1 INTRODUCTION

The sound velocity of marine sediments is exceedingly important in several research fields, such as, seafloor engineering, sedimentology, marine geophysics, and underwater geoacoustics (Hamilton, 1970, 1980; Hamilton and Bachman, 1982; Richardson et al., 2002). It is a significant factor in revealing the geological events of marine sedimentary environments and the effect of seismoacoustic propagation in the ocean (Kim et al., 2012). Since the 1970s, research on the relationships between sound velocity and granular properties of seafloor sediments has attracted global attention. Empirical equations describing the relationships can be used to predict sound velocity based on granular parameters, and vice versa.

The area offshore of Qingdao contains various types of seafloor sediments and is a suitable site to test in situ acoustic measurement techniques. The sediments covering the seabed of the study area are mainly composed of silty clay, clayey silt, and sandsilt-clay. The sediment distribution is largely controlled by tidal dynamics (Dong et al., 2006). Most research in this marine region has focused on the sedimentology and geomorphology (Li, 1983; Wang et al., 2000; Li et al., 2003), while direct acoustic measurements have rarely been reported in the study area. Recently, Wang et al. (2013) and Hou et al. (2014) introduced a test of the in situ seabed acoustic measurement system for this study site. The test results of the in situ measurement system indicated that this technique can perform well, and the in situ sound velocity of seafloor sediments was measured (Hou et al., 2014). However, the regression relationships between the in situ sound velocity and the granular properties have not been developed in this area.

The purpose of this study was to analyze the relationships between the sound velocity and the granular properties of seafloor sediments in in the areas offshore of Qingdao. The distribution pattern of the sound velocity was also studied. The predicted sound velocity, using empirical equations suggested by Hamilton, Anderson, and Liu, was compared with the measured sound velocity, and the applicability of previous empirical regressions was obtained from the results of the comparison.

2 MATERIAL AND METHOD 2.1 Study area

The study area includes Jiaozhou Bay and the sea adjacent to Qingdao (Fig. 1). Jiaozhou Bay is a semienclosed bay located in the southwest of Shandong Peninsula and adjacent to the Yellow Sea with an average depth of 7 m. The main part of Jiaozhou Bay is very shallow, averaging less than 5 m deep, with a maximum depth of about 70 m in a narrow channel close to the mouth directly affected by strong tidal currents with a maximum velocity of 150 cm/s (Zhang et al., 2006).

River transportation is the major source of terrigenous sediment (Li, 1983). The Dagu River, as the leading river in the north of Jiaozhou Bay, discharges 959 200 tons of suspended sediments annually into the bay. The reciprocating tidal current also contributes to transportation and redistribution of the sediments. A semi-diurnal tide is typical with a range of 4.8 m and a mean level of 2.8 m (Compilation Committee of "Records of Bays in China", 1993). The study area is characterized by various geomorphic units, including alluvial delta, aggradational plain, subaqueous sand ridge, offshore slope, accumulation and erosion valleys, and coastal valleys.

2.2 Data collection

To obtain the acoustic parameters, we mainly took two types of measurements: laboratory sampling and in situ measurements. The laboratory measurement of sediment acoustic properties is often different from the in situ value because of several factors, such as, temperature changes, pressure reduction, decrease of sediment rigidity, and mechanical porosity rebound (Hamilton, 1976). An in situ measurement obtains the wave transmission data directly by inserting wave transducers into the seafloor sediments. Therefore, the acoustic parameters obtained by in situ measurement are more accurate than the sampling data because the environmental conditions are kept intact. In this study, the acoustic parameters of seafloor sediments were all obtained by in situ measurement.

The water depth of the measuring stations offshore of Qingdao ranged from 2 to 43 m. The sound velocities were measured at 45 stations (Fig. 1) in 2009, 2011, 2012, and 2013 with the in situ seabed acoustic measurement system (Fig. 2). The system consists of two parts: the deck control unit and the underwater measurement unit. The underwater measurement unit emits sonic waves that propagate through the seafloor sediment, receives the returning signals, and transmits them to the deck control unit for waveform display and analysis (Hou et al., 2014). The mean operations during the test included selecting a reasonable sample interval and length, and setting suitable gain parameters based on the waveform. Then, the sound velocities were obtained after data analysis and processing.

The sediment samples were collected using a spade box system after the in situ measurement. Grain size compositions of the samples were analyzed with a Cilas-940L Laser Grain-size Analyzer, at the Institute of Oceanology, Chinese Academy of Sciences (IOCAS). The analysis results were at a 1/4 Φ interval, and the grain size parameters were calculated using moment methods. Next, the mean grain size, sorting coefficient, and grain fraction were obtained.

3 RESULT 3.1 Relationship between sound velocity and granular properties

The correlations between the in situ sound velocity and the measured granular properties of the sediments are presented in plots to illustrate the ranges and scatter of the data. The empirical regression equations of the illustrated data were established using the statistical fitting method.

3.1.1 Mean grain size

The function of the mean grain size is given in Fig. 3. The curve shows that when the velocity varies from 1 450 m/s to 1 700 m/s, the mean grain size is in the range of 3-7.5 Φ. The velocity decreases according to a nonlinear trend with increasing mean grain size, Mz, and can be best fitted by the quadratic equation V_p =2054.4-151.43Mz+10.205Mz² with a squared correlation coefficient R²=0.85. The sediment with 3-6 Φ shows a rapid velocity reduction, whereas sediment with a grain size above 6 Φ has low gradient variation.

3.1.2 Sand content

The sand content of the sediments varied from 0 to 80% and most were located at 0-50%. As we can see from the left side of the plot, the sediment sand content of several stations approximates to zero. The data are slightly scattered, but the increasing trend is evident. A reasonable correlation between sound velocity and sand content of sediments was established, as shown in Fig. 4. The curve reveals that the sound velocity increases in a linear fashion with increasing sand content. An equation was established to best fit the measured data as V _p=1489.4+1.9703S with a squared correlation coefficient, R²=0.76.

3.1.3 Clay content

The clay content-velocity relationship of surface sediment (Fig. 5) shows a reverse pattern to that of sand content-velocity, where the sound velocity decreases in a nonlinear model with increasing clay content. The clay content of the sediment was below 40%, and mainly ranged between 13%-32%. The curve demonstrates a favorable relativity and the correlation can be best described as V_p=1709.6-13.165C+0.2001C² with a squared correlation coefficient, R²=0.76.

3.1.4 Sand-clay ratio

The study of the relationship between sound velocity and the sand-clay ratio is relatively rare and seldom reported. In this study, a positive correlation was found based on Qingdao offshore experimental data (Fig. 6). The curve represents a squared correlation coefficient of 0.75 and can be best fitted by the equation V_p=1461.1+16.188SSC (SSC represents the sand-clay ratio; the specific value of total content of sand and silt to clay content). Sound velocity increased with the sand-clay ratio, which ranged from 2 to 16 with most values between 2 and 8.

3.1.5 Sorting coefficient

The sorting coefficient is used to describe the distribution of grain sizes in a sample. A sorting index equal to the square root of the ratio of the larger quartile (the diameter with 25% of the cumulative size-frequency distribution greater than itself) to the smaller quartile (the diameter with 75% of the cumulative size-frequency distribution greater than itself). The sorting is deemed to be fine when the coefficient is close to 1. As shown in Fig. 7, the correlation was fitted by the equation as V_p =1361.22+83.46So with a low squared correlation coefficient, R²=0.38. The sorting coefficient was in the range of 1.5 to 2.9, which indicates a poor sorting of the sediment. When the sorting coefficient ranged from 1.5 to 2.4, the sound velocity showed an obvious increase with the sorting coefficient. However, when the sorting coefficient was above 2.4, which means a poor sorting coefficient and a complex sediment composition, the data were scattered and without a clear trend.

3.2 Sound velocity distribution characteristics

The sound velocity distribution of surface sediment (Fig. 8) revealed that, from north to south, the velocity changed from 1 520-1 570 m/s in Jiaozhou Bay to 1 600-1 670 m/s on the Qingdao coast. It then decreased below 1 500 m/s on the central southern side and the Huangdao coast, and increased above 1 600 m/s in the Lingshan Island area. On the whole, the central part and the area near Huangdao showed a broad range of low velocities, whereas the Qingdao coast, located in the northeast, was characterized by high velocity.

In this study, the mean grain size showed a suitable variation with the sound velocity, as suggested in the distribution maps. In Fig. 9, still from north to south, the mean grain sizes were 5-6 Φ in Jiaozhou Bay, 3-5 Φ on the Qingdao coast, 6-8 Φ on the central southern side and Huangdao coast, and 3-4 Φ in the Lingshan Island area. In general, the central and western areas showed a broad range of fine grains, whereas the coastal area near Qingdao had coarse grains.

Hamilton's equations (Hamilton, 1970, 1974; Hamilton and Bachman, 1982; Bachman, 1985) have been widely used to estimate the sound velocity of seafloor sediment when measurements are not available (Bachman, 1989). Empirical equations for various environments have been reported between the granular properties and mean grain size.

Hamilton (1970) established the quadratic equation between velocity and grain sizes of 1-9 Φ.

Anderson (1974) established an equation for water depths less than 1 500 m.

Liu et al. (2013) established a quadratic equation for the South Yellow Sea sediment based on in situ measurements.

Sediment sound velocities were calculated using these empirical equations and the sediment grain size was obtained by the sampling measurement in this study. Figure 10 shows the velocity distribution characteristics of the measured velocity and the predicted value calculated using empirical equations.

As shown in Fig. 10, there is a consistency between the predicted velocity distribution characteristics and the measured values. The three predicted value distributions had similar high and low velocity areas. However, there was a major difference among the velocity ranges, which were 1 540-1 713 m/s (a), 1 463-1 574 m/s (b), 1 481-1 651 m/s (c), and 1 430-1 659 m/s (d). The velocity obtained using Hamilton's empirical equation (Fig. 10a) was higher than that calculated using Anderson's (Fig. 10c) and Liu's (Fig. 10b), and the in situ measured value (Fig. 10d). The velocity calculated by Liu's (Fig. 10b) empirical equation was lower than that of Anderson's (Fig. 10c) and the measured value (Fig. 10d). The velocity calculated by Anderson's empirical equation (Fig. 10c) had a similar range with the measured value (Fig. 10d).

4 DISCUSSION

The geoacoustic properties of solid mineral grains and pore fluid determine the acoustic properties of sediments because the two provide the transfer path for the sound wave. Hamilton (1970) and Anderson (1974) analyzed the relationship between sound velocity and porosity. A generally accepted view is that porosity has the best correlation with sound velocity, where the sound velocity decreases with increasing porosity. Sediment grain size and composition are two crucial factors that reflect porosity. In general, porosity of well-sorted sandy sediments increases with grain size (Lu and Liu, 2008). However, when considering sediment with discrete grain size and poor sorting, the correlation is no longer simple because of the effects of grain shape, arrangement, and other factors. The grain size of the sediment samples remains constant regardless of whether they are on the seabed or in the laboratory. Therefore, the relationship between sound velocity and mean grain size may allow grain size to be used as a proxy parameter for estimating the sound velocity of seafloor sediments.

High sand and silt content results in a tight structure, with coarse particles forming the frame and fine clay grains acting as the filler, which leads to high sound velocity. Conversely, when sand and silt content is low, the pores among the fine grains will be mostly filled by seawater. This will cause large porosity and looseness of structure, which leads to low sound velocity. Sound velocity decreases with clay content in the range of 0-30% and then remains at a steady value. Due to the relationship between clay content and porosity, for mixed sediments (sand and silt-clay mixtures), there will be no obvious variation of porosity when the clay content is above 30%, which causes the stabilization of sound velocity.

Many previous studies have focused on the relationship between sound velocity and grain size, sand, and clay content (Hamilton, 1970; Orsi and Dunn, 1991; Liu et al., 2013). However, the sorting coefficient, as an important indicator of sound velocity, has been ignored. The sorting coefficient was above 1.5 in the study area, which represents a complex sediment component. When the sorting coefficient was between 1.5-2.4, the sound velocity increased with the sorting coefficient. This indicates that sediment has a high sound velocity when there is a high content of coarse particles, a tight structure with coarse particles forming the frame, and fine grains as filler. When the sorting coefficient was above 2.4, the complex sediment component with an uncertain mean grain size, and sand and clay content could not result in a regular sound velocity variation. This means that other sediment parameters may have a greater influence on the sound velocity than the sorting coefficient. Conversely, the disperse data may also be caused by a measuring error. This will require further work to analyze the relationship between sound velocity and the sorting coefficient based on more reliable testing data.

Lu et al. (1998) and Hamilton (1970) suggested that high sand content leads to an increase of sound velocity. Hamilton (1970) and Kim and Kim (2001) pointed out that the increasing clay content can reduce sound velocity. The results of this study are consistent with these previous relationships between sound velocity and granular properties. The numerical span of seafloor sediment properties in Jiaozhou Bay resulted in a wide range of sound velocity, although the sites are close together.

The sediment covering the inner bay is mainly composed of silt. There is also a small amount of relatively coarse material, including grits, gravels, and shell fragments, caused by seafloor erosion. Therefore, this area shows relatively coarse grains and poor sorting. A sandy shoal occurs outside the mouth of the bay where the sediment has a mean component of coastal erosion products, such as grit, silk and shell debris, with coarse grain and a small amount of gravel in the trough. The seabed to the southwest of the study area near Lingshan Island is predominantly covered by residual sediments mainly composed of silt and fine sand. The three abovementioned areas are high velocity zones, where the physical parameter distribution of sediment is characterized by coarse grain, high sand content, low clay content, poor sorting, and a high sand-mud ratio. However, the central and western parts of the study area are an accumulational plain, primarily with siltclay and clay-silt deposits, characterized by fine grain size and increased clay content. This area is a low velocity zone with fine grain, low sand content, high clay content, a poor sorting coefficient, and a low sand-mud ratio. Based on the above analysis, there is a correlation between the distribution pattern of velocity and the sedimentary environment.

Because different sampling equipment was used in each study, discrepancies are inevitable in the empirical equations and research results. Thus, the empirical relationships can only serve as references for other marine areas. The differences between the predicted and measured sound velocity can be explained as follows: the velocity of sand sediment has a frequency dispersion below 8.9%, as discovered by Buckingham and Richardson (2002) and Gorgas et al. (2002) using the in situ measuring experiment. Hamilton's empirical equation (14a) has been used as the sampling measurement with a 200-kHz geoacoustic frequency based on areas such as, the North Pacific basin, the Bering Sea, the North Sea, and the Mediterranean Sea. Environmental factors, including temperature, water depth, and the differences in the measuring method and parameters are the main issues causing the empirical equation to show discrepancies with other studies. Liu carried out measurements on the mud areas of the south Yellow Sea using the hydraulic-driven self-contained in situ sediment acoustic measurement system (HSISAMS) with a frequency of 30 kHz. This produced a smaller velocity calculated by the empirical equation. Anderson's empirical equation is based on measuring data from 82 sediment cores and thousands of test samples in the global marine areas, which has produced a broadly representative result. As a result, Anderson's empirical equation (1974) is more applicable to predict the sound velocity of Qingdao offshore shallow sediments. Therefore, actual data in large quantities is essential to establish global empirical equations.

5 CONCLUSION

The sound velocity and granular properties of the marine sediments offshore of Qingdao were studied based on an in situ method. The correlations between granular properties (mean grain size, sand content, clay content, sand-clay ratio, and sorting coefficient) and sound velocity were established. The measured velocity distribution and the comparison with the predicted velocity values were analyzed. The following conclusions can be drawn from the results:

1) the sediments in the study area, the Qingdao offshore area, include various types, which leads to wide ranges of mean grain size, sand and clay content, sand-clay ratio, and sorting coefficient, and results in a great variation of sound velocity. The relationships established in this paper can be used to predict sound velocity in the area offshore of Qingdao from measured granular parameters, and vice versa;

2) porosity is a dominant parameter that affects sediment sound velocity. The effects of granular properties on sound velocity are attributed to the effects of granular properties in reducing and improving porosity;

3) the sound velocity is closely related to the sedimentary environment. Different measurement methods result in different sound velocity ranges. To establish global empirical equations, it is essential to obtain plenty of actual data using uniform measurement methods;

4) future research includes the comparison of in situ and laboratory measurement data, and the calibration of laboratory acoustic measurement with an in situ technique.

6 ACKNOWLEDGEMENT

We thank the crew for their help in conducting the in situ experiments during the cruises in 2009, 2011, 2012, and 2013.