recognized on the polymodal distributions of 40 samplesfrom the surface sediments of Hulun Lake.

Component

Numberofsamples

Modal size (lm) Percentage

Component descriptionMin. Max. Min. Max.

C1 40 0Æ8 2Æ3 0Æ3 9Æ3 Long-term suspensionC2 40 3Æ0 9Æ0 1Æ8 92Æ3 Offshore suspensionC3 34 8Æ8 62Æ2 0Æ5 91Æ5 Offshore suspensionC4 19 65Æ7 160Æ7 0Æ3 96Æ8 Nearshore suspensionC5 6 176Æ2 650Æ3 0Æ3 89Æ5 Nearshore saltationC6 1 739Æ5 13Æ0 Nearshore traction

15C1

10

C1C2C3C4C5C6

Num

ber

of s

ampl

es

5

00 1003010 20 40 50 60 70 80 90

Percentage (%)

Fig. 5. Frequency of the percentages of the six grain-size components, C1 to C6, in 40 samples from thesurface sediments of Hulun Lake.

30 C1

20

25

10

15 C2C3

C4

0·1 1 10 100 1000

Num

ber

of s

ampl

es

5

0

C4

C5C6

Modal size (µm, log)

Fig. 4. Frequency of the modal sizes of the six grain-size components, C1 to C6, in 40 samples from thesurface sediments of Hulun Lake.

1056 J. Xiao et al.


distinct transport mechanisms. Fluvial depositsare composed mainly of a saltation medium-sandcomponent [modal size (Ms): 200 to 400 lm] anda suspension fine-silt component (Ms: 10 to15 lm) (Middleton, 1976; Ashley, 1978; Bennett& Best, 1995). Typical loess deposits consist of ashort-suspension medium-to-coarse silt compo-nent (Ms: 16 to 32 lm) and a long-suspensionclay-to-fine silt component (Ms: 2 to 6 lm) (Pye,1987; Tsoar & Pye, 1987; Sun et al., 2002; Qinet al., 2005), whereas desert sands consist of asaltation fine-to-medium sand component (Ms:100 to 200 lm) and a suspension clay-to-fine siltcomponent (Ms: 2 to 6 lm) (Gillette et al., 1974;

Pye, 1987; Tsoar & Pye, 1987). Lake shorelinematerials are dominated by unconsolidated sandswith a low proportion of silty clay (Sly, 1978;Hakanson & Jansson, 1983). In order to under-stand the constituent components of sourcematerials of the Hulun Lake sediments, modernaeolian dust trapped on the ice in the central partof the lake and modern riverbed sands depositednearby the mouths of the Herlun and UrshenRivers were analysed for grain-size distribution.

As shown in Fig. 8, both the ice-trappedaeolian dust materials and riverbed sands arecomposed of multiple components and domi-nated by the coarsest component. The dominant

Modal size (φ)0

0·4

Percentage (%)1002 4 6 8 0 20 40 60 80

0·8

1·2

2·0

1·6

2·4

Wat

er d

epth

(m

)

3·2

2·8

C2, C3, C4, C5 C2, C3, C4, C5, C4+C5+C6

Fig. 6. Modal size and percentage of the grain-size components in 40 samples from the surface sediments of HulunLake plotted against water depth. Linear regression lines are shown. For the regression equation, correlation coef-ficient and standard error, see Table 3.



component of the ice-trapped aeolian dust dis-plays a modal size of 306 lm that is much coarserthan the modal size (100 to 200 lm) of thedominant saltation component of typical dunesands. Modern dust-storm events in the lakeregion occur mainly in spring when the northwesterly winds are strong and the lake is frozen(Xu et al., 1989). The suspension and saltationcomponents of the aeolian materials could beblown downwind because of strong winds, andthe coarse traction component would be trappedon the lake ice due to diurnal freezing andthawing of the ice surface. Consequently, thedominant component of the ice-trapped aeoliansediment has a larger modal size than that oftypical dune sands. The dominant component ofHerlun and Urshen riverbed sands shows modalsizes of 147 lm and 122 lm that are finer than themodal size (200 to 400 lm) of the dominantsaltation component of typical fluvial deposits.Both the Herlun and Urshen Rivers flow throughsand dunes and run dry during the dry years (Xuet al., 1989). The sands from the sand dunes,source materials of the fluvial deposits, appearnot to have been reworked and sorted sufficientlyby the rivers during the process of transportation,so that the dominant component of the riverbedsands displays a modal size that is finer than thatof typical fluvial deposits and corresponds withthat of typical dune sands.

The nature and distribution of clastic depositsare controlled by the dynamic condition of thetransporting media (Inman, 1949; Folk & Ward,1957; Tanner, 1964; Visher, 1969; Ashley, 1978).Fundamentally, a relationship exists between

clastic and hydraulic interactions within a lake,so that lakes can be separated into high-energyand low-energy regimes (Sly, 1989a,b). In general,the shallow water of the nearshore zone possesseshigher hydraulic energy than the deep water ofthe offshore zone, and the deposits in lakesbecome coarser with higher hydraulic energy.Taking the spatial distribution of each grain-sizecomponent within the lake, together with theseobservations, it is inferred here that the compo-nents C1, C2 and C3 to C6 of the surfacesediments of Hulun Lake represent differentsedimentary modes related to different deposi-tional processes (Table 2). Component C1 (Ms: 0Æ8to 2Æ3 lm) which exists in all samples may belongto a long-term suspension component in a fluidmedium, the transportation and deposition ofwhich depends on the intensity of turbulence.The C2 (Ms: 3 to 9 lm) and C3 (Ms: 8Æ8 to62Æ2 lm) components represent two offshore sus-pension components. The C4 (Ms: 65Æ7 to160Æ7 lm), C5 (Ms: 176Æ2 to 650Æ3 lm) and C6(Ms: 739Æ5 lm) components that occur mainly inthe nearshore samples are interpreted as near-shore suspension, saltation and traction compo-nents, respectively.

As shown in Fig. 4, the modal sizes of twoadjacent components among C2 (offshore-suspen-sion fine silt), C3 (offshore-suspension medium-to-coarse silt), C4 (nearshore-suspension finesand), C5 (nearshore-saltation medium sand)and C6 (nearshore-traction coarse sand) overlap;this implies an inherent relationship between thetwo components. The authors infer that clasticmaterials had been reworked by the hydraulic

Table 3. Relationship between the modal size and percentage of the grain-size components in 40 samples from thesurface sediments of Hulun Lake and the water depth, as well as the offshore distance.

Grain-sizecomponent

Regressionequation*

Correlationcoefficient

Standarderror (m)

Regressionequation�

Correlationcoefficient

Standarderror (km)

C2 modal size y = )0Æ39x + 5Æ07 0Æ02 0Æ9 y = )3Æ0x + 27Æ02 0Æ06 4Æ2C3 modal size y = )0Æ007x + 2Æ26 0 y = 0Æ08x + 4Æ34 0C4 modal size y = 1Æ07x ) 1Æ68 0Æ25 0Æ8 y = 4Æ97x ) 13Æ55 0Æ34 3Æ2C5 modal size y = 0Æ59x + 1Æ27 0Æ23 0Æ8 y = 4Æ49x ) 2Æ02 0Æ48 3Æ5C2 percentage y = 0Æ02x + 1Æ49 0Æ33 0Æ7 y = 0Æ07x + 1Æ35 0Æ30 3Æ6C3 percentage y = 0Æ003x + 2Æ12 0Æ01 0Æ8 y = )0Æ005x + 4Æ94 0C4 percentage y = )0Æ02x + 2Æ61 0Æ55 0Æ7 y = )0Æ06x + 5Æ48 0Æ36 3Æ2C5 percentage y = )0Æ02x + 2Æ69 0Æ70 0Æ5 y = )0Æ09x + 6Æ92 0Æ37 3Æ9C4 + C5 + C6percentage

y = )0Æ02x + 2Æ85 0Æ73 0Æ5 y = )0Æ08x + 7Æ01 0Æ52 3Æ0

In the linear regression equation, x represents modal size or percentage of the grain-size components and y representswater depth in metres or offshore distance in kilometres. For data of the grain-size components, see Figs 6 and 7.*Relationship between the modal size and percentage of grain-size components and the water depth.�Relationship between the modal size and percentage of grain-size components and the offshore distance.

1058 J. Xiao et al.


Modal size (φ)0

0

Percentage (%)1002 4 6 8 0 20 40 60 80

2

4

6

8

10

Offs

hore

dis

tanc

e (k

m)

12

C2, C3, C4, C5 C2, C3, C4, C5, C4 + C5 + C6

Fig. 7. Modal size and percentage of the grain-size components in 40 samples from the surface sediments of HulunLake plotted against offshore distance. Linear regression lines are shown. For the regression equation, correlationcoefficient and standard error, see Table 3.

8

6

Ice-trapped aeolian dust

Dominant component:

306 µm, 68·3%

Herlun riverbed sands

Dominant component:

147 µm, 76·8%

Urshen riverbed sands

Dominant component:

122 µm, 77·5%

0·1 1 10 100 10000

2

4

Vol

ume

(%)

0·1 0·11 10 100 1000 1 10 100 1000

Grain size (µm, log) Grain size (µm, log)Grain size (µm, log)

Fig. 8. Frequency distribution curves of modern aeolian dust trapped on the ice in the central part of Hulun Lakeand modern riverbed sands deposited at the mouths of the Herlun and Urshen River. The modal size and percentageof the dominant component of each sample are shown.



dynamics of lake waters during transportationwithin the lake, resulting in the transformation ofthe attribution of some particles from one com-ponent to the adjacent component. In general, thecoarse tails of finer components would bereworked to a part of the adjacent coarser com-ponent under the condition of higher hydraulicenergies (closer to the lakeshore); whereas the finetails of coarser components would be reworked toa part of the adjacent finer component under thecondition of lower hydraulic energies (closer tothe lake centre).

Relationships between the major componentsand the lake levels

The major grain-size components of the surfacesediments of Hulun Lake exhibit a spatial distri-bution within the lake that characterizes themodern hydraulic condition of the lake. Thehydraulic condition of a specific lake is closelyrelated to the water depth and size of the lake(Hakanson & Jansson, 1983; Sly, 1989a,b). Ingeneral, the hydraulic energy of lake watersintensifies in the nearshore zone. In order tounderstand implications of the grain-size compo-nents for lake levels, the relationships betweenthe modal size and percentage of the majorcomponents and the water depth (Fig. 6; Table 3),as well as offshore distance (Fig. 7; Table 3), wereanalysed.

As shown in Fig. 6 and Table 3, the modal sizesof the major components appear to have nothingto do with water depth; nevertheless, the per-centages of the major components (except com-ponent C3) in the samples are obviously related towater depth. The C2 (offshore-suspension finesilt) percentages show a positive relation withwater depth (correlation coefficient: 0Æ33),whereas the percentages of C4 (nearshore-suspen-sion fine sand), C5 (nearshore-saltation mediumsand) and C4 + C5 + C6 display negative rela-tions with water depth (correlation coefficients:0Æ55, 0Æ70 and 0Æ73, respectively). With respect tooffshore distance (Fig. 7; Table 3), both the modalsizes of C4 and C5 show negative relations withoffshore distance (correlation coefficients: 0Æ34and 0Æ48, respectively). The C2 percentages arepositively related to offshore distance (correlationcoefficient: 0Æ30), whereas the percentages of C4,C5 and C4 + C5 + C6 are negatively related tooffshore distance (correlation coefficients: 0Æ36,0Æ37 and 0Æ52, respectively). Judging from thecorrelation coefficient and the number ofthe samples, the negative correlation between

the C4 + C5 + C6 percentage and the water depthis most significant (Fig. 6; Table 3). These dataindicate that those components interpreted asindicative of nearshore environments do, indeed,constitute more of the grain-size distribution insamples collected from shallower water andmight, therefore, provide a useful means ofreconstructing water depths from, for example,sediment core data.

Application of the nearshore component–water depth model

The close relationship between the percentage ofthe nearshore components in the surface sedi-ments of Hulun Lake and the water depthprovides a new method for reconstructing thehistory of changes in the lake level during thegeological past. Figure 9 illustrates the percentageof the nearshore components of the HL06 sedi-ment core recovered in the central part of HulunLake spanning the last 11 000 cal yr (Xiao et al.,2009). The nearshore components display in-creases in the percentage during the intervals ofca 9200 to 8500 cal yr bp, 8100 to 7700 cal yr bp,6600 to 6000 cal yr bp, 4500 to 3600 cal yr bp, 3000

MAP (mm)C4 + C5 + C6 (%)200 250 300 3500 5 10 15 20 25

0

1

2

3

4

5

6

77

8

9

Age

(ca

l. ka

BP

)

10

11

Fig. 9. Percentage of the nearshore grain-size compo-nents (C4 + C5 + C6) of the HL06 sediment corerecovered in the central part of Hulun Lake spanningthe last 11 000 cal yr (Xiao et al., 2009), compared withthe mean annual precipitation (MAP, mm) recon-structed on the pollen profile of the same core (Wenet al., 2011). The chronology was derived from cali-brated ages of reservoir-effect-free radiocarbon dates(Xiao et al., 2009). Shaded bars mark the intervals oflowered lake levels and weakened regional precipita-tions during the Holocene.

1060 J. Xiao et al.


to 2600 cal yr bp and 1100 to 300 cal yr bp,denoting drops in the lake level during theseperiods. Quantitative reconstruction of the regio-nal precipitation based on the pollen profile ofthe same sediment core (Wen et al., 2011) indi-cates that the mean annual precipitation in thelake region decreased during the above episodes.The high percentages of the nearshore compo-nents in the lake sediments can be correlated wellboth in phase and in amplitude with the lowprecipitations in the lake region (Fig. 9).

Historical documents and modern observa-tions indicate that Hulun Lake shrank and thelake level fell during the low-rainfall years (Xuet al., 1989). The present data suggest that thenearshore components increased in the lakesediments when the mean annual precipitationdecreased in the lake region. Such a goodcoincidence of variations between two indepen-dent proxies does not only demonstrate thevalidity of the log-normal distribution functionmethod in fitting and partitioning polymodallake sediments but also reveals the potential ofthe major grain-size components for the researchof the palaeoenvironmental history of lakes.

CONCLUSIONS

Application of the log-normal distribution func-tion method to fitting and partitioning individual,polymodal grain-size distributions suggests thatthe modern clastic sediments of Hulun Lakecontain six distinct unimodal grain-size distribu-tions, representing six grain-size components.Each of the components retains its identity,including modal size, manner of transportationand environment of deposition, although therelative percentage varies with the hydraulicconditions throughout the lake. These compo-nents are specified from fine to coarse modes aslong-term suspension clay, offshore-suspensionfine silt and medium-to-coarse silt, and near-shore-suspension fine sand, saltation mediumsand and traction coarse sand. Grain-size compo-nents interpreted as being indicative of nearshorehydrodynamic environments contribute propor-tionately more to modern lakebed samples col-lected from shallow water. The potential value ofsuch relations for palaeoenvironmental recon-struction was demonstrated by applying the log-normal partitioning method to a sediment corefrom Hulun Lake. High percentages of the near-shore components were found to be correlatedwith low regional precipitation reconstructed

from the pollen profile of the same core. Thesuccess of the log-normal distribution functionmethod applied to partitioning the grain-sizecomponents of polymodal lake sediments provesto be a valuable approach to understanding theprocesses of transportation and deposition ofclastic materials within lakes and relating thegrain-size components to specific sedimentaryenvironments of lakes.

ACKNOWLEDGEMENTS

We thank three anonymous reviewers, DanielAriztegui and Stephen Rice for valuable com-ments and suggestions that helped to improve theearly versions of the manuscript. Special thanksare extended to Stephen Rice for careful revisionsof the final version of the manuscript. This studywas financially supported by the Ministry ofScience and Technology of China (Grant2010CB833400) and the National Natural ScienceFoundation of China (Grant 41130101).

REFERENCES

Ashley, G.M. (1978) Interpretation of polymodal sediments.

J. Geol., 86, 411–421.

Bennett, S.J. and Best, J.L. (1995) Mean flow and turbulence

structure over fixed, two-dimensional dunes: implications

for sediment transport and bedform stability. Sedimentol-

ogy, 42, 491–513.

Chinese Academy of Sciences (Compilatory Commission of

Physical Geography of China) (1984) Physical Geography ofChina: Climate. Science Press, Beijing, pp. 1–30 (in Chinese).

Folk,R.L.andWard,W.C. (1957)BrazosRiverbar:Astudyin the

significance of grain size parameters. J. Sed. Petrol., 27, 3–26.

Gillette, D.A., Blifford, D.A. and Fryear, D.W. (1974) The

influence of wind velocity on size distributions of soil wind

aerosols. J. Geophys. Res., 79, 4068–4075.

Hakanson, L. and Jansson, M. (1983) Principles of Lake Sed-

imentology. Springer, Berlin, 316 pp.

Inman, D.L. (1949) Sorting of sediments in the light of fluid

mechanics. J. Sed. Petrol., 19, 51–70.

Kranck, K., Smith, P.C. and Milligan, T.G. (1996) Grain-size

characteristics of fine-grained unflocculated sediments II:

‘multi-round’ distributions. Sedimentology, 43, 597–606.

Krumbein, W.C. (1938) Size frequency distribution of sedi-

ments and the normal phi curve. J. Sed. Petrol., 8, 84–90.

Middleton, G.V. (1976) Hydraulic interpretation of sand size

distributions. J. Geol., 84, 405–426.

Passe, T. (1997) Grain size distribution expressed as tanh-

functions. Sedimentology, 44, 1011–1014.

Prins, M.A., Postma, G. and Weltje, G.J. (2000) Controls on

terrigenous sediment supply to the Arabian Sea during the

late Quaternary: the Makran continental slope. Mar. Geol.,169, 351–371.

Pye, K. (1987) Aeolian Dust and Dust Deposits. Academic

Press, London, pp. 29–62.



Qin, X.G., Cai, B.G. and Liu, T.S. (2005) Loess record of the

aerodynamic environment in the east Asia monsoon area

since 60,000 years before present. J. Geophys. Res., 110,B01204.

Sly, P.G. (1978) Sedimentary processes in lakes. In: Lakes:

Chemistry, Geology, Physics (Ed. A. Lerman), pp. 65–89.

Springer, New York.

Sly, P.G. (1989a) Sediment dispersion: part 1, fine sediments

and significance of the silt/clay ratio. Hydrobiologia, 176/177, 99–110.

Sly, P.G. (1989b) Sediment dispersion: part 2, characterisation

by size of sand fraction and per cent mud. Hydrobiologia,

176/177, 111–124.

Sun, D.H., Bloemendal, J., Rea, D.K., Vandenberghe, J., Jiang,F.C., An, Z.S. and Su, R.X. (2002) Grain-size distribution

function of polymodal sediments in hydraulic and aeolian

environments, and numerical partitioning of the sedimen-

tary components. Sed. Geol., 152, 263–277.

Talbot, M.R. and Allen, P.A. (1996) Lakes. In: SedimentaryEnvironments: Processes, Facies and Stratigraphy (Ed. H.G.

Reading), pp. 83–124. Blackwell Publishing, Oxford.

Tanner, W.F. (1964) Modification of sediment size distribu-

tions. J. Sed. Petrol., 34, 156–164.

Tsoar, H. and Pye, K. (1987) Dust transport and the question of

desert loess formation. Sedimentology, 34, 139–153.

Visher, G.S. (1969) Grain size distributions and depositional

processes. J. Sed. Petrol., 39, 1074–1106.

Weltje, G.J. and Prins, M.A. (2003) Muddled or mixed?

Inferring palaeoclimate from size distributions of deep-sea

clastics. Sed. Geol., 162, 39–62.

Wen, R.L., Xiao, J.L., Li, Y.C., Xu, Q.H., Ma, Y.Z. and

Feng, Z.D. (2011) Pollen–climate transfer functions in-

tended for temperate eastern Asia. Global Ecol. Bioge-ogr., in review.

Xiao, J.L., Chang, Z.G., Wen, R.L., Zhai, D.Y., Itoh, S. and

Lomtatidze, Z. (2009) Holocene weak monsoon intervals

indicated by low lake levels at Hulun Lake in the

monsoonal margin region of northeastern Inner Mongolia,

China. Holocene, 19, 899–908.

Xu, Z.J., Jiang, F.Y., Zhao, H.W., Zhang, Z.B. and Sun, L.(1989) Annals of Hulun Lake. Jilin Literature and History

Publishing House, Changchun, 691 pp (in Chinese).

Zhang, J.C. and Lin, Z.G. (1985) Climate of China. Shanghai

Scientific and Technical Publishers, Shanghai, 603 pp

(in Chinese).

Manuscript received 21 November 2010; revisionaccepted 2 September 2011

1062 J. Xiao et al.


Documents

The link between grain-size components and depositional ... · The link between grain-size components and depositional processes in a ... It is widely recognized that lake sediment