Sound speed and attenuation measurements within a seagrass meadow from the water column into the seabed

Biological processes and physical characteristics associated with seagrass can greatly affect acoustic propagation in coastal regions where seagrass meadows are present. An important acoustical effect is due to bubble production by the plants, which can have a significant impact on both object detection and bottom mapping sonars by increasing clutter through reflection, absorption, and scattering of sound.^1,2 Remote sensing techniques have been demonstrated to monitor biological markers, such as photosynthetic activity from seagrass, as an assessment of marine ecosystem health.^3,4 Additionally, seismo-acoustic survey tools have been investigated to obtain carbon sink estimates for the sediment underneath seagrass beds.⁵ The presence of marine vegetation in the shallow water environment has the potential to affect the use of underwater acoustics for scientific, commercial, and military purposes.

Gas generated by seagrass photosynthetic activity can dissolve directly into the surrounding seawater or form bubbles that cling to the outside of the leaves.⁶ It has also been shown that gas production by seagrass is temporally variable over both shorter (diurnal) and longer time scales (seasonal and longer), indicating that the potential acoustic effects are also time dependent.⁷ In addition to the gas-bearing leaf tissue in the water column, the rhizomes also contain aerenchyma (gas-filled canals), which allow for diffusion of oxygen into the surrounding sediment. The density and elastic moduli of the plants themselves can also potentially affect long-range acoustic propagation by altering the effective material properties at the water-sediment interface and within the seabed when seagrass meadows are ubiquitous in the environment.^8,9

Whereas previous studies on the acoustic properties of seagrass have focused on sound propagation and backscatter in the water column, little prior work has focused on measurement of acoustic properties below the water-sediment interface where the plant rhizome and root systems exist. This letter reports preliminary in situ measurements of sound speed and attenuation in a bed of Thalassia testudinum in the water column, in the canopy, and in the sediment beneath the seagrass. For comparison, measurements were also obtained at the same depths at a nearby site with no seagrass and only bare sediment. Sediment cores were collected at both sites and analyzed for macroscopic plant biomass, sediment bulk density, and mean grain size to aid in interpretation of the acoustic data.

The field experiment was conducted in east Corpus Christi Bay near Port Aransas, Texas during March 2016. T. testudinum is a highly abundant seagrass species in this region;¹⁰ however, these measurements were acquired during the dormant growth season, and the plant leaves extended to a height of approximately 8 cm above the seafloor, and photosynthetic activity was low. The water depth at the experiment site was approximately 1 m, and the site was accessed by small watercraft operated by the University of Texas at Austin Marine Science Institute (UT:MSI). The location of the experiment site (27° 48.550′ N, 97° 07.201′ W) was determined from a handheld GPS receiver. Acoustic measurements and core samples were collected in a bare sediment site and a bed of T. testudinum only a few meters apart.

The in situ acoustic measurements were conducted using a manually deployed system developed at the Applied Research Laboratories at the University of Texas at Austin (ARL:UT).¹¹ The measurement system was configured with a single compressional wave receiver and a single source held at a fixed distance from each other, with the acoustic axis between the source and receiver parallel to the water/sediment interface. This configuration allowed the measurement system to be more portable than a multi-receiver configuration would be. In the bare sediment region, acoustic measurements were conducted 15 cm above the water-sediment interface and 6, 12, and 18 cm below the water-sediment interface. In the seagrass bed, measurements were conducted at 15 cm (above the canopy) and 5 cm (in the canopy) above the water-sediment interface, as well as 6, 12, and 18 cm below the water-sediment interface where the plant rhizome and root systems were present.

A source excitation waveform consisting of a 50 kHz, 20-cycle tone burst was produced by an arbitrary waveform generator, passed through a broadband power amplifier, then sent to the source transducer. The signal propagated through the medium (water, seagrass, or sediment) to the receiver. The measured signal was sent to a preamplifier and narrowband bandpass filter (50 kHz ± 1 kHz), and was then digitized and recorded on a computer. The signals produced by the waveform generator were also digitized and recorded in a time-synchronous manner for post-processing. For each measurement, 64 waveforms were recorded and averaged to improve the signal-to-noise ratio.

The source and receiver transducers were positioned in the water so that both were 15 cm above the bare sediment, to calibrate the source-receiver separation distance. The time-of-arrival t_w of the signal from the source to receiver was estimated from a cross-correlation between the excitation signal and the received signal in the water column. Water temperature and salinity were measured at the experiment site to obtain a value of sound speed in the water column, which was $cw=1525.6$ m/s. The measured signal time-of-arrival and sound speed in water were used to calibrate the source-receiver distance $d=cwtw=16.63$ cm, which is in good agreement with the transducer separation distance measured by a ruler at the experiment site.

A cross correlation technique was used to estimate the group speed in the bare sediment, the seagrass canopy, and the seagrass-bearing sediment.¹² The source-to-receiver travel times in the sediment and seagrass t_s were estimated using the cross correlation between the excitation signal and the received waveform. The travel time difference between the seawater path and the propagation path for each case $Δt=tw−ts$ was computed and then combined with the seawater sound speed and probe separation distance to estimate the group speed

In the absence of dipsersion, Eq. (1) yields the sound speed. Bias error can occur in the estimate of the cross correlation time delay when the single-source/single-receiver configuration is used because the excitation signal and received waveform do not have the same shape.¹³ This mismatch in waveform shape occurs primarily because the bandpass parameters of the source are different between the water and the sediment/seagrass. Dispersion and attenuation as the signal propagates through the medium can also have an effect. Both mechanisms can result in inaccuracy in the arrival time estimate. In an attempt to account for this bias error, the timing uncertainty in the water column $σtw=3.8×10−6$ s was estimated from the difference between the leading edge of the waveform and the time delay obtained from the cross correlation. Values of the timing uncertainty in the bare sediment and in the seagrass σ_ts were of the same order of magnitude as that in the water. The intrinsic timing uncertainty due to the sampling rate was negligible compared to the bias error. The uncertainty in the probe separation distance is given by $σd=cwσtw$ = 0.6 cm. From Eq. (1) and propagation of error, the uncertainty in the sound speed estimate is given by

The attenuation was estimated from the natural logarithm of the ratio of the received amplitude in the sediment or seagrass to that in pure seawater and the probe separation distance d,

where A_s and A_w are the root-mean-square amplitudes of the steady-state portion of the received waveforms recorded in the sediment/seagrass and seawater propagation paths, respectively. The geometrical spreading for each path is assumed to be equal since the probe separation distance remained constant throughout the experiment; therefore, the effect of spreading loss cancels out and Eq. (3) yields an estimate of the intrinsic attenuation in the medium. The uncertainty in the attenuation estimate depends on the error in the probe separation as well as the error in the amplitude measurements $σAs$ and $σAw$⁠, which were estimated from the standard deviation of the steady-state amplitudes for the sediment/seagrass and seawater cases, respectively. From Eq. (3) and propagation of error, the uncertainty in the intrinsic attenuation estimate is given by

Additional bias error in the estimate of attenuation in the sediment is present because the loading of the source transducer by the sediment changes the initial pressure amplitude within the sediment compared to that in water.¹³ Although this likely introduces an offset in the absolute value of attenuation measured in the sediment (both bare and seagrass-bearing), the difference in attenuation between the two cases produces a reliable estimate for the attenuation increase in the sediment due to the seagrass. For the attenuation estimates in the water column where seagrass was present, the source transducer loading is assumed to be the same as the water with no seagrass; therefore, the absolute attenuation estimates here are more accurate.

Two cores each were hand-collected in both the bare-sediment patch and the seagrass bed. The cores consisted of 7.62-cm-diameter, 30.5-cm-long plastic core liners, which were inserted manually into the sediment to a depth of approximately 20 cm and then removed, keeping the contents intact. The cores were capped underwater, stored vertically in an ice-filled cooler aboard the boat, and transported back to an onshore laboratory for analysis.

The cores were sectioned in approximately 2-cm increments to obtain the sediment bulk density, macroscopic plant biomass, and mean grain size M_z as a function of depth. The grain size is reported in units of $ϕ$⁠, where $ϕ=−log2(d/d0)$⁠, d is grain size in millimeters, and $d0=1$ mm is a reference grain size. The total wet mass of each section was measured on a laboratory scale after it was removed from the core and divided by the volume of the core sub-section to obtain estimates of bulk density. The macroscopic plant matter was separated from each section, dried in an oven, and weighed to acquire estimates of biomass. A wet-sieving procedure was then used to separate particles $≤500$ μm from the coarser grains and shell fragments, and the separated sediment fractions were dried in an oven. A dry-sieving technique was employed to determine the grain size distribution of the coarse-grain fraction, and the size distribution of the fine-grain fraction was determined using a laser diffraction particle size analyzer. The size distributions of the two fractions were combined,¹⁴ and the mean grain size M_z was obtained using the graphical method of Folk and Ward.¹⁵ 

Comparison of the acoustic data obtained from the in situ measurements and the physical properties obtained from the core analysis are shown in Fig. 1. The sound speed ratio (relative to the measured sound speed in seawater at the experiment location) and attenuation at 50 kHz are plotted as a function of depth in Figs. 1(a) and 1(b). The average values of macroscopic plant biomass, bulk sediment density, and mean grain size M_z (given in units of $ϕ$⁠) are plotted in Figs. 1(c)–1(e) with the horizontal error bars indicating the range in measured values between the two cores at each site.

The depth-dependence of the sound speed and attenuation profiles are markedly different between the two measurement locations. Well above the seagrass canopy, the sound speed and attenuation are similar to that in seawater above the bare-sediment patch. Upon entering the top-portion of the seagrass canopy, the two data sets begin to diverge. At 5 cm above the water-sediment interface, the sound speed ratio reduces to 0.97 and the attenuation increases to 104 dB/m relative to open water. In the bare sediment patch, the sound speed ratio in the sediment takes on values greater than unity, consistent for fine-sand sediments.¹⁶ At 6 cm into the seagrass-bearing sediment, the sound speed ratio is reduced to a value of 0.37 and the attenuation difference (red dotted line) takes on a maximum value of 285 dB/m relative to the bare sediment. At further increasing depth into the seagrass-bearing sediment, the attenuation decreases and the sound speed approaches that of the bare-sediment.

The acoustic behavior correlates to the macroscopic biomass profiles obtained from the cores. The greatest reduction in sound speed and increase attenuation in the seagrass bed occur in the uppermost layer of the sediment, where the biomass content is most significant. In contrast, the abundance of macroscopic plant biomass in the bare-sediment patch is negligible, taking on values less than 8 g/cm². The bulk density is approximately 5% to 15% lower in the seagrass cores than in the bare sediment cores for the uppermost 10 cm layer. Additionally, the mean grain size in the upper 10 cm varies between 1.8 $ϕ$ and 2.6 $ϕ$ in the seagrass cores compared to 0.8 $ϕ$ to 1.3 $ϕ$ in the bare sediment cores, which are characteristic of medium-to-fine sand and coarse-to-medium sand, respectively. Changes in the sediment physical parameters of this magnitude alone are not expected to result in such drastic changes to sound speed and attenuation as were observed in the experiment.¹⁶ Therefore, the seagrass tissue and any associated gas volumes within the tissue likely have a dominating influence on the acoustic properties near the water-sediment interface in the seagrass bed. Even well below the seagrass-bearing sediment layer where little significant biomass was found, the attenuation difference between the two sites is still significant (140 dB/m at 18 cm depth into the sediment). The rhizome aerenchyma allow diffusion of oxygen into the sediment to occur. Additionally, significant organic matter resulting from the decomposition of plant and other material accumulates beneath seagrass beds. Both of these processes have the potential to affect the acoustic properties of the sediment, even in the absence of detectable macroscopic biomass.

In situ acoustic measurements were conducted to determine the effects of seagrass on the acoustic propagation in the vicinity of the water-sediment interface. Measurements of sound speed and attenuation at 50 kHz were conducted in a bare-sediment patch and in a nearby bed of T. testudinum. Cores were collected to provide measurements of macroscopic biomass, bulk sediment density, and mean grain size for correlation with the acoustic data. These preliminary results indicate that seagrass can have profound effects on the sound speed and attenuation not only within the water column, as noted by previous work,^3,4,7–9 but also within the sediment beneath the seagrass. Future work will include investigations of dispersion, shear wave propagation within the sediment, and differences between various seagrass species. Additionally, the multiple receiver measurement techniques described in Refs. 11 and 13 will be employed in future experiments so that increased accuracy is achieved in the sound speed and attenuation estimates.

This work was supported by the U.S. Navy Office of Naval Research Ocean Acoustics and Ocean Engineering Programs and the ARL:UT Independent Research and Development Program. The authors are greatly thankful for logistical support provided by Professor Kenneth H. Dunton and piloting of the research vessel by Kimberly Jackson, both of UT:MSI. We also thank Jeremy King of ARL:UT for processing the core samples and Dan Duncan of the University of Texas at Austin Institute for Geophysics for assistance with the laser diffraction particle size analyzer.