Passive Acoustic Monitoring: How Marine Scientists Listen to the Ocean

Passive Acoustic Monitoring (PAM) is the observational study of marine environments using autonomous or cabled hydrophones to detect, localize, and monitor vocalizing animals and ambient ocean soundscapes. Unlike active acoustic systems (such as naval sonar or seismic airgun surveys that introduce high-energy pulses into the water column), passive acoustic sensors listen without transmitting acoustic energy into the sea. While the sensors themselves do not transmit sound or interfere with biosonar, field biologists must account for potential localized disturbance caused by support vessels during mooring deployment or array towing.

Because seawater absorbs light rapidly, visual observation of cetaceans (whales, dolphins, and porpoises) is restricted to daylight hours, calm sea states, and brief moments when animals surface to breathe. PAM overcomes these constraints, providing continuous, longitudinal observation across seasons, storms, polar winters, and deep oceanic trenches.

Deployment Architectures and Marine Hardware

In their seminal review of fixed passive acoustic observation methods, Dr. David Mellinger and colleagues categorized marine acoustic monitoring into distinct hardware platforms:

[ Ocean Passive Acoustic Monitoring Platforms ]
 ├── 1. Autonomous Seafloor Moorings (SoundTrap, HARP, EAR)
 │    └── Battery-powered; acoustic release recovery; deployments up to 1–2 years
 ├── 2. Cabled Seafloor Observatories (OOI, Ocean Networks Canada)
 │    └── Continuous fiber-optic telemetry & real-time shore streaming
 ├── 3. Towed Hydrophone Surface Arrays
 │    └── Multi-channel spatial arrays towed behind visual survey vessels
 └── 4. Mobile Autonomous Vehicles (Wave Gliders, Slocum Gliders, Drifters)
      └── Spatial survey coverage following oceanographic current boundaries

1. Autonomous Seafloor Moorings

Anchored to the benthic seabed by heavy weights, autonomous recorders float vertically in the water column using subsurface syntactic foam buoys. An acoustic release mechanism triggered from the surface allows the instrument to detach from its anchor and float to the surface for recovery. Deployments range from several months to two years, constrained primarily by battery capacity and internal solid-state storage.

2. Cabled Ocean Observatories

Networks such as the Ocean Observatories Initiative (OOI) and Ocean Networks Canada provide continuous power and high-speed fiber-optic data transmission directly to shore. These instruments stream uncompressed, multi-channel acoustic data in real time, enabling long-term monitoring of seismic events, baleen whale migration corridors, and marine heatwave anomalies.

3. Towed Hydrophone Arrays

Research vessels tow multi-channel hydrophone cables at distances of 100 to 400 meters behind the stern to distance the sensors from vessel engine noise. Time-difference-of-arrival (TDOA) between hydrophone elements provides left-right ambiguous bearings to vocalizing dolphins, which are combined with vessel tracking to triangulate precise animal coordinates.

Physics of Underwater Acoustic Propagation

The effectiveness of PAM depends on how acoustic energy travels through seawater:

1. Sound Speed in Seawater

Sound speed in water (c) is approximately 1,500 m/s, varying between 1,450 m/s and 1,540 m/s as a function of temperature (T), salinity (S), and depth/pressure (D). Herman Medwin formulated the standard empirical equation:

c ≈ 1449.2 + 4.6T - 0.055T² + 0.00029T³ + (1.34 - 0.010T)(S - 35) + 0.016D

2. Transmission Loss (TL)

As sound spreads from a source, underwater transmission loss decreases acoustic intensity according to geometric spreading and frequency-dependent absorption:

• Spherical Spreading (Deep Water): Where water depth exceeds distance, sound radiates omnidirectionally:

TL = 20 · log₁₀(r) + α · r

• Cylindrical Spreading (Shallow Coastal Water): Between the sea surface and sea floor, acoustic waves are channeled:

TL = 10 · log₁₀(r) + α · r

(where r is distance in meters, and α is frequency-dependent chemical absorption in dB/m).

Because high frequencies attenuate much faster than low frequencies (due to boric acid and magnesium sulfate relaxation in seawater), low-frequency baleen whale calls (below 100 Hz) can propagate hundreds of kilometers, whereas ultrasonic dolphin clicks (above 100 kHz) attenuate within hundreds of meters to a few kilometers.

Estimating Animal Density from Passive Acoustics

Modern bioacoustics does not stop at detecting species presence; it estimates absolute population density (D = animals per square kilometer) without visual sightings.

In a comprehensive statistical framework published in Biological Reviews, Tiago A. Marques, Len Thomas, and colleagues formalized the distance sampling framework for passive acoustics:

The acoustic point-transect density estimator is formulated as:

D̂ = [ n · (1 - ĉ) ] / [ K · π · w² · P̂ · T · r̂ ]

Where:

• n is the total count of detected vocal cues.
• ĉ is the estimated false positive detection rate, measured via manual ground-truth audits of detector output samples.
• K is the number of monitoring hydrophone stations.
• w is the truncation radius around each sensor beyond which cues are excluded.
• P̂ is the estimated probability of detecting a cue produced within the monitored area $\pi w^2$.
• T is the total cumulative recording duration.
• r̂ is the mean individual acoustic cue production rate (cues per animal per unit time).

Key Methodological Assumptions

For the acoustic density estimate to be biologically valid, several rigorous conditions must be satisfied:

1. Cue Rate Stability: The cue production rate (r̂)—typically calibrated from digital acoustic recording tags (DTAGs) on focal animals—must match the behavioral context (e.g. foraging dive bouts versus resting) and demographic composition of the monitored population.
2. Detection Function Calibration: Detection probability (P̂) depends heavily on ocean sound propagation conditions, ambient noise fluctuations, and cue source level distributions. It assumes that detection probability directly at the sensor ($r = 0$) is known or close to certainty.
3. Spatial Independence: Animals must be distributed independently of sensor locations; moorings must neither attract curious individuals nor cause avoidance.
4. Group Size Decoupling: In social delphinids where vocalizations occur in overlapping group choruses, individual cue rates must be decoupled from group size estimates to prevent severe density overestimation.

Conclusion

Passive acoustic monitoring has become the foundational observational tool of marine bioacoustics. By combining hydrophone engineering, underwater acoustic physics, and statistical density estimation, marine scientists listen to ocean basins continuously—providing the empirical data that trains modern computational models like DolphinGemma and Project CETI.