A reconstruction of late Pleistocene relative sea level in the south Bohai Sea, China, based on sediment grain-size analysis

Sedimentary Geology 281 (2012) 88–100

Contents lists available at SciVerse ScienceDirect

Sedimentary Geology journal homepage: www.elsevier.com/locate/sedgeo

A reconstruction of late Pleistocene relative sea level in the south Bohai Sea, China, based on sediment grain-size analysis Liang Yi a, b, c,⁎, Hongjun Yu a, Joseph D. Ortiz c, Xingyong Xu a, Xiaoke Qiang d, Haijun Huang e, Xuefa Shi a, Chenglong Deng f a

Key Laboratory of Marine Sedimentology and Environmental Geology, First Institute of Oceanography, State Oceanic Administration, Qingdao 266061, China State Key Laboratory of Estuarine and Coastal Research, East China Normal University, Shanghai 200062, China Department of Geology, Kent State University, Kent, OH 44242, USA d State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an 710075, China e Key Laboratory of Marine Geology and Environment, Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071, China f State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing 100029, China b c

a r t i c l e

i n f o

Article history: Received 27 December 2011 Received in revised form 23 August 2012 Accepted 24 August 2012 Available online 3 September 2012 Editor: J. Knight Keywords: Sea-level reconstruction Sediment grain size Mathematical partitioning Late Pleistocene Bohai Sea (China)

a b s t r a c t Future anthropogenic sea-level rise and its impact on coastal regions is an important issue facing human civilizations. Due to the short nature of the instrumental record of sea-level change, development of proxies for sea-level change prior to the advent of instrumental records is essential to reconstruct long-term background sea-level changes on local, regional and global scales. Here, we employ numerical methods to partition sediment grain size using a combined database of marine surface and core samples, and to quantitatively reconstruct sea-level variation since the late Pleistocene in the south Bohai Sea, China. Our sea-level reconstruction indicates that relative sea-level changes in the southern Bohai Sea track global sea-level variation for the duration of the record. The results also indicate substantial regression from 70 to 30 cal kyr BP, and potentially subarial exposure from 38 to 20 cal kyr BP. Our results document the feasibility of reconstructing relative sea-level change by numerical partitioning of sediment grain size data, demonstrating the potential for future applications. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Due to the large and growing population density in low lying, coastal regions, even small changes in sea level can have substantial societal and economic impacts (IPCC, 2007). Because the instrumental record of sea-level changes is short, reconstruction of a long-term baseline record of regional changes in sea level must rely on geological proxies of sea-level change. Records of palaeo-sea level change can provide insights to help interpret future predictions of sea-level change, but only if they are quantified relative to a measurable palaeo-coastal datum. Two of the most widely used approaches for past sea-level changes are: (1) exploitation of dated geomorphologic features such as coastal sands (Mauz and Hassler, 2000; van Heteren et al., 2000; Giannini et al., 2007), salt marsh (Madsen et al., 2007), terraces (Bryant et al., 1990; Chappell et al., 1996; Barreto et al., 2002), and other coastal sediments (Zong et al., 2003; Zong, 2004); and (2) sea-level transfer functions based on faunal assemblages such as testate amoebae (Charman et al., 2002; Roe et al., 2002; Gehrels and Newman, 2004), foraminifera (Chappell and Shackleton, 1986; Horton, 1997; Edwards and Horton, ⁎ Corresponding author at: Bei-Tu-Cheng-Xi-Lu 19#, Chao-Yang District, Beijing 100029, China. E-mail addresses: [email protected], [email protected] (L. Yi). 0037-0738/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.sedgeo.2012.08.007

2000; Horton and Edwards, 2005), and diatoms (Horton, 1997; Horton et al., 2006). Regardless of the method employed, there are two key requirements that must be fulfilled to reconstruct past sea-level change: the first is to quantify the relationship between the proxy variable and the reference water depth. The second constraint is to assess the consistency of the proxy's relationship to sea level through time, and reject any approaches that fail this test (Thomas and Varekamp, 1991). This problem exists in all applications of quantitative sealevel reconstruction and is particularly important when dealing with small surface calibration data sets (Horton, 1997; Edwards and Horton, 2000; Zong et al., 2003; Horton et al., 2006). While a variety of methods has been developed to reconstruct palaeo-changes in sea level, many regions, including the Bohai Sea, China, still lack detailed relative sea-level curves extending back to the Pleistocene. For example, coral terraces are absent in the Bohai Sea, and the poor preservation of faunal assemblages makes development of a transfer function for a relative sea-level reconstruction unfeasible. In contrast, frequent alternations between transgression and regression has presumably imprinted sea-level change on the grain size distribution of Bohai Sea sediments, which varies from medium silt to coarse sand during the late Quaternary (IOCAS, 1985). Advantages of grainsize-based relative sea-level transfer function

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

approaches are that they require smaller sample sizes, allowing for replication, faster measurement and higher spatial or temporal resolution at a fraction of the cost of detail micro-palaeontological analysis. Sediment grain size varies in response to current strength and direction (Pettijohn and Ridge, 1932; Self, 1977; McCave, 1978; Nordstrom, 1981). As a result, a variety of methods has been proposed to relate grain size statistics to depositional environment (Middleton, 1976; Ashley, 1978) or sediment transport (Gao and Collins, 1991, 1992). Because marine depositional environments exhibit differences in energy in response to water depth, depth information is imprinted on sediment grain size distributions. The challenge is to partition the depth-related information from other environmental factors that can influence grain size. Several approaches have been employed to relate variations in sediment grain size to the hydrodynamic conditions that control sediment transport. Friedman and Sanders (1978) partitioned sediment into the broad Udden–Wentworth classes (i.e. clay, silt, sand, etc.). Folk and Ward (1957) and McManus (1988) calculated (either graphically or by the methods of moments) bulk grain size statistics such as the mean, standard deviation (sorting), skewness and kurtosis. Boulay et al. (2003) extracted “components” sensitive to environmental conditions based on measured relationships between mean grain size and standard deviation. Although the relationship between traditional grain size indicators and water depth may seem strong enough to develop a transfer function, the indicative meaning of these proxies can be unclear due to interactions from other environmental factors. For example, an increase in the sand fraction may indicate stronger currents, a change in sediment supply, or a change in depositional environment related to sea-level fall. The influence of multiple processes can generate poly-modal grain size spectra and lead to multicolinearity. An effective means of reconstructing sea level based on grain size must thus decompose sediment grain size spectra to partition variance contributed by various processes on the basis of either theoretical considerations or through multivariate statistical methods. Approaches that have been developed to decompose grain size spectra generated by automated particle size analyzers, include methods that fit spectra to specific distribution functions (Kranck et al., 1996a,b; Påsse, 1997; Sun, 2004; Qin et al., 2005; Yi et al., 2010). Sun et al. (2002) fit their grain size spectra to Weibull functions using parameters determined by least squares, extracting information on the relative contribution of fluvial, lacustrine, desert and aeolian loess within each sample. Xiao et al. (2009) employed a similar approach using the lognormal distribution with grain size data from Dali Lake to reconstruct Holocene lake level changes. These approaches provide the impetus for us to apply similar methods to our sediment. This paper thus has two objectives: (1) to demonstrate the feasibility of reconstructing sea level variations in regions with poor fossil preservation using a grain size based transfer function, and (2) to reconstruct a relative sea level curve for the Bohai Sea during the late Pleistocene. 2. Study area and materials 2.1. Study area The south Bohai Sea (Laizhou Bay, Fig. 1) is located within the Yi-Shu Rift (Gao et al., 1980; Zhang et al., 2003). It formed in response to subsidence during the Cenozoic (Allen et al., 1997; Hu et al., 2001; He and Wang, 2003). The period from the Neogene to the present has been marked by tectonic quiescence and stable sedimentation rates (Wu et al., 2006; Yu et al., 2008). During the Quaternary, there were four tectonic episodes around the whole basin of the Bohai Sea (Xu et al., 2005), but how intensive these were, is poorly understood.

89

Some 2000–3000 m of fluvial, lacustrine and marine sediments have been deposited in the basin (IOCAS, 1985). Depositional environments have varied mainly between delta, estuarine and tidal flat systems (Xue and Ding, 2008). During regressions, the exposed area of the south Bohai Sea was replaced by diluvial fans (Chen et al., 1991), loess or sandy dunes (Chen et al., 1991; Zhao, 1995; Yu et al., 1999), or alluvial fans (Meng et al., 1999). The average water depth of the modern Bohai Sea is 15 m (IOCAS, 1985). The present sedimentary environment of the study area is tidal to inter-tidal (IOCAS, 1985; Xue and Ding, 2008). Based on several decades of tide data from the Yangjiaogou gauging station, the study area has a mean tidal level of 1.17 m, with a mean high tide of 1.70 m, a mean low tide of 0.47 m, a highest tide level of 4.79 m, and a lowest tide level − 0.94 m (Du et al., 2008; Chen et al., 2009). 2.2. Marine surface sediment During a cruise from 23 to 29 January 2007, the Institute of Oceanology, Chinese Academy of Sciences, collected thirty-six marine surface sediment samples from various depths along the shelf and slope of Laizhou Bay offshore of the estuary of the Zimaigaou, Xiaoqinghe and Mihe Rivers (Fig. 1B). Conditions during the field investigations were calm with little wave activity (Du et al., 2008; Chen et al., 2009). The sampling distribution was devised to collect material from a broad range of hydrodynamic depositional environments within the available time constraints. At each sampling station, water depth was measured using a sounding line and calibrated by a regional tidal level. Surface sediment was collected using a grab sampling device. Two samples from the middle of the Laizhou Bay (Fig 1A: R1, suspended materials; R2, marine surface sediment; Qiao et al., 2010) and two samples from the riverbed of the Xiaoqinghe River (Fig. 1B: H7 and H8; Du et al., 2008; Chen et al., 2009) were also collected for reference. 2.3. Core sediment Core Lz908 is located onshore near the south coast of the Bohai Sea, China (37°09′N, 118°58′E; Elevation 6 m a.s.l.; Fig. 1). The core was drilled to a depth of 101.3 m below the surface with a recovery rate of 75% during the summer of 2007 by the First Institute of Oceanography, State Oceanic Administration, China. The drill site for Lz908 was submerged until the middle of the 20th century. The upper 28.2 m of sediment contains two transgressions based on facies boundaries (Yi et al., 2012a). Given that counts of marine foraminifera reach a low of ~ 1 to 3 shells/g from 31 to 33 m (Yao et al., 2010) and that foraminiferal variation may lag sea-level variation, we chose to study the upper 34 m of the core to ensure that we collected material that spanned these last two sea-level cycles within the core. The core was sampled for this study every ~ 20 cm, yielding a total of 106 samples for analysis. Age control for core Lz908 is provided by radiocarbon dating and optically stimulated luminescence (OSL) of sediment grains. Sufficient material was found to obtain four foraminifer samples from Lz908 for radiocarbon dating. All radiocarbon measurements were conducted at the Woods Hole Oceanographic Institution, USA, at the NOSAMS Accelerator Mass (AMS) Spectrometry facility. Conventional 14 C ages were converted to calendar ages with the Calib6.0 radiocarbon calibration program (Stuiver et al., 2009) using the Bohai Sea calibration dataset (Wang et al., 2004; Wang and Fan, 2005) reported here as calibrated (cal) kyr BP. For the OSL dating, pure quartz from the fine fraction (4–11 μm) was selected following the sensitivity-corrected multiple aliquot regenerative-dose protocol developed by Lu et al. (2007) to determine the equivalent dose. All measurements were performed using a Daybreak 2200 automated OSL reader at the Qingdao Institute of Marine Geology, Chinese Geological Survey. Following Prescott and

90

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

Fig. 1. Study area (A), the position of Lz908 core (▲) and marine surface samples (B). Two samples from the middle of the Laizhou Bay (R1, suspended materials; R2, marine surface sediment; Qiao et al., 2010) and two samples from the riverbed of the Xiaoqinghe River (H7 and H8; Du et al., 2008; Chen et al., 2009) are collected for reference. The shorelines of 1855 AD and 1934 AD are modified from Xue and Cheng (1989), Xue (1993), Saito et al. (2000).

Hutton (1994) and Aitken (1998), we then measured neutron activation and cosmic ray contribution in the dose rate determination, while also taking into account influences from water content and grain size. When placed on the calibrated 14C and OSL based age model, bulk sediment variations in grain size demonstrated potential modulation

in response to the Asian monsoon intensity (Wang et al., 2001, 2008; Cheng et al., 2009), and Yi et al. (2012a) thus refined the preliminary chronology of core Lz908 by tuning it to the July insolation at 65°N synchronously, in accordance with the method of Ding et al. (1994). This orbital tuning significantly improved the core's chronology (Yi et al., 2012a: Fig. 9) and is applied here for further analyses (Fig. 2F).

Fig. 2. A, profile of Lz908 core (Yi et al., 2012a). B, grain-size distribution the borehole Lz908 from software Surfer® v8 using the natural neighbor gridding method. C–E, component percentages of clay (b4 μm), silt (4–63 μm) and sand (>63 μm) in the borehole Lz908, respectively. F, timescale framework of Lz908. The absolute dating results, i.e. radiocarbon dates and OSL ages, and astronomical calibration were initially released in Yi et al. (2012a). OSL ages were then corrected using an interval regression (Yi et al., 2012c). The bold line in F represents the astronomical timescale of Lz908 core.

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

2.4. Grain-size measurement All surface sediment and core samples were pretreated with 10– 20 ml of 30% H2O2 to remove organic matter, washed with 10% HCl to remove carbonates and mollusk fragments, rinsed with deionized water, and then placed in an ultrasonic vibrator for several minutes to facilitate dispersion. Grain-size spectra of remaining terrigenous material were measured using a Malvern Mastersizer 2000 grain size analyzer in the Sedimentology Lab at Kent State University, USA. One hundred grain size classes between 0.3 and 300 μm were exported for further analysis. As initially described by Yi et al. (2012a) sediment in core Lz908 was deposited in a rhythmic pattern during two transgression cycles (Fig. 2). The age model for the core indicates that the base of the deeper transgression extends down to ~ 130 cal kyr BP. The surface samples provide a means of placing these down core grain size variations in context (Yi et al., 2012a). 3. Methodology 3.1. Mathematical partitioning of sediment grain size Polymodal grain size spectra can be assumed to result from the superposition of multiple unimodal components (Ashley, 1978), which follow some type of theoretical distribution (Kranck et al., 1996a,b; Påsse, 1997; Sun et al., 2002; Qin et al., 2005). Decomposition then becomes a matter or selecting appropriate functions and estimating parameters (Sun et al., 2002). Both Lognormal (Xiao et al., 2009) and Weibull functions (Sun et al., 2002; Xu et al., 2010; Yi et al., 2010) have been employed. Sun et al. (2002) conducted experiments to determine the most suitable functions for various sediment types. They found that the Weibull distribution was applicable to a wider range of sediment types that the log normal distribution. Here we compare results of both Lognormal and Weibull function decomposition on seven types of grain size spectra collected from the south Bohai Sea (Xu et al., 2010). As described in Ashley (1978), a poly-modal distribution can be expressed as: f ¼ p1 f 1 þ p2 f 2 þ … þ ð1−p1 −p2 −…−pn−1 Þf n

ð1Þ

where fi represents the function for component i, given i = 1 to n components, and pi is a component's percentage in the bulk sample. Within each spectrum, there are n − 1 coefficients, pi, that need to be estimated due to closure. The Weibull function has the following form: f ðx; α; βÞ ¼

α α−1 −ðβx Þα x e βα

ð2Þ

where x represents the grain size in μm, the coefficient α determines the distribution's shape (e.g. the skewness or symmetry) and β controls the position of the central tendency of the curve, the modal grain size. In general, most grain size spectra can be fitted with a small number of components. For example, a three component system using the Weibull function can be expressed as:

f ðx; α 1 ; β1 ; m1 ; α
2 ; β2 ;a1m2 ; α 3 ; β3 Þ ¼
a2 x x − − α 1 α1 −1 α α −1 2 2 β β 1 2 m1 α1 x e þ m2 α 2 x e β1 β2
a3 x α α −1 − þð1−m1 −m2 Þ α33 x 3 e β3 : β3

ð3Þ

91

The values α1 and β1 are the parameters of the distribution function for the fine-grained component, α2 and β2 represent the medium-grained component, while α3 and β3 are those for the coarse-grained component. The percentages of each component in a sub-population are given by m1, m2, (1 − m1 − m2), respectively. This approach can be generalized to include a smaller or greater number of components as are needed to best-fit the data. Using the measured grain size data in one hundred grain size classes between 0.3 μm and 300 μm, the parameters can be estimated by General Least Squares Fitting (Kleinbaum et al., 1997). Three statistical mea2 sures, the R-squared (R 2), adjusted R-squared (Radj ) and standard error of the regression (s.e.), are used to evaluate the fit. The components derived in this way are employed to form a grain-size based relative sea-level transfer function.

3.2. Quantitative reconstruction of relative sea level Here, we attempt to integrate reference water level (RWL) and sediment grain size component information to reconstruct sea-level variability since the late Pleistocene, following three steps: First, we analyze the correlation among grain size components and sea level to ensure that there is sufficient independent information with which to develop a regression; second, we employ regression techniques to develop a series of polynomial fits; and third, we examine the reliability of the various fits using calibration–verification methods to determine the optimal transfer function. To assess the bias and precision of our transfer function we used an approach similar to those employed with faunal assemblages (e.g., Edwards et al., 2004). In addition to the maximum bias and root mean squared error (RMSE), we evaluate the leverage of outliers during the modeling process, through implementation of a leave-one-out Jackknife procedure. Jackknife is a cross-validation method which resamples the data set by computing n subsets of (x1, x2, … , xn), each consisting of all of the cases except the ith deleted case (i = 1, 2, …, n), following which the max biasjack and RMSEjack are calculated for each case (Kleinbaum et al., 1997). Ideally, the surface sediment calibration data should be collected from a range of depths that spans the full range of palaeo-depth variation. The depths, RWL of the surface sediment samples, range from 0.1 to 6.0 m. However, it is not possible to know the true range of variation of sea level through time prior to application of the transfer function. In addition, the relatively small size of the calibration data set (36 samples) makes it is plausible that the calibration data set may underestimate the true range of relative sea-level variability through time. One way to assess the sensitivity of the transfer function to results that may be extrapolated beyond the range of the calibration data set is to employ a split sample, calibration–verification process (Meko and Graybill, 1995). Withholding some of the data from the calibration process provides a more effective means of assessing the true uncertainty in the transfer function. To assess these transfer functions, we conducted two calibration–verification runs in which we withheld the ten smallest, then the ten largest samples: (1) Remove the 10 smallest samples as the verification group and the rest of the samples (n = 26) as the calibration group, do the regression analysis, and using the fitted equation to predict the values of 10 removal samples. Finally, compare the estimated and actual values of 10 removal samples to assess the capacity of the transfer function in a smaller-value dataset; (2) Remove the 10 greatest samples as a verification group, and follow the same steps above to assess the capacity of the transfer function in a greater-value dataset.

92

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

We employed two statistical measures to assess the quality of the calibration–verification process: the coefficient of determination, r(v) 2, and the ‘Reduction of Error’ (RE) defined as: h  i2 
∑ X i −X
v X^ i −X^ v r ðvÞ ¼   2 
2 ∑ X i −X
v ∑ X^ i −X^ v 2

 2 ∑ X i −X^ i RE ¼ 1−  2 ∑ X −X
i

ð4Þ

ð5Þ

c

∧

where Xi and X i are the observational and estimated values of the ver∧

ification group, X v and X v are the means of the observational and estimated data in the verification group, and X c is the observational mean in the calibration group. A transfer function with minimal bias should nominally be able to estimate the X c accurately. These measures are broadly applied in palaeo-environmental reconstruction (Cook et al., 1999, 2010; Mann and Rutherford, 2002; Zhang et al., 2004; Yi et al., 2012b). The coefficient of determination, r(v) 2, provides a measure of the variance shared in common between two variables and has a range of [0, + 1] (Cook et al., 1999). RE ranges from − ∞ to + 1, with a RE = 0 being no better than the calibration period. An RE > 0 indicates reconstruction skill in excess of the nominal result (i.e. X c ), and an RE b 0 indicates less skill than the nominal result. 4. Result and analysis 4.1. Grain-size analysis 4.1.1. Hydraulic meaning of grain size To evaluate the trend in grain size as a function of water depth, we selected samples along a transect extending from the mouth of the Xiaoqinghe River into Laizhou Bay (samples H7–H8–D1–D5–R2). Samples in this transect demonstrate on offshore-fining trend as expected (Fig. 3A). At the shallowest site, H7, the coarse material is the dominant fraction, with minor contributions from finer grain sizes. At site H8, the modal size decreases, and the central peak grows broader due to the contribution of sediment from additional populations. This pattern continues offshore towards the surface sediment sample at site R2, as the grain size spectrum becomes increasingly broad. Comparison of grain size spectra between the suspended material collected at site R1 and the underlying surface sediment at site R2 demonstrates that they are similar with the exception of the coarsest material present (Fig. 3B). These results suggest that decomposition of the sediment grain size spectra should enable us to extract grain size components that can be used to reconstruct variations in sea level.

4.1.2. Partitioning of grain size spectra Xu et al. (2010) modeled the sediment grain size in core Lz908 using a Weibull function-based model. For comparison, we generated a component model based on lognormal functions to determine which of these two functions provides a better fit to the data set, and whether the data could be adequately fitted using a smaller number of components (Table 1). The comparison demonstrates that four of the seven types of sediment grain-size spectra are best-fit by the Weibull function, as can be seen from the smaller standardized residual errors and generally smaller squared sum of residual errors, while the remaining three are equally fitted by either set of functions. Accordingly, the decomposition employed here is based on the Weibull function. Furthermore, because the vast majority of variance is explained by three components fitted by the Weibull function (Eq. (3)), and because their modal sizes can be categorized into four classes (clay, fine to medium silt, very coarse silt and very fine sand), we truncated the Weibull function-based model to include only four components, designated M1, M2, M3 and M4 from finest to coarsest modes, respectively.

4.1.3. Spatial change of grain-size components Twenty-four samples (67%) contain component M1 and most of samples contain component M2. Seventeen samples (47%) contain components M3, while twenty-three samples (64%) contain component M4. According to the percentage of each component (Table 2), the M3 and M4 components are the two major sub-populations in the grain size distribution of surface sediments. Component M1, with a modal size b 4 μm in the clay size range, likely represents the suspended load in the marine environment, which settles only under calm conditions, but once settled, is difficult to move due to the cohesive properties of clay sized particles. The spatial pattern of M1 reaches its greatest concentration offshore of the mouths of the rivers that feed into Laizhou Bay, particularly in the vicinity of the Xiaoqinghe River (Fig. 4), suggesting that it is supplied by fluvial input. The components M2 and M3, are also interpreted as suspension populations, but would be capable of settling under more turbulent conditions. Components M2 and M3 have similar spatial patterns, with a shallow coastal high near the mouth of the Xiaoqinghe River, a gap in their distribution, between water depths of b~ 3 m and then an offshore increase in water deeper than ~ 3 m. The low percentage of these components in water depth of b~ 3 m is likely due to the greater mobility of the sortable silts in these shallow depths where energy should be greater. In contrast, component M4 in the sand size range is sufficiently large that it likely represents intermittent bed load transport during high-energy currents or under storm conditions. The spatial pattern for component M4 is inversely correlated with those for M2 and M3.

Fig. 3. Profile of grain-size distribution from river mouth to shallow sea (H7-H8-D1-D5-R2) (A) and comparison between suspended (R1) and marine surface (R2) samples (B). See details in text.

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

93

Table 1 Fitting experiment results of grain-size distribution function using different functions. Sample

I-1 I-2 I-3 I-4 II-5 II-6 II-7 a

a

Sediment grain size

Lognormal function

Medium–very coarse silt

Very fine sand

Weibull function

Fitting degree (%)

Standardized residual error

Squared sum of residual error

Fitting degree (%)

Standardized residual error

Squared sum of residual error

99.22 99.48 99.83 99.81 99.88 99.90 99.93

0.91 0.83 0.61 0.72 0.66 0.69 0.61

7.78 6.43 34.80 48.61 41.23 44.95 35.48

99.99 99.99 99.98 99.99 99.97 99.96 99.94

0.26 0.24 0.43 0.48 0.68 0.71 0.84

6.23 5.53 17.60 21.76 43.88 48.01 60.64

Seven types of sediment grain-size spectra are cited from Xu et al. (2010) and employed to be partitioned using lognormal function for comparison.

Component M4 is the dominant component on the shelf in water shallower than 3 m depth. Therefore, coupling that components M3 and M4 are the two dominant fractions, the indicative meanings of partitioned grain-size components can be summarized that the M3 and M4 dominant areas represent low and high hydraulic energy regimes, respectively, and the 3-m isobath marks the edge of the shelf and a hydrodynamic boundary between the M3 and the M4 components.

4.1.4. Down-core variation of partitioned components Components M3 and M4 are present in a larger fraction and greater percentage of the core samples than components M1 and M2 (Fig. 5). Ninety samples (85%) contain component M1, all samples contain M2, forty samples (38%) contain component M3, and sixty-six samples (62%) contain component M4. The modal grain size for each component generally increases as the percentage of each component changes (Fig. 5). The decomposition R 2 was greater than 0.995 for virtually all down core samples, while the most of decomposition standard error was below 1 in virtually all samples. There was a weak inverse correlation observed between the decomposition R 2 and decomposition standard error (s.e.).

4.1.5. Homogeneity test Although the modal sizes and percentages for each component are not identical for the surface and core samples, several features are consistent between them to further test the homogeneity between surface and core samples. All four components in both surface and core samples could be approximately categorized as dominated by either clay (b 4 μm), fine to medium silt (4–16 μm), coarse silt (32– 63 μm) or very fine sand component (64–125 μm), respectively. Fig. 6 shows representative grain size distributions from both surface and core samples. The data indicate that the surface and core sediments in the southwest of Laizhou Bay consist of a strongly polymodal mixture of different grain size components and that the grain size spectral shapes observed in the surface samples are similar to the core spectra.

To confirm this qualitative similarity, we performed a one-way analysis of variance (ANOVA) to compare the surface and core grain size Weibull function components. The dependent variables in the ANOVA analysis were the modal size and percentage of each component, where the single factor independent variable was the surface or core groupings. We found no significant differences between the two data sets with the exception of the percentage of the finest component M1 (Table 3). Given the similarity between the surface and core grain size data sets, we combined them to generate a single Weibull function decomposition for use in further analyses. These results are consistent with that of Yi et al. (2012a), who did not find any major changes in sedimentation patterns within the south Bohai Sea during the late Pleistocene. This is perhaps not surprising given that varimax-rotated principle component analysis (V-PCA) results from Yi et al. (2012a) show the same data structure in all of the sediments, implying that there is no significant difference between the surface and core data sets. The results also indicate that grain size variation in components extracted from sediment samples represents hydrodynamic changes (Yi et al., 2012a). It is thus reasonable here to integrate these sediment data sets together to reconstruct past sea-level changes. 4.2. Relation, calibration–verification, and reconstruction of sea-level change 4.2.1. Correlation analysis The Weibull function components partition the sediment into potentially independent modes that can be related to sedimentological processes. We conducted a correlation analysis to determine which of these components or combinations of them would be most effective at reconstructing relative sea level through time. The correlation coefficient against water depth for the modal size of each component based on an analysis of the surface samples is − 0.63, − 0.54, − 0.64 and − 0.49 for components M1, M2, M3 and M4, respectively. All correlations were significant at the p b 0.05 level. The inverse correlations indicate that the greater the water depth, the smaller the component modal size as energy levels in the depositional setting decreased.

Table 2 Characteristics of the four grain-size components recognized on the polymodal distributions from the borehole Lz908 sediments. Sample

Component Number of Modal size (β1–β4, μm) samples Range Mean ± standard deviation

Surface samples M1 M2 M3 M4 Borehole M1 samples M2 M3 M4

24 33 17 23 90 106 40 66

1.0–4.0 2.4 ± 1.1 4.6–12.5 7.3 ± 2.5 23.4–63.3 43.8 ± 13.5 67.1–132.6 100.4 ± 19.2 0.9–5.5 2.2 ± 1.1 3.3–18.4 8.2 ± 3.0 22.7–58.9 43.7 ± 11.0 64.3–145.2 100.0 ± 22.1

Percentage (m1–m4, %) Range

Mean ± standard deviation

0.5–23.3 0.7–29.0 28.5–89.5 4.7–100 0.9–20.1 2.3–43.8 45.1–90.0 66.5–96.9

4.1 ± 4.9 11.5 ± 9.9 70.9 ± 16.7 83.3 ± 27.5 6.5 ± 4.2 14.5 ± 10.2 69.9 ± 13.6 85.4 ± 7.9

Interpretation of component

Long-term suspension component in fluid and marine mediums Suspension population Suspension population Traction transport in an estuarine environment with lower energy Long-term suspension component in fluid and marine mediums Suspension population Suspension population Traction transport in an estuarine environment with lower energy

94

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

Fig. 4. Spatial changes of each component of surface sediments. M1, M2, M3 and M4 from finest to coarsest modes, respectively, are the four components partitioned from grain-size spectra by Weibull function.

Because we found a significant difference between the percentage of component M1 in the surface and core samples (p b 0.01, Table 3) and because its indicative meaning is unclear, we do not include it as part of a sea-level transfer function. Because components M3 and M4 exhibit a strong inverse correlation (Figs. 4 and 5), and because of their clear indicative meanings with respect to sea-level changes, we integrate these two as a new component, which we refer to as the combined major component (Mmc). Because there is a strong positive correlation between M2 and Mmc (r = 0.68), there would be a serious multicollinearity problem if we present both in the transfer function (Kleinbaum et al., 1997). Therefore, only Mmc will be employed to construct the sea-level transfer function. The correlation between water depth and Mmc improves considerably (r = − 0.82), which demonstrates that the hydrodynamic information captured in Mmc is sufficient to develop a transfer function to relate grain size variability to water depth.

4.2.2. Regression analysis To construct the transfer function, we set Mmc as the independent variable and the RWL as the dependent variable. We then compare various polynomial regressions using the normal least squares fitting technique, with Bootstrap resampling (Cook, 1990) for all regression analyses to optimize the coefficients and obtain the most reliable model.

(1) The first order linear regression RWL ¼ 7:06−0:092  M mc :

ð6Þ

The correlation coefficient r is 0.82, the explained variance R 2 2 is 67%, the adjusted explained variance Radj is 66%, F value is 60.39, and all parameters are significant at p b 0.01 level. The linear model Eq. (6) also demonstrates small values of RMSEjack and max biasjack (Fig. 7). There is no obvious difference between the values of RMSE and max bias, obtained during the calibration–verification process indicating that there are no large leverage effects within the data set. (2) Second order polynomial regression 2

RWL ¼ 8:44−0:127  M mc þ 0:001  Mmc :

ð7Þ

The multiple-correlation coefficient rm is 0.84, the R 2 is 71%, the 2 Radj is 69%, F value is 40.42, and all parameters are significant at 2 p b 0.01 level. In Eq. (7), the relation of RWL to Mmc and Mmc , which should be positive, is in part negative. We thus exclude Eq. (7) because the obtained parameters do not make physical sense. In addition, the inclusion of an additional term does not result in a significant increase in the explained variance relative to the linear model.

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

95

Fig. 5. Down-core variation of each component of Lz908 sediments. Mode and content of each component are displayed as solid and dash lines, respectively. The right panel showing decomposition R2 and standardized error (s.e.) is employed for helping choosing the best fit of partitioning estimations. The abbreviations are same as in Fig. 4.

(3) Three-order polynomial regression 2

RWL ¼ 7:84−0:096  Mmc þ 0:0001  M mc þ 0:0001 

3 Mmc :

ð8Þ

2 is 68%, F value is 26.21, and The rm is 0.84, the R 2 is 71%, the Radj none of the parameters in this model were significant (p > 0.30). Because Eq. (8) is not statistically significant, we only accept the linear transfer function Eq. (6) as the most appropriate model. Following verification steps, the reliability of Eq. (6) is assessed. The assessment demonstrates a large r(v) 2 and positive RE (Table 4) indicating is predictive power. Therefore, combining the regression analyses, the Eq. (6) was chosen for the sea-level reconstruction from core samples (Fig. 8).

5. Discussion In their classic study, Zhao et al. (1978) identified three transgressions in the western Bohai Sea during the late Pleistocene. Later, Zhao (1986) and Wang et al. (1986) identified the palaeo-shorelines associated with these three transgressions. To build upon these results, we have reconstructed late Pleistocene changes in relative sea level for the south Bohai Sea, using a transfer function based on information extracted from sediment grain size spectra. Our transfer function (Eq. (6)) accounts for 67% of the sea-level variance in our calibration data set. This reconstruction quantitatively extends the regional relative sea-level history to the late Pleistocene, providing a comparatively long dataset to evaluate the relationship between regional and

global sea levels (Fig. 8). Our results indicate that sea level was higher than today during most of the late Pleistocene. Measurements from all by 98 samples have positive relative sea level predictions, with the lowest observed sea-level (negative reconstruction values) approximately from 60 to 15 cal kyr BP. Calibrating RWL to relative sea level requires information regarding changes of the whole depositional basin, subsidence or uplift history, and variations in depositional process. The irregular tectonic history around the Bohai Sea during the late Quaternary (Xu et al., 2005), which likely shifted depositional centers within the whole catchment basin, makes this task difficult during that time. However, the south part of the Bohai Sea experienced tectonic quiescence and stable sedimentation (Wu et al., 2006; Yu et al., 2008). Previous studies also indicate that sediment accumulation rate was relatively stable in pre-Holocene strata (Yi et al., 2012a,c). In this context, we hypothesize that the south Bohai Sea is similar to an isolated depositional basin, in which the local water level serves approximately as an indicator of regional relative sea-level changes. Hence, in the following discussion regarding sea-level changes, we will calibrate our relative sea-level curves against global records (Lambeck and Chappell, 2001; Waelbroeck et al. 2002; Siddall et al. 2003). 5.1. Sea-level history of the southern Bohai Sea in the late Pleistocene At the beginning of the late Pleistocene, RWL in the southern Bohai Sea rose quickly to 2.9 m (relative to present values), then after dropped and remained less than 2.0 m. RWL rose up again from 110 to ~ 100 cal kyr BP, and stood at 4.4 m, the second highest level during the last 140 cal kyr BP. RWL dropped to ~ 2 m at

96

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

Fig. 6. Selected grain-size distributions and partitioning results of surface (left panel) and core (right panel) samples. The abbreviations are same as in Fig. 4.

97 cal kyr BP, and this low stand lasted 6 kyr with a slight rising at 94 cal kyr BP. The highest RWL in the reconstructed series occurred between 89 and 79 cal kyr BP with maximal values of almost 5 m and an average level greater than 4 m. Sea level fell gradually from 79 to 60 cal kyr BP, then entered a prolonged, low stand from 60 to 17 cal kyr BP, with negative values that may indicate subaerial exposure during the last glacial maximum (negative values in Fig. 8, 18– 23 cal kyr BP), although these values lie within the uncertainty of the reconstruction. From 20 cal kyr BP onward, RWL rose quickly again and formed the first transgression during the Holocene around the Bohai Sea. During the Holocene, RWL has not yet reached a value greater than during 130–70 cal kyr BP.

Wang et al. (1986) and Zhao (1986) reconstructed shorelines for the three transgressions during the late Pleistocene using data from more than 100 cores along the west coast of the Bohai Sea. They stated that the second transgression, occurring during MIS3 (57– 29 cal kyr BP), was the largest transgression of the late Pleistocene and was more expressed than other transgressions. Later, this sea-level pattern was broadly reported in sediment around the Bohai Sea (see reviews by Zhao, 1995; Wang and Tian, 1999; Liu, 2009). Meanwhile, others have reported a warmer (2–4 °C higher) and wetter (precipitation 50–100% higher) climate than the present in west China for this time period (see a review by Shi et al., 2001). However, in agreement with global sea-level curves and climate

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100 Table 3 The one-way ANOVA results of the modal size and percentage of each component between Lz908 core sediments and marine surface samples. Components

Variance

Sum of squares

Mean square F value Sig. level

M1 Content

Between groups Within groups Total Between groups Within groups Total Between groups Within groups Total Between groups Within groups Total Between groups Within groups Total Between groups Within groups Total Between groups Within groups Total Between groups Within groups Total

251 1540 1791 0.048 122 122 161 13,852 14,013 15 880 895 36 11,709 11,745 5 8408 8413 101 20,561 20,662 30 37,615 37,646

251 14

Modal size

M2 Content

Modal size

M3 Content

Modal size

M4 Content

Modal size

0.05 1

17

b0.01

0.04

0.84

161 102

1.58

0.21

15 7

2.28

0.13

36 217

0.17

0.68

5 155

0.04

0.85

101 242

0.42

0.52

30 442

0.07

0.79

97

Table 4 Calibration–verification of sea-level transfer function Eq. (6)⁎. Statistics Calibration Verification Calibration Verification Full calibration (0.1–4.0 m) (4.0–6.0 m) (0.7–6.0 m) (0.1–0.7 m) (0.1–6.0 m) r r(v)2, r2 RE

0.64 0.41 –

0.74 0.55 0.64

0.88 0.78 –

0.36 0.13 0.45

0.82 0.67 –

⁎ All regression analyses were conducted by using the normal least square fitting with the Bootstrap resampling technique.

5.2. Sea level consistency between local and global pattern

reconstructions, in our quantitative reconstruction of relative sea levels, we do not observe a high stand during MIS3, but rather a substantial regression during 70–30 cal kyr BP with low sea level, and potentially subaerially exposed land from 38 to 20 cal kyr BP. The sediment deposited during this period could be deposited as parts of diluvial fan (Chen et al., 1991), loess/sandy dune (Chen et al., 1991; Zhao, 1991, 1995; Yu et al., 1999) or alluvial fan (Meng et al., 1999). Our reconstruction is also somewhat consistent with the results of the BQ-1 core drilled on the west coast of the Bohai Bay, which inferred that “transgression” in the MIS3 was the smallest one of the late Pleistocene and might subarial exposure (Yan et al., 2006).

Changes in the relative positions of sea and land surfaces are indicative of vertical land movements, changes in ocean volume, or, in most cases, of both factors (Lambeck and Chappell, 2001). Interglacial and interstadial sea levels since the late Pleistocene have been estimated from dated terraces in New Zealand (Pillans, 1983) and Sumba (Pirazzoli et al., 1991), and the most detailed and reliable record is based on coral terraces at Huon Peninsula, Papua New Guinea by Chappell et al. (1996) and Yokoyama et al. (2001). The RWL series of the south Bohai Sea compares well with these records, although slight lead or lags are observed (Fig. 8). Moreover, similar high stands during late MIS5 were also observed in the western North Atlantic (Potter and Lambeck, 2004) and the northeastern North Carolina, USA (Parham et al., 2007). These high stands were caused by isostatic rebound following MIS6 (Muhs et al., 2002). However, although the tectonic movement around 100–80 cal kyr BP in the Bohai Sea might result in a uplift of surrounding mountains and deflation of the depositional basin (Xu et al., 2005) and thus may imply a relation between sea level fluctuation and regional tectonic activities, more evidence is needed to prove that high sea levels occurred in the early or late MIS5.

6. Conclusions By combining grain-size data from marine surface and downcore sediments in the south Bohai Sea, new insights into regional relative sea-level changes since the late Pleistocene are obtained: (1) The grain size of surface and core samples can be mathematically partitioned using the Weibull distribution into four components. These four components with differing modal sizes and percentages could be interpreted as a long-term suspension component, which only settles under low turbulence conditions, sortable silt and very fine sand components transported by suspension during greater turbulence and bedload transport component, respectively. (2) Through regression and rigorous verification techniques, the reference water level could be reconstructed from sediment grain size. The reconstruction quantitatively extends the regional relative sea-level history to the late Pleistocene, providing a comparatively long dataset to evaluate regional sea-level variability. (3) We find no evidence of a sea-level high stand during MIS3 but rather a substantial regression during 70–30 cal kyr BP and potentially exposed land during 38–20 cal kyr BP. These results for the south Bohai Sea are in good agreement with published global sea-level records for the late Pleistocene, implying similarities between local and global sea-level patterns.

Fig. 7. Observed vs. estimated reference water level (RWL) from the transfer function Eq. (6). See details in text.

Therefore, we conclude that grain-size based sea-level reconstruction provide results that are comparable to other reconstruction methods and demonstrates great potential application for future works.

98

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

Fig. 8. Relative sea levels with 95% confidence interval in the south Bohai Sea since the late Pleistocene (bottom panel) and comparison with global sea level changes (A, Lambeck and Chappell, 2001; B, Waelbroeck et al., 2002; C, Siddall et al., 2003). The negative meters below sea level indicate an exposure to land but not exact values of water depths. See details in text.

Acknowledgment

References

The authors are grateful to Prof. Jasper Knight (the editor) and one anonymous reviewer for their helpful suggestion and comments, which have improved this manuscript. We also wish to thank Dr. Yongqiang Zong in the University of Hong Kong for discussion. This research was supported by the State Oceanic Research Project for Public Benefit of China (201105020), Chinese Offshore Investigation and Assessment (908-01-ZH2) and Open Research Fund of State Key Laboratory of Estuarine and Coastal Research (SKLEC201208). C.D. was supported by the National Basic Research Program of China (2012CB821900) and the National Natural Science Foundation of China (40925012). Part of the work was completed while the lead author conducted a post-doc at the Department of Geology, Kent State University, USA.

Aitken, M.J., 1998. An Introduction to Optical Dating. Oxford University Press. 279 pp. Allen, M.B., Macdonald, D.I.M., Xun, Z., Vincent, S.J., Brouet-Menzies, C., 1997. Early Cenozoic two-phase extension and late Cenozoic thermal subsidence and inversion of the Bohai Basin, northern China. Marine and Petroleum Geology 14, 951–972. Ashley, G.M., 1978. Interpretation of polymodal sediments. Journal of Geology 86, 411–421. Barreto, A.M.F., Bezerra, F.H.R., Suguio, K., Tatumi, S.H., Yee, M., Paiva, R.P., Munita, C.S., 2002. Late Pleistocene marine terrace deposits in northeastern Brazil: sea-level change and tectonic implications. Palaeogeography, Palaeoclimatology, Palaeoecology 179, 57–69. Boulay, S., Colin, C., Trentesaux, A., Pluquet, F., Bertaux, J., Blamart, D., Buehring, B., Wang, P., 2003. Mineralogy and sedimentology of Pleistoncene sediment in the South China Sea (ODP Site 1144). In: Prell, W.L., Wang, P., Blum, P., Rea, D.K., Clemens, S.C. (Eds.), Proceedings of the Ocean Drilling Program. Scientific Results, College Station, TX (Ocean Drilling Program), pp. 1–21. Bryant, E.A., Young, R.W., Price, D.M., Short, S.A., 1990. Thermoluminescenge and uranium–thorium chronologies of pleistocene coastal landforms of the Illawarra region, New South Wales. Australian Geographer 21, 101–112.

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100 Chappell, J., Shackleton, N.J., 1986. Oxygen isotopes and sea level. Nature 324, 137–140. Chappell, J., Omura, A., Esat, T., McCulloch, M., Pandolfi, J., Ota, Y., Pillans, B., 1996. Reconciliation of late Quaternary sea levels derived from coral terraces at Huon Peninsula with deep sea oxygen isotope records. Earth and Planetary Science Letters 141, 227–236. Charman, D.J., Roe, H.M., Gehrels, R.W., 2002. Modern distribution of saltmarsh testate amoebae: regional variability of zonation and response to environmental variables. Journal of Quaternary Science 17, 387–409. Chen, M., Chen, J.G., Wang, J.F., 1991. The evolution trend of swamping, salting and desertization along the coast of Bohai Sea and its relationship to global change. Quaternary Sciences 11, 113–122 (in Chinese with English abstract). Chen, B., Huang, H.J., Mei, B., 2009. Characteristics of sediment transportation near Xiaoqing Estuary. Marine Geology & Quaternary Geology 29, 35–42 (in Chinese with English abstract). Cheng, H., Edwards, R.L., Broecker, W.S., Denton, G.H., Kong, X., Wang, Y., Zhang, R., Wang, X., 2009. Ice age terminations. Science 326, 248–252. Cook, E.R., 1990. Bootstrap confidence intervals for red spruce ring-width chronologies and an assessment of age-related bias in recent growth trends. Canadian Journal of Forest Research 20, 1326–1331. Cook, E.R., Meko, D.M., Stahle, D.W., Cleaveland, M.K., 1999. Drought reconstructions for the continental United States. Journal of Climate 12, 1145–1162. Cook, E.R., Anchukaitis, K.J., Buckley, B.M., D'Arrigo, R.D., Jacoby, G.C., Wright, W.E., 2010. Asian monsoon failure and megadrought during the last millennium. Science 328, 486–489. Ding, Z.L., Yu, Z.W., Rutter, N.W., Liu, T.S., 1994. Towards an orbital time scale for Chinese loess deposits. Quaternary Science Reviews 13, 39–70. Du, T.Q., Huang, H.J., Yan, L.W., Liu, G.W., Song, Z.J., 2008. Characteristics of suspended matters in winter water off the Xiaoqinghe River estuary. Marine Geology & Quaternary Geology 28, 41–48 (in Chinese with English abstract). Edwards, R.J., Horton, B.P., 2000. Reconstructing relative sea-level change using UK salt-marsh foraminifera. Marine Geology 169, 41–56. Edwards, R.J., van de Plassche, O., Gehrels, W.R., Wright, A.J., 2004. Assessing sea-level data from Connecticut, USA, using a foraminiferal transfer function for tide level. Marine Micropaleontology 51, 239–255. Folk, R.L., Ward, W.C., 1957. Brazos River bar: a study in the significance of grain size parameters. Journal of Sedimentary Petrology 27, 3–26. Friedman, G.M., Sanders, J.E., 1978. Principles of Sedimentology. John Wiley & Sons, New York, NY, USA. Gao, S., Collins, M.B., 1991. A critique of the “McLaren Method” for defining sedimenttransport paths. Journal of Sedimentary Petrology 61, 143–146. Gao, S., Collins, M.B., 1992. Net sediment-transport patterns inferred from grain-size trends, based upon definition of “transport vectors”. Sedimentary Geology 80, 47–60. Gao, W.M., Li, J.L., Sun, Z.Y., 1980. Formation and evolution of the Yi-Shu continental rift. Seismology and Geology 2, 11–18. Gehrels, W.R., Newman, S.W.G., 2004. Salt-marsh foraminifera in Ho Bugt, western Denmark, and their use as sea-level indicators. Geografisk Tidsskrift-Danish Journal of Geography 104, 97–106. Giannini, P.C.F., Sawakuchi, A.O., Martinho, C.T., Tatumi, S.H., 2007. Eolian depositional episodes controlled by Late Quaternary relative sea level changes on the Imbituba– Laguna coast (southern Brazil). Marine Geology 237, 143–168. He, L., Wang, J., 2003. Cenozoic thermal history of the Bohai Bay Basin: constraints from heat flow and coupled basin-mountain modeling. Physics and Chemistry of the Earth, Parts A/B/C 28, 421–429. Horton, B.P., 1997. Quantification of the Indicative Meaning of a Range of Holocene Sealevel Index Points From the Western North Sea. Unpublished Thesis, University of Durham, Durham City, UK, 509 pp. Horton, B.P., Edwards, R.J., 2005. The application of local and regional transfer functions to the reconstruction of Holocene sea levels, north Norfolk, England. The Holocene 15, 216–228. Horton, B.P., Corbett, R., Culver, S.J., Edwards, R.J., Hillier, C., 2006. Modern saltmarsh diatom distributions of the Outer Banks, North Carolina, and the development of a transfer function for high resolution reconstructions of sea level. Estuarine, Coastal and Shelf Science 69, 381–394. Hu, S., O'Sullivan, P.B., Raza, A., Kohn, B.P., 2001. Thermal history and tectonic subsidence of the Bohai Basin, northern China: a Cenozoic rifted and local pull-apart basin. Physics of The Earth and Planetary Interiors 126, 221–235. IOCAS (Institute of Oceanology, Chinese Academy of Sciences), 1985. Bohai Sea Geology. Science Press, Beijing, China. 232 pp (in Chinese). IPCC (Intergovernmental Panel on Climate Change), 2007. IPCC Fourth Assessment Report: Climate Change 2007 (AR4). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Kleinbaum, D., Kupper, L., Muller, K., Nizam, A., 1997. Applied Regression Analysis and Other Multivariable Methods. Duxbury Press, Pacific Grove, CA, USA. 816 pp. Kranck, K., Smith, P.C., Milligan, T.G., 1996a. Grain-size characteristics of fine-grained unflocculated sediments I: ‘one-round’ distributions. Sedimentology 43, 589–596. Kranck, K., Smith, P.C., Milligan, T.G., 1996b. Grain-size characteristics of fine-grained unflocculated sediments II: ‘multi-round’ distributions. Sedimentology 43, 597– 606. Lambeck, K., Chappell, J., 2001. Sea level change through the last glacial cycle. Science 292, 679–686. Liu, T., 2009. Loess and Arid Environment. Anhui Science & Technology Press, Hefei, China. 537 pp (in Chinese). Lu, Y.C., Wang, X.L., Wintle, A.G., 2007. A new OSL chronology for dust accumulation in the last 130,000 yr for the Chinese Loess Plateau. Quaternary Research 67, 152–160.

99

Madsen, A.T., Murray, A.S., Andersen, T.J., Pejrup, M., 2007. Temporal changes of accretion rates on an estuarine salt marsh during the late Holocene — reflection of local sea level changes? The Wadden Sea, Denmark. Marine Geology 242, 221–233. Mann, M.E., Rutherford, S., 2002. Climate reconstruction using ‘pseudoproxies’. Geophysical Research Letters 29, 139.1–139.4. Mauz, B., Hassler, U., 2000. Luminescence chronology of Late Pleistocene raised beaches in southern Italy: new data of relative sea-level changes. Marine Geology 170, 187–203. McCave, I.N., 1978. Grain-size trends and transport along beaches: an example from eastern England. Marine Geology 28, 43–51. McManus, J., 1988. Grain size determination and interpretation. In: Tucker, M. (Ed.), Techniques in Sedimentology. Blackwell, Oxford, UK, pp. 63–85. Meko, D.M., Graybill, D.A., 1995. Tree-ring reconstruction of upper Gila River discharge. Journal of the American Water Resource Association 31, 605–616. Meng, Q., Han, M., Zhao, M., Jiang, A., 1999. A preliminary study of the Mihe River alluvial diluvial fan and the palaeochannels. Journal of Shandong Normal University (Natural Science) 14, 46–50 (in Chinese with English abstract). Middleton, G.V., 1976. Hydraulic interpretation of sand size distribution. Journal of Geology 84, 405–426. Muhs, D.R., Simmons, K.R., Steinke, B., 2002. Timing and warmth of the Last Interglacial period: new U-series evidence from Hawaii and Bermuda and a new fossil compilation for North America. Quaternary Science Reviews 21, 1355–1383. Nordstrom, K.F., 1981. Differences in grain-size distribution with shoreline position in a spot environment. Northeast Geology 3, 252–258. Parham, P.R., Riggs, S.R., Culver, S.J., Mallinson, D.J., Wehmiller, J.F., 2007. Quaternary depositional patterns and sea-level fluctuations, northeastern North Carolina. Quaternary Research 67, 83–99. Påsse, T., 1997. Grain size distribution expressed as tanh-functions. Sedimentology 44, 1011–1014. Pettijohn, F.J., Ridge, J.D., 1932. A textural variation series of beach sands from Cedar Point, Ohio. Journal of Sedimentary Petrology 2, 76–88. Pillans, B., 1983. Upper Quaternary marine terrace chronology and deformation, South Taranaki, New Zealand. Geology 11, 292–297. Pirazzoli, P.A., Radtke, U., Hantoro, W.S., Jouannic, C., Hoang, C.T., Causse, C., Best, M.B., 1991. Quaternary raised coral-reef terraces on Sumba Island, Indonesia. Science 252, 1834–1836. Potter, E.K., Lambeck, K., 2004. Reconciliation of sea-level observations in the Western North Atlantic during the last glacial cycle. Earth and Planetary Science Letters 217, 171–181. Prescott, J.R., Hutton, J.T., 1994. Cosmic ray contributions to dose rates for luminescence and ESR dating: large depths and long-term time variations. Radiation Measurements 23, 497–500. Qiao, S., Shi, X., Zhu, A., Liu, Y., Bi, N., Fang, X., Yang, G., 2010. Distribution and transport of suspended sediments off the Yellow River (Huanghe) mouth and the nearby Bohai Sea. Estuarine, Coastal and Shelf Science 86, 337–344. Qin, X.G., Cai, B.G., Liu, T.-S., 2005. Loess record of the aerodynamic environment in the East Asia monsoon area since 60,000 years before present. Journal of Geophysical Research 110, B01204. Roe, H.M., Charman, D.J., Roland Gehrels, W., 2002. Fossil testate amoebae in coastal deposits in the UK: implications for studies of sea-level change. Journal of Quaternary Science 17, 411–429. Saito, Y., Wei, H., Zhou, Y., Nishimura, A., Sato, Y., Yokota, S., 2000. Delta progradation and chenier formation in the Huanghe (Yellow River) delta, China. Journal of Asian Earth Sciences 18, 489–497. Self, R.P., 1977. Longshore variation in beach sands, Nautla area, Veracruz, Mexico. Journal of Sedimentary Research 47, 1437–1443. Shi, Y., Yu, G., Liu, X., Li, B., Yao, T., 2001. Reconstruction of the 30–40 ka enhanced Indian monsoon climate based on geological records from the Tibetan Plateau. Palaeogeography, Palaeoclimatology, Palaeoecology 169, 69–83. Siddall, M., Rohling, E.J., Almogi-Labin, A., Hemleben, C., Meischner, D., Schmelzer, I., Smeed, D.A., 2003. Sea-level fluctuations during the last glacial cycle. Nature 423, 853–858. Stuiver, M., Reimer, P.J., Reimer, R.W., 2009. CALIB 6.1.1. http://calib.qub.ac.uk (website program). Sun, D.H., 2004. Monsoon and westerly circulation changes recorded in the late Cenozoic aeolian sequences of Northern China. Global and Planetary Change 41, 63–80. Sun, D.H., Bloemendal, J., Rea, D.K., Vandenberghe, J., Jiang, F.C., An, Z.S., Su, R.X., 2002. Grain-size distribution function of polymodal sediments in hydraulic and aeolian environments, and numerical partitioning of the sedimentary components. Sedimentary Geology 152, 263–277. Thomas, E., Varekamp, J.C., 1991. Palaeoenvironmental analysis of marsh sequences (Clifton, Connecticut): evidence for punctuated rise in relative sea-level during the Holocene. Journal of Coastal Research 11 (SI), 125–158. van Heteren, S., Huntley, D.J., Van de Plassche, O., Lubberts, R.K., 2000. Optical dating of dune sand for the study of sea-level change. Geology 28, 411–414. Waelbroeck, C., Labeyrie, L., Michel, E., Duplessy, J.C., McManus, J.F., Lambeck, K., Balbon, E., Labracherie, M., 2002. Sea-level and deep water temperature changes derived from benthic foraminifera isotopic records. Quaternary Science Reviews 21, 295–305. Wang, H., Fan, C., 2005. The 14C database (II) on the circum-Bohai Sea-Coast. Quaternary Sciences 25, 141–156 (in Chinese with English abstract). Wang, Q., Tian, G.Q., 1999. The neotectonic settings of late Quaternary transgression on the eastern coastal plain of China. Journal of Geomechanics 5, 41–48 (in Chinese with English abstract). Wang, Q., Li, F., Li, Y., Gao, X., 1986. Shoreline changes in west-southern coastal plain of the Bohai Sea since 150 ka. In: Qin, Y.S., Zhao, S.L. (Eds.), Late Quaternary Sea-Level Changes. China Ocean Press, Beijing, China, pp. 62–71.

100

L. Yi et al. / Sedimentary Geology 281 (2012) 88–100

Wang, Y., Cheng, H., Edwards, R.L., An, Z., Wu, J., Chen, C.-C., Dorale, J.A., 2001. A highresolution absolute-dated Late Pleistocene monsoon record from Hulu Cave, China. Science 294, 2345–2348. Wang, H., Li, F., Fan, C., Frechen, M., van Strydonck, M., Fei, D., Wang, Y., 2004. The 14C database (I) on the circum-Bohai Sea-Coast. Quaternary Sciences 24, 601–613 (in Chinese with English abstract). Wang, Y., Cheng, H., Edwards, R.L., Kong, X., Shao, X., Chen, S., Wu, J., Jiang, X., Wang, X., An, Z., 2008. Millennial- and orbital-scale changes in the East Asian monsoon over the past 224,000 years. Nature 451, 1090–1093. Wu, S., Yu, Z., Zou, D., Zhang, H., 2006. Structural features and Cenozoic evolution of the Tan-Lu Fault Zone in the Laizhou Bay, Bohai Sea. Marine Geology & Quaternary Geology 26, 101–110 (in Chinese with English abstract). Xiao, J.L., Chang, Z.G., Si, B., Qin, X.G., Itoh, S., Lomtatidze, Z., 2009. Partitioning of the grain-size components of Dali Lake core sediments: evidence for lake-level changes during the Holocene. Journal of Paleolimnology 42, 249–260. Xu, J., Ma, Z., Cheng, G., Long, Z., Deng, Q., Gao, X., Zhang, G., Cai, D., Zhang, J., Zhao, J., 2005. Estimating times of Quaternary tectonic episodes in the Bohai Sea based on geomorphic features of surrounding mountains areas. Quaternary Sciences 25, 700–710 (in Chinese with English abstract). Xu, X.Y., Yi, L., Yu, H.J., Li, N.S., Shi, X., 2010. The differences of grain-size parameters estimated with graphic and moment methods in coastal sediments. Acta Oceanologica Sinica 32, 80–86 (in Chinese with English abstract). Xue, C.T., 1993. Historical changes in the Yellow River delta, China. Marine Geology 113, 321–329. Xue, C.T., Cheng, G.D., 1989. Shelly ridges in west coast of Bohai Sea and Holocene Yellow River Delta system. In: Yang, Z.G., Lin, H.M. (Eds.), Quaternary Processes and Events in China Offshore and Onshore Areas. China Ocean Press, Beijing, China, pp. 117–125 (in Chinese). Xue, C., Ding, D., 2008. Weihe River–Mihe River delta in south coast of Bohai Sea, China: sedimentary sequence and architecture. Scientia Geographica Sinica 28, 672–676 (in Chinese with English abstract). Yan, Y., Wang, H., Li, F., Li, J., Zhao, C., Lin, F., 2006. Sedimentary environment and sealevel fluctuations revealed by Borehole BQ1 on the west coast of the Bohai Bay, China. Geological Bulletin of China 25, 264–279 (in Chinese with English abstract). Yao, J., Yu, H.J., Xu, X.Y., Yi, L., Su, Q., 2010. Deposition characteristics in brine aquifers and brine formation in Laizhou Bay Area. Advances in Marine Science 28, 473–477 (in Chinese with English abstract). Yi, L., Yu, H.J., Xu, X.Y., Liu, Y., Yao, J., Zhao, N., 2010. Influences of carbonate contents on the grain-size measurements of borehole sediments from southern shore of Laizhou Bay. Advances in Marine Science 28, 325–331 (in Chinese with English abstract).

Yi, L., Yu, H., Ortiz, J.D., Xu, X., Chen, S., Ge, J., Hao, Q., Yao, J., Shi, X., Peng, S., 2012a. Late Quaternary linkage of sedimentary records to three astronomical rhythms and the Asian monsoon, inferred from a coastal borehole in the south Bohai Sea, China. Palaeogeography, Palaeoclimatology, Palaeoecology 329–330, 101–117. Yi, L., Yu, H.J., Ge, J.Y., Lai, Z.P., Xu, X.Y., Qin, L., Peng, S.Z., 2012b. Reconstructions of annual summer precipitation and temperature in north-central China since 1470 AD based on drought/flood index and tree-ring records. Climatic Change 110, 101– 130. Yi, L., Lai, Z.P., Yu, H.J., Xu, X.Y., Su, Q., Yao, J., Wang, X.L., Shi, X., 2012c. Chronologies of sedimentary changes in the south Bohai Sea, China: constraints from luminescence and radiocarbon dating. http://dx.doi.org/10.1111/j.1502-3885.2012.00271.x. Boreas. Yokoyama, Y., Esat, T.M., Lambeck, K., 2001. Coupled climate and sea-level changes deduced from Huon Peninsula coral terraces of the last ice age. Earth and Planetary Science Letters 193, 579–587. Yu, H.J., Han, D.L., Liu, X., Shan, Q.M., 1999. Study on the disintegration of marine stratum in the northern shelf region of nearshore seas of China. Acta Oceanologica Sinica 21, 83–89 (in Chinese with English abstract). Yu, Z., Wu, S., Zou, D., Feng, D., Zhao, H., 2008. Seismic profiles across the middle Tan-Lu fault zone in Laizhou Bay, Bohai Sea, eastern China. Journal of Asian Earth Sciences 33, 383–394. Zhang, Y., Dong, S., Shi, W., 2003. Cretaceous deformation history of the middle Tan-Lu fault zone in Shandong Province, eastern China. Tectonophysics 363, 243–258. Zhang, Z., Mann, M.E., Cook, E.R., 2004. Alternative methods of proxy-based climate field reconstruction: application to summer drought over the conterminous United States back to AD 1700 from tree-ring data. The Holocene 14, 502–516. Zhao, S., 1986. Transgression and coastal changes in Bohai Sea and its vicinities since the Late Pleistocene. In: Qin, Y.S., Zhao, S.L. (Eds.), Late Quaternary Sea-Level Changes. China Ocean Press, Beijing, China, pp. 53–62. Zhao, S., 1991. China Shelf Sea desertization and its derived deposits during the last stage of late Pleistocene. Oceanologia et Limnologia Sinica 22, 285–293 (in Chinese). Zhao, S., 1995. Shelf Desertification. Ocean Press, Beijing, China. 194pp (in Chinese). Zhao, S., Yang, G., Cang, S., Zhang, H., Huang, Q., Xia, D., Wang, Y., Liu, F., Liu, C., 1978. Transgression's stratas and shoreline changes in the south coast of Bohai Bay, China. Oceanologia et Limnologia Sinica 9, 15–25 (in Chinese). Zong, Y., 2004. Mid-Holocene sea-level highstand along the Southeast Coast of China. Quaternary International 117, 55–67. Zong, Y., Shennan, I., Combellick, R.A., Hamilton, S.L., Rutherford, M.M., 2003. Microfossil evidence for land movements associated with the AD 1964 Alaska earthquake. The Holocene 13, 7–20.