The Effect of Attenuation from Fish on Passive Detection of Sound Sources in Ocean Waveguide Environments

Abstract

Attenuation from fish can reduce the intensity of acoustic signals and significantly decrease detection range for long-range passive sensing of manmade vehicles, geophysical phenomena, and vocalizing marine life. The effect of attenuation from herring shoals on the Passive Ocean Acoustic Waveguide Remote Sensing (POAWRS) of surface vessels is investigated here, where concurrent wide-area active Ocean Acoustic Waveguide Remote Sensing (OAWRS) is used to confirm that herring shoals occluding the propagation path are responsible for measured reductions in ship radiated sound and corresponding detection losses. Reductions in the intensity of ship-radiated sound are predicted using a formulation for acoustic attenuation through inhomogeneities in an ocean waveguide that has been previously shown to be consistent with experimental measurements of attenuation from fish in active OAWRS transmissions. The predictions of the waveguide attenuation formulation are in agreement with measured reductions from attenuation, where the position, size, and population density of the fish groups are characterized using OAWRS imagery as well as in situ echosounder measurements of the specific shoals occluding the propagation path. Experimental measurements of attenuation presented here confirm previous theoretical predictions that common heuristic formulations employing free space scattering assumptions can be in significant error. Waveguide scattering and propagation theory is found to be necessary for accurate predictions.

1. Introduction

Acoustics is the primary means of sensing self-radiating sources in the ocean such as manmade vehicles, geophysical phenomena, and vocalizing marine mammals [1,2,3]. Attenuation from fish, however, can significantly decrease detection range for passive sensing in the ocean [4,5]. This makes it important to accurately predict attenuation from fish in long-range passive sensing. Attenuation from fish has been previously measured from fluctuations in the intensity of long-range acoustic signals caused by diel or seasonal shoaling patterns of the fish, including sardines in the Bristol Channel and groups of anchovies and sardines in the Gulf of Lion [4,5,6,7]. In these experiments fish population densities were inferred from the reductions in signal intensity, since it is difficult to measure instantaneous population distributions across long ranges with conventional survey methods. More recently, wide area Ocean Acoustic Waveguide Remote Sensing (OAWRS) was used to observe fish shoals over thousands of square kilometers and simultaneously measure attenuation from these shoals within the active OAWRS transmissions [8]. Here the effect of attenuation from fish on passive ocean acoustic waveguide remote sensing (POAWRS) of ship-radiated tonals is investigated, where wide-area OAWRS imagery of herring groups in Norwegian spawning grounds is used in conjunction with in situ echosounder data to measure the size and population density of the fish groups occluding the propagation path.

POAWRS has been previously applied to detect, localize, and classify vocalizations from fin whales in the Norwegian Sea [9], baleen whale species and toothed whale species in the Gulf of Maine [10,11,12], and sperm whales along the US east coast [13] over continental shelf-scale regions. Temporal–spatial distributions of marine mammal vocalizations measured by POAWRS in the Gulf of Maine have been compared with herring population density distributions in order to provide insights into the predator–prey dynamics in that ecosystem, and it was shown that marine mammals will spatially converge on fish spawning grounds and divide into separate foraging areas specific to each marine mammal species [3].

POAWRS has also been used to detect and characterize ship-radiated sound from surface vessels [14,15,16]. During previous OAWRS surveys of spawning herring in the Gulf of Maine and spawning capelin in Finnmark, Norway, ship-radiated tonals from research vessels were continuously measured with POAWRS over more than 10 h and at ranges of up to 30 km [14]. During the OAWRS survey of Norwegian herring spawning grounds, POAWRS detections of ship-radiated tonals from a fishing vessel less than 20 km away were lost to ambient background noise for periods of more than 1.5 h [14]. Here it is experimentally and theoretically shown that these detection losses are caused by attenuation from the herring shoals. The conditions determining whether attenuation from fish will significantly affect passive sensing are investigated here, and they are found to be similar to previously-demonstrated conditions determining whether attenuation will affect active sensing [8]. Attenuation from herring shoals did not significantly affect ship-radiated tonals in the Gulf of Maine since the frequencies of the tonals were well below the resonance peak of the shoaling herring in the region. Attenuation from capelin shoals did not significantly affect tonals in the Finnmark region because the size and population density of the capelin shoals were not sufficiently high. No attenuation was observed in one-way acoustic transmissions during an OAWRS survey of herring in the Gulf of Maine [17], likely because the direct propagation path between the source and receiver did not cross the fish shoals and the transmission frequencies were below the swimbladder resonance peak of the shoaling fish [8].

Here, POAWRS is used to measure reductions in ship-radiated sound due to attenuation from fish. A previous investigation of attenuation from fish measured reductions in active OAWRS transmissions [8]. Reductions in the intensity of ship-radiated tonals due to attenuation from fish are predicted using an analytical formualtion for acoustic propagation and scattering through an ocean waveguide with inhomogeneities [18]. This formulation has been previously shown to be consistent with experimental measurements of attenuation from fish to active OAWRS transmissions [8], as well as measurements of attenuation and temporal coherence loss in the presence of surface gravity waves, near-sea-surface air bubbles and internal waves [19,20,21,22]. Reductions in ship-radiated sound measured here are consistent with the predictions of the waveguide attenuation model, where the position, size, and population density of the fish groups occluding the propagation path are fully characterized using OAWRS imagery as well as in situ echosounder measurements. Common heuristic approaches that employ free space scattering assumptions for attenuation from fish are also investigated. While it has been theoretically shown that these heuristic approaches can be in disagreement with formulations that incorporate waveguide scattering and propagation effects [8], here it is experimentally confirmed that the heuristic approaches can be in significant error for predicting attenuation in a waveguide environment.

It is experimentally shown that reductions in the intensity of ship-radiated tonals increase with the size and population density of fish groups occluding the propagation path, with reductions of more than 4 dB commonly observed for the Norwegian herring shoals studied here. These intensity reductions are shown to lead to prolonged detection losses as dense herring shoals continuously occlude the propagation path. Such intensity reductions and corresponding detection losses may be significant in other regions where dense fish groups congregate, and may present significant impediments to passive sensing of other acoustic sources such as vocalizing marine mammals and underwater vehicles.

2. Materials and Methods

2.1. Measuring Reductions in Ship Tonal Intensity Due to Fish-Attenuation

Ship-radiated tonals from the FV Artus are detected by the POAWRS system over a three-hour period during a survey of herring spawning grounds near Ålesund, Norway (Figure 1). Passive acoustic data is beamformed in the direction of the Artus using a large-aperture densely-sampled horizontal coherent receiver array towed behind the RV Knorr [14,15,16], where the relative bearing of the Artus is determined using GPS data from both ships. Concurrent wide-area scattering strength maps generated by the monostatic OAWRS system are used to monitor herring groups within a 30 km radius of the Knorr (Appendix B of [8]). Active OAWRS transmissions do not interfere with the passive measurements shown here since the frequency of the Artus tonal studied here (1584 Hz) is well above the OAWRS sensing frequency (955 Hz).

Significant reductions in the received level of ship-radiated tonals from the Artus are observed when three herring shoals occlude the propagation path between 3:15–3:26 on 21 February (Figure 2). The decrease in sound pressure level of the Artus tonal due to attenuation from fish $\mathrm{\Delta }SP{L}_{\mathrm{measured}}$ is measured as the decrease in the received level beamformed in the direction of the Artus ($RL$) corrected for transmission loss from spreading losses and seafloor scattering ($TL$) after the fish groups occlude the propagation path, according to

$$\mathrm{\Delta }SP{L}_{\mathrm{measured}}={(RL+TL)}_{\mathrm{no}\mathrm{fish}}-{(RL+TL)}_{\mathrm{fish}}$$

(1)

where ${(RL+TL)}_{\mathrm{no}\mathrm{fish}}$ corresponds to measurements where OAWRS imagery confirms that no significant fish groups occlude the propagation path from the Artus to the receiver (Figure 2A), and ${(RL+TL)}_{\mathrm{fish}}$ corresponds to measurements where significant fish groups are observed in OAWRS imagery along the propagation path (Figure 2B). The received level $RL$ is calculated from the average received intensity at frequencies between $f_c\pm B/2$, where $f_c$ = 1584 Hz is the center frequency of the Artus tonal and B = 27 Hz is the 3 dB-down bandwidth of the Artus tonal (Figure A3). Transmission loss from spreading losses and seafloor scattering ($TL$) is defined as

$$TL=10{log}_{10}\left({\left(4\pi \right)}^4〈{\left|G\left(\mathit{r}|{\mathit{r}}_{\mathit{S}};f,c\left({\mathit{r}}_{\mathit{w}}\right),d\left({\mathit{r}}_{\mathit{w}}\right)\right)\right|}^2〉\right)$$

(2)

where $G\left(\mathit{r}|{\mathit{r}}_{\mathit{S}};f,c\left({\mathit{r}}_{\mathit{w}}\right),d\left({\mathit{r}}_{\mathit{w}}\right)\right)$ is the Green function between the source location ${\mathit{r}}_{\mathit{S}}=({\mathit{\rho }}_{\mathit{S}},z_S)$ and the receiver location $\mathit{r}$. A parabolic equation model [23] is used to calculate the Green functions in the range-dependent environment, where the conditional expectation over the sound speed is determined by averaging five Monte-Carlo realizations. Each Monte-Carlo realization employs randomly-selected sound-speed profiles measured in the relevant environment (Figure A5) every 500 m along the propagation path. It has been previously shown that five Monte-Carlo realizations are sufficient to converge on the mean transmission loss in continental shelf waveguide environments using this method [24]. In the example shown in Figure 2 (3:15–3:26 on 21 February), ${(RL+TL)}_{\mathrm{no}\mathrm{fish}}$ is equal to 164.6 dB with a standard deviation of 1.6 dB, and ${(RL+TL)}_{\mathrm{fish}}$ is equal to 160.7 dB with a standard deviation of 1.4 dB. The decrease in the Artus tonal due to attenuation from fish $\mathrm{\Delta }SP{L}_{\mathrm{measured}}$ is then 3.9 dB with a standard deviation of 2.1 dB. The distance from the Artus to the receiver increases from 9.1 km to 12.5 km during this time period, leading to an increase in the magnitude of transmission loss of approximately 1.2 dB. Variations in the speed of the Artus were less than 0.1 m/s between 3:15 and 3:26 (Figure A4C). The bearing of the receiver with respect to the Artus changed by less than 1° between 3:15 and 3:26 (Figure A4B), demonstrating that the directivity of the Artus signal did not contribute to measured intensity reductions.

2.2. Predicting Reductions in Ship Tonal Intensity Due to Fish-Attenuation

Reductions in the intensity of ship-radiated tonals due to attenuation from fish are predicted using a normal-mode-based analytical theory for acoustic propagation and scattering through inhomogeneities in an ocean waveguide [8,18]. Predicted reductions depend on the population density distribution with respect to depth and range along the propagation path. Since it is difficult to precisely characterize variations in population density for an individual shoal with respect to depth and range along the propagation path, each shoal is characterized here assuming a constant areal population density $n_A$, as well as mean shoal depth $z_m$, shoal vertical thickness H, mean range along the propagation path ${\rho }_m$ and horizontal width along the propagation path W. These variables are determined using a combination of OAWRS scattering strength maps and in situ echosounder measurements of the three shoals (Figure 3).

The population density ($n_A$) and vertical distribution ($z_m$, H) of the herring shoals occluding the propagation path are determined from in situ echosounder measurements of the shoals. Echosounder measurements of volumetric population density $n_V$ and areal population density $n_A$ are calculated according to Appendix G of [8]. The shoals are segmented by selecting regions where the areal population density is greater than the critical population density at which herring groups were found to form (0.2 fish/m²) [25,26], and the mean population density $n_A$ for each shoal is determined by averaging across these segments (Figure 3D–F). The mean depth $z_m$ and vertical thickness H for each shoal are chosen so that the average volumetric population density with respect to depth $n_V\left(z\right)$ is greater than 10${}^{-3}$ fish/m³ at depths between $z_m-H/2$ and $z_m+H/2$ (Figure 3G–I).

The horizontal distribution of shoals along the propagation path (${\rho }_m$, W) is determined using OAWRS scattering strength measurements along the propagation path from the source (FV Artus) to the receiver (RV Knorr). OAWRS scattering strength is typically measured by correcting transmissions for source level, areal resolution footprint, spreading loss and seafloor attenuation [25,26,27,28], however the OAWRS transmissions in this environment have been previously shown to be affected by attenuation from herring shoals [8], so these measurements are referred to as “fish-attenuated scattering strength” (Appendix A). The mean range of each shoal along the propagation path ${\rho }_m$ and horizontal width of each shoal along the propagation path W are chosen so that the fish-attenuated scattering strength is 5.6 dB greater than background seafloor scattering at ranges between ${\rho }_m-W/2$ and ${\rho }_m+W/2$ (Figure 3J), where 5.6 dB is the standard deviation of an acoustic measurement after saturated multipath propagation when the time-bandwidth product is one [29,30]. Measured values of $n_A$, $z_m$, H, ${\rho }_m$, and W for the three herring shoals studied here are shown in

Table 1. Parameters Characterizing the three Herring Shoals in Figure 3.

	Areal Density (Fish/m2)	Shoal Depth (m)	Shoal Vertical Thickness (m)	Distance Along Propagation Path (km)	Shoal Horizontal Thickness (km)
Shoal 1	1.7	62.5	45	1.01	0.38
Shoal 2	3.7	72.5	25	2.81	0.22
Shoal 3	5.1	77.5	15	3.99	1.15

.

3. Results

The predictions of the waveguide attenuation model are found to be in agreement with measured reductions in the Artus tonal due to attenuation from herring, where the size, position, and population density of the three shoals occluding the propagation path are fully characterized by OAWRS imagery as well as in situ echosounder measurements (Figure 4). The reduction in sound pressure level $\mathrm{\Delta }SPL$ is modeled according to Equation (10) of [8], where the volumetric population density $n_V$ is expressed with respect to range $\rho$ and depth z assuming discrete shoals with constant population densities, according to:

$$n_V(\rho ,z)=\left\{\begin{array}{cc}(n_{A,k}/H_k)\hfill & {\rho }_{m,k}-W_k/2<\rho <{\rho }_{m,k}+W_k/2\&z_{m,k}-H_k/2<z<z_{m,k}+H_k/2\hfill \\ 0\hfill & \mathrm{otherwise}\hfill \end{array}\right.$$

(3)

where $n_{A,k}$ is the areal population density of shoal k, $z_{m,k}$ is the mean depth of shoal k, $H_k$ is the vertical thickness of shoal k, ${\rho }_{m,k}$ is the mean range of shoal k along the propagation path and $W_k$ is the horizontal width of shoal k along the propagation path (Table 1). In the upward-refracting underwater environment studied here, dominant lower order acoustic modes are concentrated in the upper water column. Since the observed fish groups are concentrated in the lower water column, these lower order modes will experience almost no attenuation from the fish groups (Figure 4F). Modeled $\mathrm{\Delta }SPL$ is found to be in agreement with measurements at the receiver depth used in this experiment. Free-space-like factored formulations that ignore these depth-dependent modal attenuation effects (Equation (14) of [8]) overestimate the loss due to attenuation from fish by more than 5 dB (Figure 4G).

Reductions in the intensity of the Artus tonal and corresponding detection losses are shown to be strongly correlated with dense herring groups occluding the propagation path, where the criteria for valid detections are defined in Appendix C. The population density of herring groups between the source and receiver is calculated at each time interval by interpolating an OAWRS population density map (Figure 5) between the known locations of the Artus and the Knorr. During a period between 3:28 and 5:07, the tonal from the Artus is not detected by the POAWRS system. This loss of detection is shown to be correlated with herring groups occluding the propagation path with horizontal widths greater than 1 km and population densities above 0.2 fish/m², which is the critical population density at which large herring shoals were found to form [25,26] (Figure 6).

Reductions in the Artus tonal due to attenuation from fish are predicted between 3:10 and 6:00 from population density measurements using Equation (10) of [8]. In the absence of individual echosounder measurements of herring depth for each shoal, attenuation is modeled assuming the same depth distribution for all shoals. The volumetric population density $n_V$ with respect to range $\rho$ and depth z is then expressed as:

$$n_V(\rho ,z)=n_A\left(\rho \right)p\left(z\right)$$

(4)

where $n_A\left(\rho \right)$ is the areal population density measured by OAWRS (Figure 5) and $p\left(z\right)$ is the probability distribution function of herring depth, which is determined from echosounder measurements during the experiment and explicitly defined in Appendix B. Predicted reductions in the intensity of the Artus tonal are shown to be consistent with measured reductions and corresponding detection losses (Figure 7). Variations in ship speed are less than 0.5 m/s during this time period, indicating that changes in the source level of the Artus tonal over time are not significant enough to lead to the detection losses observed here (Figure A4).

While attenuation from herring in Ålesund waters led to prolonged detection losses for tonals from the FV Artus, attenuation from fish did not affect POAWRS detections of ship-radiated tonals during previous surveys of herring in the Gulf of Maine and Norwegian capelin off the coast of Finnmark [14]. Reductions in the intensity of ship-radiated tonals due to attenuation from fish is predicted in all three environments, and the presence or absence of significant attenuation is found to be consistent with theoretical predictions. Attenuation is negligible for tonals from the RV Delaware II in the Gulf of Maine since the frequencies of the tonals were well below the resonance peak of the spawning herring in that environment (Figure 8A). Predicted losses due to attenuation from Finnmark capelin are less than 1 dB because the horizontal width of observed capelin shoals is not large enough to lead to significant attenuation (<0.25 km, Figure 8B). Significant attenuation from Ålesund herring is predicted at the frequency of the tonal from the FV Artus, which is consistent with observed intensity reductions during the survey (Figure 8C). This analysis demonstrates that the waveguide attenuation formulation can be used to determine the conditions for significant attenuation from fish for long-range passive sensing in ocean waveguide environments. Predicted losses due to attenuation from fish are used to estimate the source level of the Artus signal in Appendix D.

4. Discussion

The effect of attenuation from fish on passive sensing of ship-radiated tonals is investigated, and it is shown that measured reductions in ship-radiated tonals are in agreement with a normal-mode based formulation for acoustic propagation through inhomogeneities in a waveguide. Common heuristic approaches that employ free space scattering assumptions disagree with measurements by more than 5 dB. While the potential for such heuristic approaches to incorrectly predict attenuation from fish in a waveguide environment has been previously discussed in theory [8], to our knowledge the results shown here are the first experimental evidence that attenuation formulations ignoring waveguide physics can be in significant error. It is shown that the waveguide attenuation formulation can be used to determine the conditions for significant attenuation for long-range passive sensing of self-radiating sources in the ocean such as manmade vehicles, geophysical phenomena, and vocalizing marine mammals. Attenuation from fish may also limit the ability of vocalizing marine mammals to echolocate in regions with dense fish shoals [31,32].

5. Conclusions

Following the theoretical work in Ratilal and Makris [18] and Duane et al. [8], it is experimentally shown that attenuation from dense herring groups in Norwegian spawning grounds can reduce the intensity of ship-radiated tonals and lead to prolonged detection losses for passive sensing of surface vessels in an ocean waveguide. Such detection losses can create significant impediments to passive acoustic sensing of underwater vehicles, marine life, and geophysical phenomena. Here, wide-area OAWRS imagery is used in conjunction with echosounder data to measure the size, position, and population density of every major fish shoal occluding the propagation path from source to receiver. Reductions in signal intensity due to attenuation from fish are predicted using an analytical theory for acoustic propagation through inhomogeneities in an ocean waveguide, and measured intensity reductions are found to be in agreement with theoretical predictions. Common heuristic approaches that employ free space scattering assumptions for attenuation from fish groups are found to be in disagreement with measurements.