Direct measurements of sediment geoacoustic properties in the New England Mud Patch and shelf break^a)

In shallow-water environments, knowledge of the seabed geoacoustic properties is important for accurate predictions of acoustic propagation and bottom scattering. To build an understanding of the geoacoustic properties of fine-grained (muddy) sediments, a series of experimental measurements were performed in the New England Mud Patch (NEMP), which is an area of fine-grained sediment accumulation on the New England shelf (Goff , 2019). The central experiment was known as the 2017 Seabed Characterization Experiment (SBCEX17), and it took place in a region 30 km long in the east-west direction and 11 km wide in the north-south direction that encompassed one of the mud basins within the NEMP (Wilson , 2020, 2022). Additional experimental work continued at the SBCEX17 site in 2021 and 2022 and extended the study area to nearby mud basins with the NEMP and shelf break to the south where coarser material is exposed at the seafloor.

The stratigraphy of the NEMP was originally surveyed by Twichell and McClennen (1981), who created an isopach map covering a 13 000 km² area of the acoustically transparent lens, representing the fine-grained sediment that makes up the NEMP (see Fig. 1). Cores collected at this time indicated that the sediment lens was composed of silt plus clay (Bothner , 1981). A denser set of subbottom reflection profiles was collected in support of SBCEX17 (Goff , 2019), which provided a pseudo-three-dimensional stratigraphic interpretation that resolved the oblique sand ridge morphology beneath the mud layer (see the inset image in Fig. 1).

The SBCEX17 site and surrounding area were extensively sampled in 2016 to provide a detailed parameterized lithostratigraphic model of the region based primarily on sediment grain size and texture, in addition to subtle changes in the mineral content, physical properties, and geophysical response of the sediments (Chaytor , 2022). This study used 89 cores, including 54 piston cores (plus trigger cores, when recovered), 23 vibracores, and 12 gravity cores. The gravity cores were collected using the acoustic coring system (ACS), which uses acoustic probes mounted to the penetrating tip of the corer to obtain an in situ record of the compressional and shear wave speed as the core penetrates the seabed (Ballard , 2020).

This paper builds on the previous work to provide additional insights on the acoustical properties of fine-grained sediments using in situ direct measurements of seabed properties obtained using the ACS and ex situ direct measurements made with a custom core logger. In 2022, 36 additional gravity cores were collected in the NEMP and on the shelf break with the ACS. These collections provided 35 in situ records of the sediment sound speed profile and 17 measurements of shear wave speed near the depth of maximum penetration. Sediment was collected for 33 cores, and sound speed and attenuation were measured in the shipboard laboratory using a custom core logger (Venegas, 2019). Comparison between in situ sound speed profiles measured by the ACS and ex situ profiles measured by core and resonance logger (CARL) provided an assessment of the effects of sediment disturbance from gravity coring operations. This work examines the depth- and range-dependences of the mud layer within the SBCEX2017 site as well as sediments south of the NEMP near the shelf break (encompassing the study site indicated in Fig. 1). Cores collected south of the NEMP support the New England shelf break acoustics (NESBA) experiment (Ozanich , 2022) as well as acoustic propagation data collected in the 2021 and 2022 Seabed Characterization Experiments.

The ACS was developed as an alternative to laboratory measurements of sediment acoustic properties from cores, which can be inaccurate because of changes in pressure, temperature, and geotechnical properties (i.e., bulk density, plasticity, compressibility, permeability, and other parameters affecting consolidation and shear strength) of the sediment caused by removal of the core from the seabed and its subsequent transport to the laboratory. The ACS uses two sets of transducers mounted below the penetrating tip of a sediment corer to make in situ measurements of geoacoustic properties as the corer penetrates the seabed (see Fig. 2). Compressional wave measurements are obtained with rod-mounted piezoelectric cylinders, and shear wave measurements are obtained with bimorph bender elements mounted in flat blades. A second compressional and shear receiver was included in the design to estimate attenuation, but high variability in the measured signal amplitudes during penetration ultimately limited the success of this approach. The ACS continuously transmits and receives pulsed signals as the core penetrates the seabed to obtain an in situ record of geoacoustic properties as a function of depth (Ballard , 2020).

The ACS differs from other sediment acoustic systems such as the acoustic lance (Fu , 1996) and the sediment acoustic-speed measurement system (SAMS; Yang and Jackson, 2020), which both use a source positioned above the seafloor to measure the average sound speed over a set of depth intervals within the seabed. The ACS uses short horizontal propagation paths, which are aligned with the horizontally stratified seabed, to determine localized measurements of wave speed. The acoustic data are acquired during penetration to provide a high-resolution profile of sediment sound speed. Furthermore, the design of the ACS, which embeds the source and receivers in the seabed, enables the measurement of shear waves.

Although the ACS uses the same basic design as its previous iteration (Ballard , 2020), several key changes were implemented to improve mechanical robustness, measurement resolution, and signal quality. For the compressional wave measurements, larger receiver transducers were used to overcome a weak point of the previous design, which led to the bending of probes when penetrating stiffer sediments. The sizes of the piezoelectric elements are listed in Table I. Electrical connectors were added at the probes to improve ease of removing the core liners filled with sediment and enable more rapid redeployment.

The sampling rate of the compressional recording system was increased from 1 to 3 MHz compared to results reported in Ballard (2020). For the source/receiver separation and nominal sediment sound speed of the SBCEX17 site, this change reduced the uncertainty in the estimated sound speed from 20 to 7 m/s. The signal generation electronics were redesigned to accommodate a longer sequence of tone burst signals. For the 2022 deployment, twice the number of signals were used compared to the 2016 deployment, providing a denser sampling of the frequency range. The center frequencies of the tone bursts were 30, 40, 50, 80, 90, 100, 150, and 200 kHz. The tone burst durations were between 0.2 and 0.4 ms depending on frequency and weighted with a Tukey window.

Finally, preamps were mounted directly to the receiver probes to boost the signal before transmission through the cables to the instrumented pressure vessel (IPV). This change made it possible to detect the shear wave signal arrivals near the depth of maximum penetration when the speed of the core had slowed and the penetration noise was reduced. The shear system also used tone bursts, and three frequencies were used: 400, 800, and 1200 Hz. The tone burst durations were between 6 and 18 ms depending on frequency and weighted with a Tukey window.

In addition to the acoustic probes, the ACS includes a pressure sensor located on the IPV to measure the system's depth and a nine-axis inertial measurement unit (IMU) to compute its angle of declination as the corer penetrates the seabed. The ACS was used with a gravity core designed with the IPV rigidly attached to the corer above the 300 kg ballast weight and stabilizing fins (see Fig. 2).

On board the vessel, the cores were equilibrated to ambient conditions and then underwent noninvasive acoustic measurements using a custom-built, vertically oriented, broadband acoustic CARL. High-frequency (170–1000 kHz) measurements of compressional wave travel time and amplitude were made through the diameter of the core at 2–5-cm intervals along the length of the core to provide vertical profiles of sound speed and attenuation. Eight tone bursts were used with center frequencies of 170, 220, 280, 360, 460, 600, 770, and 1000 kHz. The duration of each tone burst was limited to minimize the effects of multipath from the walls of the core liner. Sound speed was calculated according to Eq. (2) using travel time measured through a core liner filled with degassed, de-ionized water as the reference measurement, t_w. In addition to uncertainty in the length of the propagation path, the CARL sound speed uncertainty also includes uncertainty in the thickness of the core liner (Venegas, 2019).

A total of 36 cores were collected over a 10-day period from 15 April to 25 April 2022. The ACS was initially deployed with both sets of compressional and shear wave probes attached to the nose cone. Seventeen cores (AC-1–AC-17) were collected in the NEMP, and five cores (AC-18–AC-22) were collected on the shelf break with this configuration. Because of difficulties penetrating into the coarser sediment and shell hash present on the shelf between the 200- and 400-m isobaths, the shear wave probes were removed from the ACS, and six additional cores (AC-23–AC-28) were collected in the shelf break region with only the compressional wave probes attached. The removal of the shear probes reduced the surface area of the core, and penetrations up to 1.3 m were achieved in the region between the 200- and 300-m isobaths, where the highest sound speeds were measured. Finally, eight more cores (AC-29–AC-36) were collected in the NEMP with only compressional wave probes attached. For this final set of cores, the gravity core was configured with two 3-m barrels connected by a coupler, and penetrations over 4 m were achieved for all of the deployments in this configuration.

In situ compressional wave profiles were collected from all cores except AC-36 for which a failure of the ACS recording system caused a loss of data. The 30 kHz tone burst had poor signal-to-noise ratio (SNR), and sound speed could not be estimated at this frequency. Although there were occasionally missing data points at selected frequencies at some penetration depths, the vertical sampling is generally dense enough that there was no significant loss of information.

Shear wave measurements were limited to cores that had shear wave probes attached. There were four cases of failure resulting from mechanical damage to the probe assemblies, and no data were collected. These failures occurred for AC-12, which was located in the southwest corner of the SBCEX17 site, where the probes sampled the coarser material of the transition layer beneath the mud, at AC-17, where the transition layer is exposed at the seafloor, and at AC-20, located on the shelf break. Core AC-13 was also negatively impacted because a faulty preamp damaged during the AC-12 deployment was not replaced before collection at this site. Furthermore, the penetration noise when the core was moving fast on entry into the seabed caused poor SNR, and shear wave speed estimates were only obtained near the depth of maximum penetration when the core's trajectory had slowed. Finally, the 400 Hz tone burst had poor SNR due to low source output, and shear speed could not be estimated at this frequency. Sediment samples were collected for all cores except AC-22,-26, and -27. These cores were collected near the shelf break, where the sediment consisted of coarser material, and the lack of cohesion enabled the sediments to flow out through the core catcher.

The penetration depth (determined by the pressure sensor on the IPV) was compared to the length of sediment collected within the core barrel to quantify the effect of core shortening, which results from friction of sediment against the tube wall, causing thinning by lateral extrusion in front of the corer (Ballard , 2020; Blomqvist, 1991; Duck , 2019; Emery and Dietz, 1941; Lebel , 1982). After removing one outlier, the penetration-to-recovery ratio was consistently greater than one with an average value of 1.64 ± 0.31 [M (mean) ± standard deviation (SD)]. Overall, the penetration-to-recovery ratio reported here was consistent with that of the 2016 collection, 1.76 ± 0.30 (M ± SD). No relationship between sound speed ratio and penetration-to-recovery ratio could be defined.

The angle of entry into the seabed was determined using the IMU. There were six cores that had entry angles greater than 15°, and five of these were collected in a higher sea state preceding or following a major storm event. The entry angles of the remaining 29 cores were closer to vertical with average value of 5.8° ± 3.3° (M ± SD).

As described in Sec. II B, the sound speed measured in the water column above the seafloor was used to calculate the reference travel time measurement, t_w. Sound speed was measured by the ship's CTD each morning and evening on days that cores were collected as well as when the sampling location was a significant distance from the last CTD cast. The measured sound speed profiles are displayed in Fig. 3 and their locations are shown in Fig. 4.

Several sites were sampled multiple times during the 2022 experiment. In some cases, two cores were collected in rapid succession to test repeatability of the ACS measurements. In other cases, locations were initially sampled with the shear wave probes attached to obtain a measurement of the shear wave speed and then sampled again at a later time without the shear wave probes to obtain a record of the compressional wave speed at greater depths within the seabed. The intercomparison of ACS measurements at a frequency of 100 kHz is shown in Fig. 5. Cores AC-01 and AC-02, AC-04 and AC-05, and AC-13 and AC-14 were collected in the same location for the purpose of testing repeatability of the ACS measurements. The remainder of the cores were collected later without the shear wave probes attached. The greater penetration depths achieved without the shear wave probes can be clearly observed in Fig. 5. For the configuration with a single 3 m tube, some records extend farther than 3 m into the seabed because the entire core barrel and lower portion of the weights were buried in the sediment.

Overall, there is good agreement between the cores collected at the same locations as indicated by the overlap in the data points and their error bars. Small variations are caused by heterogeneities, including the presence of bivalves, which have been observed throughout the mud layer (Chaytor , 2022). Although the ship was positioned in approximately the same location for the comparisons, each measurement samples a different position within the seabed, therefore, some-small scale variations are expected. There are a few cases for which one core appears biased higher than its counterpart; for example, AC-14 appears to have a higher sound speed ratio, on average, compared to AC-13 [see Fig. 5(a)]. However, there is generally overlap in the measurements' error bars. Furthermore, the variation in the sound speed ratio between two cores at one location is much less than that of cores from other locations; for example, compare AC-13 and –14, which have a sound speed ratio close to 0.975 to cores AC-06 and AC-29, which have a sound speed ratio around 1.01.

Although a more detailed description of depth-dependence within the mud layer will be provided in Sec. III C, the influence of the stratigraphy on individual cores presented in Fig. 5 is discussed here. The general structure of the seabed in the NEMP includes a mud layer with weak depth-dependence, a 1–2-m-thick transition layer with strong depth-dependence in geoacoustic properties (increasing sound speed, density, and attenuation with depth) due to increasing sand content in the mud, followed by a sand layer with more uniform properties (Chaytor , 2022; Goff , 2019; Jiang , 2023). The measurements depicted in Figs. 5(a)–5(e) are all well contained within the mud layer as determined by the high-resolution survey (Goff , 2019). However, for core AC-29 in Fig. 5(f), the mud base is at a depth of 5.8 m, and the measured sound speed shows a slight increase with depth below 3.8 m, which is associated with the transition layer.

The comparison in Fig. 5 concentrates on cores collected along the main diagonal line through the center of the SBCEX17 site plus cores from the northeast and northwest corners. Intercomparisons between sound speed profiles collected in 2016 and 2022 from the southwest part of the SBCEX17 site are contained in Garcia (2024b). The agreement between the calculated sound speed profiles is very good, including the measured increase in sound speed as the cores penetrate into the transition layer. The two different data sets were collected with different versions of the ACS and different approaches in the data analysis. The agreement between the 2016 and 2022 data sets is a sign of the robustness of the measurements.

Range-dependence of the surficial sound speed in the study area is summarized in Fig. 4. Surficial sound speed was calculated as the mean of the upper 25 cm of the in situ profile measured by the ACS at 100 kHz. The lowest surficial sound speeds are observed within the NEMP, whose location is indicated by the contours representing the mud layer thickness. Core AC-17, which is located between the 0- and 2-m thickness contours, has a notably higher sound speed than the rest of the cores within the NEMP. The mud layer may be patchy in this region, and the influence of the transition layer between the mud and sand layers is likely presenting itself at the seafloor and may also be the cause of the increase in sound speed.

Within the shelf break region, the fastest sound speeds are observed between the 200- and 350-m isobaths. The sediment in this region is believed to be continuous with the sand layer underlying the NEMP to the north. High-resolution seismic profiles in this area appear opaque with minimal penetration into the seabed (Knott and Hoskins, 1968; O'Leary, 1988). Finally, the ACS measurements collected near the 400-m isobath have a surficial sound speed close to one. The presence of fine-grained sediments on the slope is documented through sediment grabs (McMullen , 2014). High-resolution seismic profiles show a finely layered structure in the upper seabed (O'Leary, 1988; ten Brink , 2014).

The range- and depth-dependences of the measured sound speed profiles are examined in Fig. 6 as a function of latitude. The top panel shows the influence of bathymetry on the surficial sound speed, and the three regions described above can be clearly observed in Fig. 6: the sound speeds measured on the relatively flat shelf in the NEMP, where the sound speed ratio is predominately less than one; the upper portion of the slope, where shelly sandy material is exposed at the seafloor; and the sound speed ratio is greater than one and water depths >350 m, where surficial sound speeds are close to one.

The lower panel of Fig. 6 reveals the depth-dependent structure of the sound speed profiles measured by the ACS. The depth of the mud base at 70.6° longitude provided the ability to delineate the location of the NEMP, and the SBCEX and southern mud basins can be observed. Within the NEMP, the cores are relatively homogeneous compared to cores collected in the shelf break region. Assuming that the base of the sand layer follows the trajectory of the base of the mud layer, the gray dashed line indicates the extrapolated boundary of sand base. Within the region bounded by the locations where mud base and sand base intersect the seafloor, the sandy sediment is exposed near the surface of the seabed, and higher sound speeds are observed. Cores AC-23, AC-26, and AC-27 show evidence of a high speed layer below 0.5 m. Cores AC-24 and AC-28 do not penetrate deep enough to sample this layer, and it is possible that the presence of the coarse material caused the failure to penetrate. Further offshore, cores AC-18–AC-21 all show evidence of layering in the upper 2 m of the seabed.

To understand the depth-dependence of the mud layer, the sound speed profiles collected in the NEMP were interpolated onto a uniform depth grid and averaged together to obtain the mean sound speed profile as a function of depth. Averaging was necessary because nonuniformities within the sediment matrix (such as bivalve shell fragments; see Chaytor , 2022, for further description) result in small scale fluctuations that make it difficult to observe the depth-dependence in a mean sense (see Fig. 5). Additionally, the gradient in the sound speed profile is on the order of the measurement uncertainty, making it challenging to identify trends from a single measurement. All sound speed profiles collected in the NEMP are displayed together in Fig. 7. The profiles have been normalized by the mean sound speed in the upper 25 cm to remove spread caused by the range-dependence from results. The normalization is only applied to improve visualization of the data, and it does not change the shape of the profiles. The mean profile is overlaid on the measured profiles (left panel), and it is also plotted without the measurements for closer inspection (right panel). There is less sample support below 3 m because only the last seven cores collected without the shear wave probes attached penetrate to 4.6 ± 0.2 m (M ± SD).

The data show a strong negative gradient of –28 s⁻¹ (⁠ $r2=0.82$⁠, p < 0.0001) in the upper 35 cm of the sound speed profile. Below the upper 35 cm, the normalized, averaged sound speed ratio calculated from the ACS measurements shows a positive gradient that extends to a depth of approximately 2.25 m with a gradient of 6.2 s⁻¹ (⁠ $r2=0.92$⁠, p < 0.0001). The deepest section of the ACS measurements extends from 2.5 to 4.5 m and is characterized as an isovelocity layer (see Fig. 7).

The sediments collected by the gravity core were logged using CARL. Because the CARL measurements are made on the sediment collected by the gravity core, they were linearly stretched to match the depth of the penetration for comparison with the ACS measurements. The same stretching procedure was applied in previous work, and good qualitative agreement was achieved between the ACS sound speed measurements and physical seabed properties (Ballard , 2020).

The ACS and CARL sound speed profiles for core AC-32 are plotted together in the left panel of Fig. 8 at frequencies of 200 and 220 kHz, respectively. The highest frequencies of the ACS measurements overlap with the lowest frequencies measured by CARL, and closest frequencies between the two data sets are used for this comparison. After accounting for the stretching factor, the depth sampling interval for CARL was 7.1 cm. There are gaps in the record between around 0.2, 1.5, and 3.0 m, where the core was cut into sections. There are additional data gaps where low SNR prohibited a measurement. Causes of low SNR are bubbles that become trapped in the core when the liner is removed from the steel tube and capped as well as from outgassing of biogenic processes in the sediments. The ACS sampling interval depends on penetration speed. For core AC-32, it was approximately 4 cm over the upper 1.5 m of the record and then decreased over the remaining 3 m as the core was slowed by friction. In comparing the values of the estimated sound speed ratio, the CARL measurements are biased faster than the ACS measurements in the upper 1 m of the core. This bias, however, is within the uncertainty of both measurements and, therefore, shows good agreement overall. Below the 1 m, the sound speeds from the ACS and CARL are more consistent with one another.

To assess comparison of the ACS and CARL measurements from all cores collected and logged in the NEMP, the average difference between the sound speed ratios calculated from the ACS and CARL measurements are plotted in the right panel of Fig. 8. The averaged difference between ACS and CARL sound speed ratios shows a linear trend of 7.6 s⁻¹ in the upper 2.25 m of the seabed. There is a relative bias of 12 m/s near the seafloor, and the two measurements converge with depth. Between 2.25 and 3.5 m, the sound speed ratios estimated by the two systems appear equivalent. There is less sample support below 3.5 m because only the subset of cores collected without the shear wave probes penetrated to these depths.

Combining the ACS and CARL measurements provides a broadband examination of sound speed dispersion. The top panel of Fig. 9 shows sound speed as a function of frequency averaged over 50 cm depth intervals for core AC-32, which was collected near the center of the SBCEX17 site and had a penetration depth of nearly 4.5 m (see Fig. 8), and for core AC-25, which was collected on the shelf break. Although AC-25 only penetrated 124 cm, it was among the cores with highest sound speeds for which sediment was retrieved for logging. The difference in the estimated sound speeds between the two sites can be clearly observed in the measurements. There is also greater frequency dispersion in the sound speed measured at AC-25, which has more sand content than the sediment collected in AC-32.

First, consider AC-32. There is a trend of increasing sound speed across the entire frequency band. A slight hump observed at 50 kHz may be the result of a transducer resonance. The shallowest depth interval measured by CARL for AC-32 has slightly elevated speeds compared to the deeper sections. This depth interval was contained in a separate short core section (see Fig. 8), and disturbance to the sediment fabric of the delicate upper layer of the seabed is a possible cause for the offset in the measurements. Over the remainder of the core, the ACS measurements show a trend of increasing sound speed with depth, whereas the CARL measurements show a pattern of decreasing sound speed with depth, which is most pronounced at the highest frequencies.

Core AC-25 measurements were limited to two depth intervals because of limited penetration into the coarse material present at the seafloor. The larger error bars recorded for core AC-25 are evidence of variability of the sound speed measurements within each depth interval. The ACS measurements show a general trend of increasing sound speed with frequency, and the CARL measurements continue the trend of increasing sound speed as a function of frequency to 770 kHz. The decrease in sound speed at 1 MHz is attributed to multiple scattering caused by the sand fraction present in the AC-25 samples (Kimura, 2011).

Attenuation measured by CARL is depicted in the lower panel of Fig. 9. The sandier sediment sampled by AC-25 has greater values of attenuation and a frequency exponent of 1.3 for both depth intervals. The increase in attenuation with depth is attributed to layering, noting that the deeper sediments also have lower sound speed. The attenuation measured for the AC-32 samples has a more complicated dispersion pattern that changes with depth. At the lowest frequencies, attenuation decreases with depth, whereas at the highest frequencies, attenuation is highest at the deepest depths. The frequency exponent is 1.0 for the shallowest depth interval and increases to 1.8 at the deepest depth interval.

Shear wave speed was calculated according to the method described in Sec. II C, using a reference measurement collected in a laboratory water-saturated sand to calculate the effective distance between the shear wave probes of the ACS. The shear wave speed in the sand was found using multiple receivers such that the effects of the system response could be eliminated from the laboratory measurement. The shear wave speed in the sand was 84.8 and 88.9 m/s at 800 and 1200 Hz, respectively. The 400 Hz signals had low SNR in the laboratory and field measurements. Using the measured travel time in the sand for the source/receiver pair mounted on the ACS nose cone, effective distances of 22.04 and 22.75 cm were determined for the 800 and 1200 Hz tone bursts, respectively.

As a result of poor SNR during penetration, shear wave signals could only be identified in the data when the core had slowed to a near stop at its maximum penetration depth. Within the NEMP, 14 measurements of the shear wave speed were collected at depths ranging from 2.4 to 3.6 m below the seafloor, and three measurements were collected on the shelf break at depths ranging from 1.4 to 2.7 m below the seafloor. A composite plot of the shear wave speed as a function of depth revealed no discernible pattern, and the assumption is that range-dependence rather than depth-dependence has the greatest influence on the measurements over the narrow range of depths sampled.

The shear wave speeds at 1200 Hz were approximately 12 ± 6 m/s (M ± SD) faster than the 800 Hz measurements, indicating frequency dispersion in the data set. Overall, the 800 Hz data had higher SNR, and the estimated wave speeds were constrained to a tighter range of values. The estimated shear wave speeds at 800 Hz are compared to the estimated compressional wave speeds at 100 kHz over approximately the same depth interval in Fig. 10. The highest compressional and shear wave speeds were measured for AC-11, which penetrated into the transition layer. The wide error bars on the compressional wave speed result from the rapid increase in sound speed measured over the depths near maximum penetration. The lowest shear wave speed was measured at AC-14, located on the west side of the SCBEX17 site, where the smallest mean grain size and low sound speeds were also observed. The compressional and shear wave speeds are fairly well correlated with $r2=0.74$ (p < 0.001). The black line in Fig. 10 represents a linear regression between compressional and shear wave speed, $cs=0.232cp−218$⁠.

The large number of cores collected within the NEMP provides insight into the spatial variability of the sediment properties within the upper 5 m of the SBCEX17 site and southern mud basin. An initial assessment of the data showed a trend of higher sound speeds on the east side of the sampled area (Garcia , 2024b). The range-dependence of the surficial sound speed is put into context of sediment physical properties using a gridded map of sediment mean grain size within the SBCEX17 site (Chaytor , 2022). The surficial sound speed ratios, calculated as the mean of the upper 25 cm of the in situ profiles measured by the ACS at 100 kHz, are plotted as a function of longitude in Fig. 11. Cores AC-11 and AC-12 were omitted from this comparison because the mud base is within a few meters of the seafloor. Cores AC-15 and AC-16 from the southern mud basin were included, but AC-17 was excluded because of thinning and patchiness of the mud layer. A third-order polynomial fit was applied to the data to illustrate the higher rate of increase in the sound speed ratio on the east side of the study area. This trend is consistent with a larger change in the mean grain size on the east side of the SBCEX17 site [see Fig. 9(a) from Chaytor , 2022]. Interpolating the mean grain size data to the location of the ACS cores and extrapolating to cores AC-15 and AC-16, a linear relationship between sound speed ratio and mean grain size was determined with $r2=−0.94, p≤0.0001$ (see inset plot in Fig. 11). Although this comparison was based on measurements of the surficial sound speed using data from the upper 25 cm seabed, the trend of increasing sound speed to the east can also be observed in the depth-dependent profiles shown in Fig. 5.

The range-dependence of the acoustic properties of the NEMP can also be noticed in the CARL measurements. Figure 12 shows the sound speed ratio and attenuation at 280 kHz averaged over the 1–1.5 m depth interval as a function of longitude. Samples collected near the seafloor were subject to the most disturbance from the coring activity, and more consistent properties were observed in the CARL measurements below a penetration depth of 1 m (see Sec. IV C for additional discussion on sediment disturbance). The same 21 cores employed for the ACS analysis are used to examine the range-dependence of the CARL measurements. However, there is more variability in the CARL measurements, which could be attributed to sediment disturbance, irregularities in the thickness of the core liner, or other factors. The increased scatter in the measurements made the third-order polynomial fit unsuccessful, and a linear fit was applied to the data to illustrate the increase in sound speed to the east. Making the comparison to the gridded mean grain size, the correlation coefficient was weaker than that observed for the ACS measurements but still credible with $r2=−0.64$⁠, p = 0.0033 (see inset plot in upper panel of Fig. 12). The weaker correlation could be attributed to a discrepancy in the collection depth of the grain size (upper 5 cm) and CARL measurements (1–1.5 m).

The range-dependence of attenuation in the NEMP is displayed in the lower panel of Fig. 12. Higher attenuation is observed on the east side of the NEMP, where the mean grain size is larger. The relationship between mean grain size and attenuation is reasonably strong with $r2=−0.79$⁠, p < 0.0001 (see inset plot in lower panel of Fig. 12).

The range-dependence of the sediment sound speed within the SBCEX17 study area and greater NEMP has also been documented by comparisons of inference results from different sites. Jiang (2023) compared results from three sites obtained via trans-dimensional (trans-D) inversion of reflection-coefficient data, employing the Viscous Grain Shearing theory sediment model and spherical wave reflection data predictions. All three sites were in the SBCEX17 study area, and the estimated sound speed ratios at 0.1 m below the seafloor at a frequency of approximately 1 kHz show excellent agreement with the range-dependent in situ measurements depicted in Fig. 11. Direct measurements from the SAMS also show a general trend of increasing sound speed moving from sites on the west to east sides of the SBCEX17 site, but the interpretation of the results is complicated because the measurements at each location are integrated over different depth intervals (Yang and Jackson, 2020).

The sound speed ratio profile calculated from the ACS measurements presented in Sec. III C showed a strong negative gradient in the upper 35 cm of the seabed. This result is consistent with a complementary set of direct measurements obtained using a multicorer system equipped with acoustic probes. The upper 10 cm of sediment is also known to be the portion of the seabed that is most affected by bioturbation (Boudreau, 1998). Acoustic sensitivity to infauna and their activities, such as feeding, locomotion, and tube or borrow building, has been established by laboratory and field studies (Dorgan , 2020; Lee , 2022). Below the upper 35 cm, the normalized, averaged sound speed ratio calculated from the ACS measurements showed a positive gradient that extends to a depth of approximately 2.25 m, which overlies an isovelocity layer (see Fig. 7). This description of the seabed is consistent with compaction resulting from overburden pressure until the sediment reaches an equilibrium state. For the sediments in the NEMP, this transition could mark the transition between a silt suspension and gravity driven particle interactions. The change in properties as a function of depth is consistent with field observations, which found the consistency of the sediment near the seafloor to be gelatinous and jiggly and the sediment near maximum penetration to be sticky and firm.

Regardless of the physical mechanism responsible for the sound speed profile observed in situ, the description of an upper mud layer a few meters thick with a weak sound speed gradient, g < 10 s⁻¹, followed by a mud layer with uniform properties that extends to about 1–2 m above the mud base is consistent with inference results from Jiang (2023), which approximated a gradient of 6 s⁻¹ from the stacked layers estimated by trans-D inference over an isovelocity mud layer. Additional support for the weak sound speed gradient comes from direct measurements from SAMS, in which Yang and Jackson (2020) estimated a linear gradient of 7.7 s⁻¹ between depths of 1 and 3 m. Yang and Jackson (2020) also proposed an exponential gradient as the SAMS measurements trend toward a constant value below 2.5 m. The SAMS measurements could also be described by the piecewise linear profile proposed in Fig. 7. Note, SAMS consistently sank 30–50 cm into the seabed and did not sample the surficial sediments where the negative gradient is observed in the ACS data.

The effects of disturbance from gravity coring operations on the sediment sound speed ratio can be assessed from a comparison of the ACS and CARL measurements. The ACS measurements are collected in situ with the goal of obtaining direct measurements of the acoustic properties of the seabed in their natural undisturbed state. The CARL measurements are collected after the sediment has been removed from the seabed and come to equilibrium under laboratory conditions. Overall, the ACS and CARL measurements were in good agreement over a broad frequency range as observed from the smooth transition between the lower frequency ACS measurements and higher frequency CARL measurements (see the top panel of Fig. 9). Furthermore, the direct comparison between the ACS and CARL sound speed ratio profiles showed good agreement with overlapping error bars for almost all depths (see the left panel of Fig. 8). The greatest difference was observed near the seafloor, where the CARL measurements were less than 1% greater than the ACS measurements on average. Because both measurements have an uncertainty on the order of 0.5%, this level of disparity is within the combined uncertainty of the measurements. Nevertheless, this section attempts to understand causes for the differences in the in situ and ex situ measurements to gain insight on the effects of the coring process on marine sediments.

Some of the main factors believed to have caused the greatest differences between ACS and CARL measurements are (1) a bow wave in front of the cutting edge caused by restricted water flow through the corer; (2) core shortening, which is the reduction of the sediment core length by physical compaction, sediment thinning, and/or sediment bypassing; and (3) liquefaction/deformation during core recovery and transport on deck (Duck , 2019). The bow wave can wash away fluffy surficial sediment before the corer reaches the sediment–water interface. Previous ex situ measurements on the sediments collected using the ACS in the NEMP found the upper 5–20 cm of sediment were lost based on ²¹⁰Pb dating (Ballard , 2020). Depending on the size of the bow wave, the direct measurements may or may not be affected due to their position below the cutting edge of the corer. Core shortening caused by friction between the sediment and inner tube wall of the corer is believed to be the dominant reason for the differences between the penetration depth and length of collected sediment. Studies have shown that core shortening patterns can be uniform, progressive over depth, or a mixture of different patterns (Duck , 2019). Nonuniform shortening will be a source of error in the comparisons between the ACS and CARL profiles as a linear stretching factor was applied to the ex situ measurement depths. Nevertheless, for sound speed profiles collected on the shelf break for which distinct layers can be observed in the measurements, the linear stretching factor provides good qualitative agreement in the shape of the sound speed profile as a function of depth (see Fig. 13). Liquefaction and deformation of the surficial sediment is believed to be a possible source of disturbance as the cores were laid horizontally on deck during recovery to remove the core liners from the steel barrels. The surficial fluffy layer was observed to tilt, and entrainment of gas into the sediment as the core liner was re-orientated to vertical for processing and storage is possible.

Comparing the average sound speed ratio and average difference profiles (see the right panels of Figs. 7 and 8), similarities can be observed. The averaged ACS sound speed profile has a positive gradient in the upper 2.25 m; the gradient in the difference profile was also observed at the same range of depths. The magnitude of the difference gradient is similar to that of the gradient observed in the ACS sound speed ratio. It is believed that the disturbance of the sediments during the coring process, especially core shortening and liquefaction, affected the delicate structure of the fluffy shallow sediments. Below 2.5 m, the averaged ACS sound speed ratio profile is that of isovelocity, and the averaged difference profile is approximately zero. The consistency in the estimated sound speed ratios between the ACS and CARL measurements below 2.5 m suggests that this sediment was minimally disturbed by the gravity coring operation over this depth interval. Moreover, the constant value of the sound speed ratio below 2.5 m suggests that the sediment geoacoustic properties have reached a steady state under the overburden pressure of the sediments above.

The shear wave speeds estimated for the NEMP are greater than that expected for fine-grained sediments if the depth- and frequency-dependences of the shear wave properties are not accounted for. Jackson and Richardson (2007) defined a regression based on measurements from the in situ sediment geoacoustic measurement system (ISSAMS) that corresponds to shear wave speeds between 20 and 55 m/s for the range compressional wave speeds presented in Fig. 10. One notable difference between the ISSAMS and ACS measurements is that the ISSAMS measurements were collected in the upper 30 cm of the seabed, whereas the ACS shear wave speeds were collected between 1.4 and 3.6 m below the seafloor. Assuming a power-law shear wave speed profile, the difference in measurement depths could explain the difference in the measurements.

Bowles (1997) compiled shear wave speed profiles in fine-grained sediments from more than 20 sources and found that they tend to separate into high-velocity and low-velocity groups. It was determined that sediment texture accounted for the separation such that the higher-velocity profiles reflect fine-grained sediments with a coarse-grained component. The sediments of the NEMP and shelf break contain a percentage of coarse-grained particles consistent with the grouping of higher-velocity profiles. Nevertheless, even within this group of profiles, the shear wave speeds are between 80 and 100 m/s for the range of depths measured by the ACS, approximately 40 m/s less than the in situ measurements. However, many of the profiles included in Bowles (1997) analysis were inferred from low-frequency (<20 Hz) interface waves, and the frequency dispersion must be accounted for when making comparisons to high-frequency (800 Hz) direct measurements (Ballard , 2018).

Finally, Potty and Miller (2020) examined Airy phase arrivals from explosive charges recorded in the northeast corner of the SBCEX17 site. Geoacoustic parameters for the seabed were estimated using a grid search to find the best match to modal arrival times. The mud layer most affected Airy phases of the two lowest order modes in the 20–50 Hz frequency range. The best match was produced with a shear wave profile with a positive linear gradient in a 6-m-thick layer of mud with shear wave speeds between 20 and 50 m/s. Combining low-frequency propagation measurements like these with the high-frequency direct measurements from the ACS provides an opportunity to use sediment acoustic models (Buckingham, 2020; Chotiros, 2024) with inference algorithms (Bonnel , 2024; Jiang , 2023) to estimate shear wave dispersion over a broad frequency range.

This paper reported on a new set of in situ sound speed profiles collected with the ACS in the NEMP and shelf break area to the south. Compared to a previous version of the ACS, higher resolution sound speed measurements were obtained, which provided new insights into the range- and depth-dependences of the mud layer within the NEMP. An intercomparison of sound speed profiles collected in the same locations showed that the results were repeatable, including accounting for changing the configuration of the probes on the nose cone of the gravity core.

An examination of 22 cores collected within the NEMP exhibited range-dependence of the geoacoustic properties of the mud with slower sound speeds and lower attenuation on the west side of the SBCEX17 site and an increase in sound speed and attenuation on the east side. The higher rate of change in sound speed on the east side of the SBCEX17 site was well correlated with a gridded mapping of the mean grain size. In the shelf break region to the south of the NEMP, the highest sound speeds were observed between the 200- and 350-m isobaths, where the sediment is acoustically opaque to seismic surveys. Further offshore, near the 400-m isobath, the finer grained sediments associated with the slope are present, and a sound speed ratio close to unity was observed.

The depth-dependence of the mud layer was examined by averaging across all profiles collected in the NEMP, excluding those profiles that were influenced by the sand layer. The results showed a surficial layer approximately 35 cm thick with a negative sound speed gradient believed to be caused by the presence of infauna and their activities. The remainder of the mud column included a weak positive sound speed gradient that extends to a depth of 2.25 m below the seafloor and overlies an isovelocity layer. Based on other data sets, it is believed that the isovelocity mud extends to the transition layer, but the deepest ACS measurements only characterized the upper ∼5 m of the seabed.

Comparison between in situ sound speed profiles measured by the ACS and ex situ profiles measured by CARL provided an assessment of the effects of sediment disturbance from gravity coring operations. The comparison showed a difference averaged over all cores of approximately 12 m/s at the seafloor that decreased with depth. Below 2.5 m, the average difference was close to zero, suggesting that these sediments were minimally disturbed. Notably, the greatest difference was observed near the seafloor, where the high porosity mud consists of a delicate structure that is most susceptible to disturbance.

The broadband frequency dispersion of the sound speed and attenuation was analyzed using a combination of the ACS and CARL measurements. The differences in the sound speed dispersion between a core collected in the center of the SBCEX17 site and a core collected in sandier sediments were distinct. The core from the sandier sediment had significantly greater sound speed and higher attenuation. The core from the NEMP had a complicated dispersion pattern that changed as a function of depth within the seabed.

Finally, a direct in situ measurement of shear wave speed was obtained with the ACS. The shear speed estimates were provided near the depth of maximum penetration when the penetration noise was at a minimum. The shear wave speed measurements were well correlated with compressional wave speed measurements from approximately the same depth interval.

The ACS provides a unique data set that includes direct in situ measurement of acoustic parameters with collected sediment for post-experiment laboratory analysis. Future work will use the geoacoustic properties (porosity, bulk density, grain size distribution, etc.) that were derived from physical measurements of the sediments to evaluate their predictive capability to compare with the measured wave speeds. When combined with low-frequency inferences of acoustic properties, the complementary data sets provide measurements of sound speed, sound attenuation, and shear speed over a wide frequency band, which are necessary for constraining sediment acoustic models.

This work was supported by the U.S. Navy Office of Naval Research under Grant Nos. N00014-21-1-2245, N00014-21-1-2254, N00014-20-1-2781, N00014-22-1-2128, N00014-21-1-2261, and N00014-20-1-2458.

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.