Proceedings of the Institute of Acoustics

 

Vector acoustic observations and modeling of underwater noise from a ship closing towards the sensor location over a 20 km range

 

P. H. Dahl, Applied Physics Laboratory, University of Washington, Seattle, USA
D. R. Dall’Osto, Applied Physics Laboratory, University of Washington, Seattle, USA

 

1 INTRODUCTION

 

Vector acoustic observations of underwater noise emissions from a merchant ship are made as function range in waters of depth 77 m on the New England Mud Patch, an area approximately 100 km south of Cape Cod, MA. The measurements took place in conjunction with the 2022 Seabed Characterization Experiment (SBCEX). The primary objective of SBCEX is the study of sound propagation involving sea floors consisting of fine-grained sediments; in this region manifested as a mud-like layer of order 10 m thickness, commencing at the water-sediment interface (Fig. (1)). The observations were made with the Intensity Vector Autonomous Recorder (IVAR2) a system that records four coherent channels of acous- tic data continuously, one channel for acoustic pressure and three for acoustic acceleration from which acoustic velocity is obtained.

 

Stratigraphic analysis of experimental area identifies sub bottom layer structure over the first ∼ 50 m of seabed commencing with the first, low-speed, mud layer. The observed ship transect (via AIS) spans more than 20 km in closing and 10 km in opening about IVAR position, and the stratigraphy data over this range span is incorporated in the form of adiabatic modeling. The focus of this presentation is on low frequency observations (approximately 25 Hz) that probe deeper sediments beyond this observed stratigraphy.

 

The paper is organized as follows: In Sec. 2a brief description of the data based on the ship transit through the SBCEX experimental area is given along with details on data processing. Section 3 gives a geoacoustic description of the upper sediments based in part on the known stratigraphy and informed by several SBCEX published results, and describes a proposed geoacoustic model for the deeper sediments. Evidence for characterizing a large ( ∼ 1000 m) portion of deeper sediments with compressional sound speed increasing approximately linearly with depth is provided in Sec. 4 by way of observations and modeling of vertical kinetic energy. Conclusions are provided in Sec. 4.

 

2 DATA DESCRIPTION

 

The M/V Glen Canyon is a 285-m container vessel, powered by a 11-cylinder diesel-electric engine. The IVAR2 sensor was located at 40.4416 ^◦ N, 70.5282 ^◦ W (during this particular deployment of SBCEX), with the closest point of approach (CPA) of the Glen Canyon with IVAR2 occurring at 21:03 UTC 22 May, 2022, while at a speed pf approximately 20 kt. The CPA as determined by Automatic Identification System (AIS) data is of order 10 m from the IVAR2 locaion on the seabed in terms of horizontal range on the sea surface, although such a small length scale relative to size of the ship can remain indeterminate without impacting this analysis.

 

 

Figure 1: Experimental area for SBCEX approxiimately 100 km south of Cape Code, MA (blue rectangle). Insets show path of the M/V Glen Canyon over the IVAR2 sensor (green dot) in terms mud thickness as linked to two-way travel based on time high-resolution chirp sonar (left) and water depth (right).

 

 

Figure 2:  Acoustic pressure as a function of time and frequency for the transit of the M/V Glen Canyon over the IVAR2 sensor. Time is relative to CPA with negative and positive time represent- ing closing and opening in range from IVAR2, respectivley. Inset shows the spectral line near 25.4 Hz during closing phase and downshift to 25.13 Hz during opening phase. Plot color scale is relative pressure spectral level, increasing in level from blue, green to red.

 

A time-frequency analysis of the ship passage closing and opening about CPA (Fig. 2) shows broad band and line spectral components. The latter are of primary interest, with the 25.4-Hz spectral line being representative. In view of the speed, this line represents an upshifted frequency from the original source emission near 25.28 Hz. The upshifted-line undergoes a small down-shift of ∼ 0.27 Hz, consistent with the approximate propagation angle, ∼ 30 ^◦, relative to horizontal near CPA.

 

2.1 Potential and Kinetic Energy Estimates

 

In this study, vector acoustic observations of the underwater ship noise at the IVAR2 site are limited to estimates of complex, narrow band acoustic pressure p̂(t ) and acoustic velocity V̂_x,y,z (t ) in the x, y, z directions as a function of time t. The center frequency of the spectral line (Fig. 2) subject to further analysis is 25.4 Hz with bandwidth ± 0.25 Hz, although numerous other spectral lines associated with underwater noise emissions from the ship are available for study. Data are initially recorded at sample rate 16000 Hz, and baseband demodulated and decimated to a sample rate of 100 Hz for this low-frequency analysis. From these quantities the acoustic kinetic energy E_k, and the vertical component kinetic energy E_kz  are evaluated as follows:

 

 

and

 

 

where ρ_w is sea water density equal to 1028 kg/m³. For reference the acoustic potential energy E_p(t ) is evaluated as

 

 

where c_w is sound speed measured on the IVAR2 platform and equal to 1491 m/s. The time T=12s, is the period over which a moving average is taken to yield the energy quantities as a function of time t, which are expressed as their decibel equivalents in dB re pJ/m³.

 

3 UPPER SEDIMENT STRUCTURE AND MODEL FOR DEEP SEDIMENTS

 

The upper sediment structure in the SBCEX experimental area has been studied extensively using a combination of high-resolution chirp acoustic reflection data¹, e.g., producing the map of two-way travel time data in Fig. 1, along with sediment cores and various acoustic propagation measurements providing data for inversion of seabed properties^2,3. Three prominent horizons from the two-way travel map in Fig. 1 are extracted corresponding to the course of the M/V Glen Canyon closing towards IVAR2 from range 20 km and opening another 10 km [Fig. 3 (a)] The travel time horizons have been converted to three primary geoacoustic descriptors: mud, mud-to-sand, and sand, along with layer thicknesses based on sound speeds informed by these studies and others^4,5. The upper sediment structure directly below the IVAR2 sensor is of particular interest with details provided in Table 1

 

Table 1: Geoacoustic model for upper sediment structure directly below the IVAR2 sensor

 

Several if not most geoacoustic representations of the sediment structure in the SBCEX experimental site published thus far terminate with a half space commencing at a sediment depth of order 100 m for which the sound speed is typically between 1800 and 2000 m/s. This parsimonious representation for deeper sediments is of course completely justified in terms of the information content of the acoustic observations subject to analysis geoacoustic inversion, which in turn depends on frequency range and measurement type. However, a recent study⁶ based on inversion of low-frequency (less than 100 Hz) ship noise from the SBCEX site suggests higher speeds within the deeper sediments. Moreover, a 1996- era⁷reconstruction of sonabuoy data for an area approximately 50 km east of the SBCEX site also points to evidence for similarly higher speeds. Though both studies provide resolution at only very large length (depth) scales of order 100 m or greater, a proposed model for the deep sediment compressional sound speed as function of depth is nonetheless in part informed by the two studies [Fig. 3 (b)]. The model also includes a basement halfspace commencing at depth 2000 m, which is assigned compressional speed of 5000 m/s. This depth corresponds to the two-way travel time estimate (1550 ms) as determined by our team using Mk128 SUS explosive charges deployed during the SBCEX experiment. 

 

 

Figure 3: (a) Range-varying layer thicknes of the upper sediment structure in terms of descriptors mud, mud-to-sand, and sand corresponding to the course-over-ground of the M/V Glen Canyon closing towards IVAR2 sensor (black square). (b) Proposed model for the compressional sound speed in the deep sediments; yellow shade identifies an approximate linear gradient of 0.8 m⁻¹

 

A prominent feature of the profile in Fig.3(b) is the continuous increase of compressional speed with increasing depth over a 1500-m layer commencing at the bottom of the sand layer [Fig.3(a)]. Within this layer the compressional speed c(z) varies linearly in 1/c²(z), where in this case z is depth into the layer. Parameterized in this manner the compressional sound speed increases approximately linearly as in c(z) ≈ c0 + gz, where c0 is speed at the top of the layer (1800 m/s) and linear gradient parameter g = 0.8 s⁻¹. The profile Fig.3(b), referred to as 1/c²-linear is designed to operate most effectively with the the normal mode code ORCA⁸ which is used in our modeling approach based on normal modes. In the following we provide subtle but durable evidence for a gradient in compressional sound speed based on observations and modeling of vertical kinetic energy.

 

4 EVIDENCE FOR DEEP SEDIMENT SOUND SPEED GRADIENT BASED ON OBSERVATIONS AND MODELING OF VERTICAL KINETIC ENERGY DENSITY

 

Figure 4(a) displays the vertical component kinetic energy E_kz  at frequency centered at 25.4 Hz as a function range closing towards IVAR2, where CPA is reached at range = 0. Note that relatively precise association of range to the IVAR2 site, with observation time (e.g., with time given Fig. 2) is possible due to the precise (and unchanging) course, and speed of the M/V Glen Canyon, as determined by both AIS data and bearing observations made from IVAR2. The pattern of E_kz  over the closing 20 km observation pattern appears to be dominated by 2 modes, however four shaded regions separated by approximately 4 km identify subtle perturbations of the 2-mode pattern. This is most easily seen for the two regions closest to CPA where energy levels are higher. We postulate that the perturbations are caused by quasi-spatially periodic arrivals having been refracted upward from the deep sediment; in short they can be referred to as refracted arrivals. 

 

 

Figure 4: (a) Measured vertical component kinetic energy E_kz at frequency centered at 25.4 Hz as a function range closing towards IVAR2, where CPA is reached at range = 0. Four shaded areas separated by ∼ 4 km identify glimpses of refracted arrivals that produce small but notable changes in the 2-mode interference pattern. (b) Corresponding modeled results showing the same four shaded areas based a complete mode summation (sold line) and partial sum based on modes 9-26 (dashed line). (c) Measured total kinetic energy E_k (black line) and corresponding modeled result (red line).

 

A model for E_kz [Fig. 4(b)] using a mode-based formulation for which the essential modal quantities as a function of properties of the waveguide is generated with the normal mode code ORCA. From these quantities we compute both dynamic (pressure) and kinematic (velocity) properties of the acoustic field⁹. The model incorporates the range dependence as embodied by the upper sediment structure [Fig.3(a)] including slowly varying water depth, while assuming the deep sediment structure [Fig.3(b)] is range independent. This range dependent result is achieved using adiabatic mode theory¹⁰. Note that a nominal shear speed profile was also accommodated in the modeling, however its effect on the modeling results is not strongly evident. 

 

The model result (solid, red line) is necessarily scaled by a single constant as the exact source level of the ship is not subject of this study. More importantly, however, the model result for E_kz shows a similar pattern of perturbation of the 2-mode pattern over approximately the same 4-km interval. For example, the shaded areas in Fig.4(a) are repeated in Fig. 4(b). The model result also more clearly illustrates a 2-mode pattern with exception of the shaded areas; for example summing just the first two modes yields nearly the same 2-mode pattern absent the perturbations. For a given frequency, the profile in Fig.3(b) supports a greater number of modes relative to one terminated by a halfspace below the sand layer in Fig.3(a). An analysis of the contribution-strength of higher-order modes suggest that modes 9 through 26 dominate, and summing this contribution yields the second result in Fig. 4(b) shown by the dashed, red line. The group of higher-order modes, taken together, assume a ray-like appearance responding to convergence zones separated by ∼ 4 km. 

 

Finally, Fig. 4(c) compares model result with corresponding observations of the total kinetic energy E_k at 25.4 Hz using same scaling constant as Figs. 4(a) and (c). With total kinetic energy the effects of the gradient are less evident; however the correspondence with between model and data over this 20 km range nonetheless is quite reasonable. The correspondence degrades somewhat between ranges 14 and 9 km before CPA, and reasons for this will be examined in future work.

 

5 CONCLUSION

 

Vector acoustic observations of underwater noise emissions from a merchant ship at 25.4 Hz made as part of the 2022 SBCEX experiment are presented as a function of range from a vector sensor, over a 20 km span during which the vessel closed on the sensor location at constant speed and course. The measurements were made in waters of depth 77 m on the New England Mud Patch for which upper sediment structure stratigraphy has been heavily studied and consists of mud-like layer over the first ∼ 10 m of sediment. The low frequency observations from this study made possible the probing of deep sediment structure beyond that of the more studied upper sediment stratigraphy. The observations in the form of kinetic energy density and its component associated with vertical velocity centered about 25.4 Hz provide subtle but durable evidence that can be interpreted as refracted arrivals converging every ∼ 4 km. 

 

Based on this interpretation, and informed by additional information from two studies^6,7, we postulate a compressional speed profile in [Fig.3(b)] that embodies a continuous increase of compressional speed with increasing depth over a 1500-m layer which will produce upward refraction. The profile is also made consistent with a two-way travel time estimate determined team using Mk128 SUS explosive charges deployed during the SBCEX experiment. Modeling of the equiivalent observations based on normal modes provides additional confirmation of refracted arrivals that are associated with higher-order modes of the acoustic field. 

 

ACKNOWLEDGMENTS

 

This study has been funded by the U.S. Office of Naval Research.

 

REFERENCES

 

1. J. A. Goff, A. H. Reed, G. Gawarkiewicz, P. S. Wilson, and D. P. Knobles, “Stratigraphic analysis of a sediment pond within the New England Mud Patch: New constraints from high-resolution chirp acoustic reflection data,” Marine Geology, Vol. 412, pp. 81–94, (2019).

2. P. S. Wilson, D. P. Knobles, and T. B. Neilsen, “Guest Editorial: An Overview of the Seabed Characterization Experiment,” IEEE Journal of Oceanic Engineering, Vol. 45, No. 1, pp. 1–13, (2020).

3. P. S. Wilson, D. P. Knobles, and T. B. Neilsen, “Guest Editorial: Continued Exploration of Fine-Grained Sediments from SBCEX2017,” IEEE Journal of Oceanic Engineering, Vol. 47, No. 3, pp. 497–502, (2022).

4. D. P. Knobles, D. C. Escobar-Amado, M. J. Buckingham, W. S. Hodgkiss, P. S. Wilson, T. B. Neilsen, J. Yang, and M. Badiey, “Statistical inference of sound speed and attenuation dispersion of a fine-grained marine sediment,” IEEE Journal of Oceanic Engineering, Vol. 47, No. 3, pp. 553–564, (2022).

5. S. E. Dosso and J. Bonnel, “Hybrid seabed parameterization to investigate geoacoustic inversion in the New England Mud Patch,” IEEE Journal of Oceanic Engineering, Vol. 47, No. 3, pp. 620–634, (2022).

6. D. P. Knobles, T. B. Neilsen, P. S. Wilson, W. S. Hodgkiss, J. Bonnel, and Y. T. Lin, “Maximum entropy inference of seabed properties using waveguide invariant features from surface ships,” Journal of the Acoustical Society of America, Vol. 151, pp. 2885–2896, (2022).

7. J. Diebold, Lamont–Doherty Earth Observatory, (1996).

8. E. K. Westwood, C. T. Tindle, and N. R. Chapman, “A normal mode model for acousto-elastic ocean environments,” Journal of the Acoustical Society of America, Vol. 100, pp. 3631–3645, (1996).

9. P. H. Dahl, D. R. Dall’Osto, and W. S. Hodgkiss, “Active intensity vortex and stagnation point singularities in a shallow underwater waveguide,” Journal of the Acoustical Society of America, Vol. 154, pp. 1482–1492, (2023).

10. F. B. Jensen, W. A. Kuperman, M. B. Porter, and H. Schmidt, Computational Ocean Acoustics, 2nd ed., American Institute of Physics, New York, pp. 408–410, (2011).