Sound Velocity in a Thin Shallowly Submerged Terrestrial-Marine Quaternary Succession (Northern Adriatic Sea)

Estimating sound velocity in seabed sediment of shallow near-shore areas submerged after the Last Glacial Maximum is often difficult due to the heterogeneous sedimentary composition resulting from sea-level changes affecting the sedimentary environments. The complex sedimentary architecture and heterogeneity greatly impact lateral and horizontal velocity variations. Existing sound velocity studies are mainly focused on the surficial parts of the seabed sediments, whereas the deeper and often more heterogeneous sections are usually neglected. We present an example of a submerged alluvial plain in the northern Adriatic where we were able to investigate the entire Quaternary sedimentary succession from the seafloor down to the sediment base on the bedrock. We used an extensive dataset of vintage borehole litho-sedimentological descriptions covering the entire thickness of the Quaternary sedimentary succession. We correlated the dataset with sub-bottom sonar profiles in order to determine the average sound velocities through various sediment types. The sound velocities of clay-dominated successions average around 1530 m/s, while the values of silt-dominated successions extend between 1550 and 1590 m/s. The maximum sound velocity of approximately 1730 m/s was determined at a location containing sandy sediment, while the minimum sound velocity of approximately 1250 m/s was calculated for gas-charged sediments. We show that, in shallow areas with thin Quaternary successions, the main factor influencing average sound velocity is the predominant sediment type (i.e. grain size), whereas the overburden influence is negligible. Where present in the sedimentary column, gas substantially reduces sound velocity. Our work provides a reference for sound velocities in submerged, thin (less than 20 m thick), terrestrial-marine Quaternary successions located in shallow (a few tens of meters deep) near-shore settings, which represent a large part of the present-day coastal environments.


Introduction
In geophysical (acoustic/seismic) investigations of the subsurface, velocity modeling is essential for converting two-way travel time of the observed reflections into the depth domain. Velocity data are routinely extracted from multi-channel seismic data [1]; however, in many circumstances, particularly in shallow near-shore settings, obtaining offshore multi-channel data is not feasible. Restricted navigation, legal constraints, busy marine traffic, relatively low resolution of the acquired data and the surveying cost itself often make acquisition and maneuvering with streamers and seismic sources impractical or even impossible. In such settings, high-resolution single-channel seismic and acoustic surveys provide a common alternative, but the velocity data must then be obtained by other means, such as in situ measurements, laboratory core logging, and geo-acoustic modeling [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. Due to costly offshore core drilling, these approaches are mostly focused on the upper few meters of the seafloor sediment. Consequently, the acoustic properties of surficial seafloor sediments have been well known for some time [4,5,11,15,19], but sound velocity in thicker (more than 10 m) sedimentary sequences has rarely been investigated (e.g., [17]). Therefore, when velocity data for depth conversion of single-channel seismic or acoustic data are not acquired during surveying, a velocity value corresponding to the surficial sediment grain size (e.g., [19]) or a previously published value from a nearby location is usually used. Whereas this approach is sufficient for geophysical surveys of uniform sedimentary layers, it produces significant uncertainties when dealing with pronounced lateral and vertical variability in sediment composition and architecture.
Typical example of such complex settings are shallow continental shelf areas drowned during the global sea-level rise that followed the Last Glacial Maximum (LGM) lowstand when the global mean sea level was approximately at −130 m [21,22]. During transgression, earlier terrestrial depositional environments (e.g., alluvial plains) were overlain by terrestrial and marine-derived sediments deposited in fluvial, estuarine, and open marine settings, resulting in complex sedimentary architecture and highly variable sedimentary types [22][23][24], which markedly affect the sub-bottom propagation of acoustic waves [1]. Precise mapping of the 2D and 3D geometries of the sedimentary bodies in these near-shore shallow environments and their appropriate time-to-depth conversion are essential for reliable interpretation of high-resolution acoustic and seismic surveys, not only for unraveling their depth and depositional history, but also for geotechnical site assessments in various engineering projects.
During the LGM lowstand, vast areas of the presently submerged continental shelves were exposed (e.g., Figure 1a) and amounted to approximately 40% of additional landmass in Europe and 5% globally [25,26]. Therefore, the shallow-most near-shore parts of presently submerged continental shelves extend over a considerable area globally and represent an important and often poorly studied geological environment. Due to the steadily increasing interest of the geological and archaeological research communities in shallow, presently submerged, and often buried landscapes [27][28][29][30][31][32][33][34][35][36][37][38][39][40][41], accurate depth conversion is crucial for future geological studies and paleoenvironmental reconstructions in such settings. We investigate an example of a transgressed and submerged alluvial plain in the Bay of Koper (Gulf of Trieste, northern Adriatic Sea; Figure 1) to provide sound velocity values for thin (up to 20 m thick) Quaternary sediments deposited in terrestrial-marine sedimentary environments located in shallow near-shore environments a few tens of meters deep. Our work ranks among the few studies that are not limited only to the surficial seafloor sediments, but also include the entire sedimentary succession from the seafloor to the base of the sediment on the bedrock.
The Bay of Koper is located in the southeastern part of the Gulf of Trieste, which represents the northeasternmost extension of the Adriatic Sea ( Figure 1). The seabed morphology of the Bay is smooth with depths ranging up to 20 m in the open part ( Figure 1b). The main fluvial source draining into the Bay is the Rižana river with its mouth located in the reclaimed eastern part of the Bay within the Port of Koper complex. The smaller Badaševica stream is located west of the city of Koper.
The hinterland of the Bay of Koper is composed of Eocene turbidites (flysch) comprising interbedded sandstones and marlstones, which are overlain by Quaternary alluvial and paralic sediments in the valleys of Rižana and Badaševica (Figure 1b; [65,66]). Offshore the Eocene succession is unconformably overlain by Quaternary terrestrial and paralic sediments topped by Holocene marine sediments [55,56,59,62,[67][68][69][70][71]. The Quaternary succession in the Bay of Koper, which was recognized as a submerged fluvial valley of the Rižana river (Figure 1b, [68][69][70][71]), is up to a few tens of meters thick and is composed of a lower alluvial part and an upper paralic part; however, alternations of terrestrial and paralic sedimentary environments have also been observed [68,72]. The alluvial sediments are generally composed of fine-grained clastic sediments with occasional gravelly and sandy horizons, whereas the paralic sediments are mostly composed of silty clay [68][69][70]. The Holocene marine cover comprises fine-grained bioclastic sediment with the surficial sediments showing a clear zonation: sandy silt near the coastline, clayey silt in the central part, and silt in the outer part of the Bay [62,[68][69][70][71]. Holocene sedimentation rates in the Gulf of Trieste are relatively low and amount to a few millimeters per year [44,57,69,73]. Repeated multibeam bathymetric surveys in the Bay of Koper do not show significant changes in the seafloor morphology [74] and therefore imply a low-energy sedimentary environment. Figure 1. Geographical location of the study area. (a) Regional map of the northern Adriatic, which was entirely subaerially exposed during the Last Glacial Maximum (LGM). Bathymetry data from [75]. (b) Geological map of the hinterland of the Bay of Koper. Two red rectangles in the Bay of Koper mark the areas investigated in this study. The red rectangle in the Bay of Muggia marks the study area of [76]. Bathymetric data are simplified after [77] and [78]. Geological data are simplified after [65,66,[79][80][81][82]].

Materials and Methods
We used archive geotechnical reports from the borehole database of the Geological Survey of Slovenia. Boreholes were mainly located in the NE part of the Bay ( Figure 2) and were drilled in the late 1980s and early 1990s for geotechnical investigations supporting various infrastructure projects of the Port of Koper. Borehole metadata and descriptions are provided in Table 1 and Figure 5. The boreholes were drilled with rotary drilling and were cored. Sediment core samples were used for geomechanical testing and were not preserved. We therefore reconstructed the borehole sedimentary logs ( Figure 5) from borehole descriptions contained in the geotechnical reports.
Sub-bottom sonar profiles were acquired in June 2016 on board vessel Lyra with the Innomar SES-2000 Compact sub-bottom sonar. Profile transects were designed to directly cross the borehole locations. Navigation north of the second pier was obstructed by a containment boom and very shallow water depths near the coastline. For this reason, some of the sub-bottom profiles are located at some distance from the borehole locations (Table 1 and Figure 2b). We used a transmitter frequency of 8 kHz. A total of 7 sub-bottom sonar profiles were acquired (Figure 2). At shallow depths, the seafloor was also observed visually to distinguish between sedimentary, rocky, and seagrass-covered seabed. The sub-bottom profiles were visualized and interpreted in the IHS Markit Kingdom software (Version 2018, IHS Markit, London, UK). Two-way travel times (TWT s ; in milliseconds) of the seafloor (sf TWT ) and the top of the weathered/compact bedrock (d TWT ) were determined at borehole locations (Table 2).      The thickness of Quaternary sediments in the borehole (th b , in meters) was derived from the geotechnical reports and was calculated as the depth from the top of the core to the top of the bedrock represented by weathered or compact Eocene turbidites (see Figure 5). The thickness of Quaternary sediments from the sub-bottom profiles (th TWT , in miliseconds) was obtained by subtracting the TWTs from the top of the bedrock (d TWT , in miliseconds) and the seafloor (sf TWT , in miliseconds) at the borehole location (see Figures 3 and 4; Table 2). When the profiles did not directly overlie the borehole (Figure 2b and Table 1), the part of the profile closest to the borehole was used to determine th TWT (Figure 4a,d). The average sound velocity in Quaternary sediments at the borehole location (v, in meters per second; Table 2) was calculated using the following formula:

Boreholes
Borehole logs are provided in Figure 5. The Quaternary succession (including the Holocene marine sediment) is composed of fine-grained clastics with occasional gravelly horizons. Only core A-III-8/90 contains sandy horizons. Soil-rich and peat horizons are present in some of the boreholes. Horizons with bivalves and/or gastropods occur in all boreholes; however, no remarks on the species or their environment are provided in the borehole geotechnical reports. Due to the lack of detailed descriptions of the boreholes, we did not attempt to interpret the sedimentary environments. However, it is clear that the Quaternary sediments comprise terrestrial-marine deposits. The bottom parts of all the boreholes consist of weathered and/or compact bedrock built of Eocene interbedded sandstones and marlstones.

Sub-Bottom Sonar Profiles
The sub-bottom profiles with superimposed boreholes are shown in Figures 3 and 4. Boreholes more than 5 m away from the nearest sub-bottom profile were projected orthogonally to the profile (Table 1, Figure 4a,d). The seafloor is marked by the first strong sub-horizontal reflection and is indicated by a blue arrow. The Quaternary sequence is indicated by a light blue overlay. Quaternary sediments are seen as (1) acoustically transparent units (Figures 3 and 4), (2) units containing onlapping or concordant reflection geometries (Figures 3 and 4), and (3) units with sigmoidal (prograding) reflection configurations (Figures 3a and 4b,c). Eocene bedrock is indicated by a light orange overlay. The often undulating unconformity at the top of the bedrock is expressed as (1) a medium-to-high amplitude reflection under which deeper reflections are not observed (Figure 3) or (2) an up to 5 ms TWT thick medium-to-high amplitude reflection unit with a well-definable top from which downward short shallow-dipping reflections emerge ( Figure 4). The acoustic record does not discriminate between weathered or compact bedrock.
In addition to the two described units, the sub-bottom profiles contain several other features. Columnar-shaped reflections (gas flares) within the seawater column are located above the seafloor and are indicated by a light red overlay in Figure 3b,c and Figure 4c. Rough seafloor morphologies with plentiful diffraction hyperbolas are observed above the dredged areas (Figure 2a,b, Figures 3c and 4a,c), which were excavated to accommodate ships with larger drafts in the Port of Koper. Slightly rougher seafloor morphologies are also produced by seagrass (most commonly Posidonia sp.) meadows (marked by a light green overlay in Figures 3a and 4d), which were visually recognized during sub-bottom sonar acquisition. Significant, yet variable reflection degradation is seen directly beneath the areas covered by seagrass. Within the Quaternary sedimentary column, peat layers produce medium-to-high amplitude, sub-horizontal, 1-2 ms TWT thick reflections (Figures 3b and 4b). All sonar profiles show many small diffraction hyperbolas scattered between 25 and 15 ms TWTs within the sedimentary column (Figures 3 and 4). Seafloor multiples appear on the majority of the profiles (Figure 3a,b and Figure 4a-c) due to the shallow acquisition depths. Figure 3a also contains a multiple of the top of the bedrock.

Average Sound Velocity in Quaternary Sediments
Thicknesses of Quaternary sediments from drilling reports and geophysical data along with the calculated average sound velocities at borehole locations are provided in Table 2. The mean, median, and standard deviation for the whole dataset are 1523.6 m/s, 1539.7 m/s, and 144.0, respectively.

Sound Velocity Variation
The sound velocity variation in the presented dataset is significant (Section 3.3). Some of the calculated velocity values can be considered less reliable due to significant distance between the respective borehole and its closest sub-bottom sonar profile ( Table 2). If we omit the two boreholes that are separated by more than 20 m from their nearest sub-bottom profile (A-III-4/88 and A-III-10/90), the mean, median, and standard deviation for the dataset become 1556.4 m/s, 1569.2 m/s, and 143.1, respectively. Since the standard deviation remains relatively high, we discuss the principal influences on the sound velocity scattering below.

Influence of Overburden
Sound velocity in sediments strongly depends on porosity, which is in turn influenced by overburden and compaction [1,4]. Here, the overburden comprises the combined weight of the water and sedimentary columns at the borehole locations. In Figure 6, we plot the calculated velocity against the thickness of Quaternary sediments taken from borehole logs and sonar profiles (a proxy for the weight of the sedimentary column), the depth of the seafloor taken from sonar profiles (a proxy for the weight of the water column), and the depth of the top of the bedrock taken from sonar profiles (a proxy for the weight of the water and sedimentary column). A strong scattering of plotted data points demonstrates that there is no relation between the determined sound velocity and these parameters. Additionally, in the correlation plots, the x-axis values of the minimum (red) and maximum sound velocities (green) are often quite similar (Figure 6a,c,d), again suggesting that velocities are uncorrelated to the overburden thickness. This shows that, in thin sedimentary successions located in shallow water depths, overburden does not significantly influence the sound velocity. For clarity, A-III-and V-prefixes are removed from borehole labels. Data points in orange mark less reliable velocities (boreholes more than 5 m away from the nearest sub-bottom sonar profile; Table 2 and Figure 2b). Data points in green and red mark the maximum and minimum calculated velocities ( Table 2).

Influence of Grain Size
The influence of mean grain size on sound velocity in surficial marine sediments is well known [4,5,19,83,84]. Although granulometric analyses of the cored sediments used in our study were unavailable, general grain size classes could still be determined from borehole geotechnical logs (see Section 2). Therefore, an estimation of the influence of grain size on the sound velocity in our dataset is possible. In Figure 7, the calculated sound velocities at our study site were added to the plot, correlating sound velocity with the mean grain size of surficial sediments of continental shelves from [19]. Clearly, our calculated velocities (at boreholes located close to the acquired profiles) correspond well with the expected sound velocity range of the predominant grain size class determined from the borehole logs ( Figures 5 and 7), even though the dataset of [19] is based on surficial sediment samples. Boreholes penetrating exclusively clay (A-III-6/88 and V5/95 Istrska) are in the lower sound velocity spectra, whereas boreholes mainly encountering silt (A-III-5/88, A-III-9/90 and A-III-11/90) are in the middle velocity spectra, corresponding to silt mean grain size velocities. Maximum sound velocity was calculated for Borehole A-III-8/90, which is the only core containing sandy sediment. Although clay dominates in this borehole, the amount of sandy sediment seems to be sufficient enough to significantly increase the average velocity in the Quaternary succession. A special case is presented by Borehole A-III-13/90, which penetrated clay; however, its calculated velocity corresponds to the mean grain sizes of silt. This discrepancy can be attributed to (1) an inadequate geotechnical description of the core, (2) an abrupt increase in sound velocity within the gravelly horizons ( Figure 5), or (3) the presence of an overconsolidated layer within the sequence. In conclusion, the generally good agreement between the calculated velocities and expected velocities for the predominant grain size classes indicates that the composition of the stratal succession is a major factor influencing velocity variations in our study area. Figure 7. The relation between sound velocity and mean grain size for continental shelf sediments (after [19]) with added velocities from our study. Our estimated sound velocities and the velocity value from [76] for a similar setting in the Gulf of Trieste (see Section 4.3) are shown by white horizontal overlays. Asterisk at Borehole A-III-11/90 indicates that this borehole is separated by more than 5 m from the nearest sub-bottom profile. Pie charts display the proportions of sediment types in each borehole core (for details, see Figure 5).

Influence of Gas Presence
Abundant diffraction hyperbolas are present within the Quaternary succession (Figures 3 and 4; Section 3.2). They could be produced by reflections from gravel horizons; however, their occurrence does not correlate with gravel layers determined in the boreholes (Figures 3 and 4). The diffraction hyperbolas more likely indicate low concentrations of gas in the sedimentary column, commonly encountered in high-resolution geophysical profiles [85][86][87]. Reflective features in the water column (Figure 3b,c and Figure 4c; Section 3.2), which commonly result from gas-bubble plumes emitting from the seafloor [88][89][90], further indicate gas occurrence. The lowest sound velocity in our dataset was calculated for Borehole A-III-7/88, which penetrates a hyperbola-dense zone clearly visible on the corresponding sub-bottom sonar profile (Table 2 and Figure 3c). Since even minor gas concentrations (1-2%) dramatically reduce sound velocity in sediments [83,84,[91][92][93], we attribute this significantly lower velocity value to gas in the Quaternary sediment. In the northern Adriatic seabed, gas seeps are commonly observed and are attributed to both deep and shallow sources [94][95][96][97][98]. Since the diffraction hyperbolas in our sonar profiles are constrained only to a narrow zone in the uppermost part of the Quaternary sequence (Figures 3 and 4), we propose that the gas (probably methane) originates from a degradation of organic matter contained in the Holocene paralic and marine sediments and/or Late Pleistocene terrestrial sequences [62,97,99]. Gas production related to biological processes in seagrass meadows can also greatly hinder the propagation of acoustic signals [100][101][102]; however, the contribution of this effect is difficult to determine from our dataset since only a single borehole (V-5/95) is located within a meadow (Figure 3a). Nevertheless, significant signal attenuation below the meadows can be observed in the geophysical data (Section 3.2, Figures 3a and 4d).

Are Average Sound Velocities an Oversimplification?
Using average sound velocities for depth conversion of a highly heterogeneous Quaternary succession can be considered a gross oversimplification as the velocity strongly varies with grain size [4,5,19,83,84]. However, when comparing the sub-bottom sonar profiles and the borehole logs used in our study ( Figure 5, Figure 3, and Figure 4), a good alignment between the reflections and the main sedimentological boundaries from the borehole logs is apparent (Figure 3b,c and Figure 4a,b). Especially peat layers prove to be very effective reflectors, which has already been noted in the northern Adriatic Sea by other authors [43,47,[50][51][52]76,97,103,104]. This demonstrates that average velocity can be quite effectively used for robust depth conversion of sonar profiles in thin and shallow Quaternary successions.

Comparison with the Sound Velocity from the Bay of Muggia
An earlier study in a similar geological setting [76] reported sound velocity in Late Quaternary sediments from the neighboring Bay of Muggia (Figure 1b), which comprise Rosandra river deposits submerged in the Holocene transgression. There, the Quaternary succession is between 20 and 30 m thick and is composed of clay and silt with occasional sand and peat horizons. The water depth extends between 18 and 21 m. The Bay of Muggia site is therefore quite similar to our study site both in sediment thickness and composition. Using P-wave seismic refraction [76] led to a sound velocity value of 1595 m/s, which fits within the range of velocities estimated in our study (Figure 7). This implies that, also in the Bay of Muggia, the sound velocity in the Quaternary sediment is largely controlled by the sediment type and further corroborates sedimentary type as the major factor influencing sound velocity in shallow, thin, terrestrial-marine Quaternary sedimentary environment successions.

Choosing the Appropriate Velocity for the Depth Conversion of Geophysical Data
The results of our study show that sound velocity in thin (up to 20 m thick) submerged terrestrial-marine Quaternary successions located in near-shore areas few tens of meters deep is mostly controlled by the predominant grain size class of the succession (Section 4.1.2). The sound velocity for depth conversion in these settings can be chosen based on the predominant grain size class. We show that the sound velocity vs. grain size relationships previously documented in surficial sediments [19] are also valid for buried and submerged Quaternary successions (Section 4.1.2). Therefore, published values for sound velocity of surficial sediments can be utilized for depth conversion of shallow offshore high-resolution geophysical data, as long as the selected grain size corresponds to the predominant grain size class of the Quaternary succession.
Terrestrial-marine Quaternary successions often contain significant amounts of degrading organic matter; consequently, locally present gas further influences sound velocity in these settings (Section 4.1.3). Different gas indicators can easily be recognized from high-resolution geophysical data [85][86][87][88][89][90][91][92][93], facilitating the mapping of low-velocity areas. As the velocity decrease associated with the presence of gas is often quite variable e.g., [93], we suggest avoiding detailed velocity analysis in gas-rich areas.

Conclusions
We used geophysical and borehole data to determine sound velocities through the Quaternary fill of a submerged alluvial plain containing terrestrial, paralic, and marine sediments. Our study shows that an average sound velocity through the Quaternary sedimentary column is sufficient for depth conversion of high-resolution geophysical profiles acquired in thin (up to 20 m thick) Quaternary successions in shallow (up to 20 m) water depths. We find that, in these settings, the main factor influencing sound velocity is the sediment type (i.e. mean grain size) contained within the studied sedimentary column, whereas overburden effects do not show any influence. However, where gas is present in the sedimentary column, it reduces sound velocity by a few hundred meters per second and becomes the dominant factor influencing sound velocity.
We found that, for a good approximation of the average sound velocity at a borehole, the velocity typical for the most represented sediment type in the borehole column can be employed. Nevertheless, in highly heterogeneous sedimentary settings, such as the Bay of Koper investigated in this study, significant lateral variations in average velocity will occur within a small area, necessitating a careful selection of multiple, most representative values, if relying on velocities published in the literature.
Using our study area in the northern Adriatic, we provided reference values for sound velocity in thin, mud-dominated Quaternary sedimentary successions in shallow coastal areas. Velocity values determined in our study correlate well with the sound velocity vs. grain size relationships previously documented in surficial sediments [19], showing that these published values can also be used for shallow sub-bottom sedimentary sequences.