Underwater Acoustics FAQ

Sound is how the ocean is observed. Seawater blocks light and radio within meters but carries sound for kilometers, which makes acoustics the primary tool for studying marine life, monitoring human activity and measuring the ocean itself. These are the questions we get asked most often, answered from first principles, with the limits stated.

Contents

• The basics of sound underwater 
  
  • What is underwater acoustics? 
  •  How fast does sound travel underwater?
  • Why does sound travel so much farther underwater than light or radio?
  • What does dB re 1 µPa mean? 
  • What is the SOFAR channel? 
  • What frequency ranges matter?
• What makes sound in the ocean
  
  • Main sources of ocean sound 
  • Ambient noise and the noise floor 
  • Measuring wind and rain acoustically 
  • Marine mammal vocalization frequencies 
  • What vessel noise looks like 
  • Reef, fish, and seagrass sound
• Hydrophones and how measurement works 
  
  • What is a hydrophone? 
  • Reading a sensitivity spec 
  • Sample rate needs 
  • Dynamic range, bit depth, gain
  •  Choosing a hydrophone
  • Deployment depth 
  • Hydrophone self-noise 
  • Calibration
• Passive, active, and everything called "acoustic" 
  
  • Passive vs. active acoustics 
  • What is PAM? 
  • Telemetry vs. PAM 
  • Acoustic modems and releases
• What makes a modern acoustic system
  
  • Recorders vs. reporting systems 
  • The value of real-time data 
  • What a duty cycle misses 
  • How long a system can listen 
  • Why process data on-device 
  • What "AI-enabled" means 
  • Updating models after deployment 
  • Networks vs. single recorders
• Operating an acoustic system in the real ocean 
  
  • Choosing a duty cycle 
  • What ends a deployment 
  • One hydrophone or an array? 
  • Locating a sound source 
  • Keeping a mooring quiet 
  • Where to place a hydrophone 
  • Servicing and data recovery 
  • Biofouling
• Reading and analyzing acoustic data 
  
  • Reading a spectrogram 
  • Sound pressure level 
  • Sound exposure level 
  • Decidecade bands 
  • How much data recording produces 
  • Automated detection accuracy 
  • Analysis software
• How underwater acoustics is used 
  
  • Marine mammal protection 
  • Vessel detection & maritime domain awareness 
  • Offshore construction & wind 
  • Ports and harbors 
  • Fisheries and aquaculture

The basics of sound underwater

What is underwater acoustics?

Underwater acoustics is the study of how sound behaves in water and the use of sound to sense and communicate beneath the surface. It exists as a field because seawater blocks light and radio within meters but carries sound for kilometers, which makes sound the only practical way to observe most of the ocean. It covers mapping the seabed, tracking animals, detecting vessels, measuring human noise, and moving data between submerged instruments. One distinction organizes all of it: passive acoustics means listening to sound something else made, active acoustics means transmitting a pulse and reading the echo. Most confusion in this subject comes from a single word covering both.

How fast does sound travel underwater, and why is it faster than in air?

Sound travels at about 1,500 meters per second in seawater, roughly four and a half times faster than in air, because water is far stiffer and resists compression more. The figure is not fixed. Speed rises about 4 m/s per degree Celsius, 1.3 m/s per unit of salinity, and 1.6 m/s per 100 meters of depth, ranging across the ocean from roughly 1,450 to 1,550 m/s. That variation matters more than the average: sound bends toward slower water, so gradients refract it into ducts that carry unexpectedly far and shadow zones where a loud source is inaudible. Detection range is never a simple function of distance.

Why does sound travel so much farther underwater than light or radio?

Sound travels much farther underwater than light or radio because seawater absorbs electromagnetic energy strongly but absorbs acoustic energy only weakly. Seawater conducts electricity, so radio at any useful data rate dies within meters, and light fades within tens of meters even in clear water. Sound loses energy far more slowly, and how slowly depends sharply on frequency. A 100 Hz signal loses very little over 100 km; a 100 kHz signal is gone within a few kilometers. That creates a trade nothing avoids: low frequencies travel far and describe little, high frequencies describe precisely and travel barely at all. Every acoustic instrument sits somewhere on that line, and choosing where is the first decision in any study.

What does dB re 1 µPa mean, and why can't you compare underwater decibels to decibels in air?

dB re 1 µPa is a sound pressure level measured against a reference of one micropascal, the underwater convention. Airborne acoustics references 20 micropascals instead, so the same physical sound produces two very different numbers. Decibels are always relative, which makes the reference part of the unit rather than a footnote. Converting needs two corrections: about 26 dB for the change of reference, and roughly 36 dB for the difference in acoustic impedance between air and water. A given intensity therefore reads about 62 dB higher underwater. This is the most common error in press coverage of ocean noise: a source at 180 dB re 1 µPa is not comparable to a 180 dB jet engine. A decibel figure without its reference is not a measurement.

What is the Sofar Ocean channel?

The Sofar Ocean channel, short for Sound Fixing and Ranging, is a horizontal layer where ocean sound speed reaches its minimum, typically near 1,000 meters at mid latitudes. Sound entering it is refracted back toward the axis rather than escaping, so it can travel thousands of kilometers with little loss. It exists because temperature falls with depth, slowing sound, while pressure rises, speeding it up; the crossover is the minimum. Identified in the 1940s and used to locate downed aircraft, it now supports basin-scale acoustic thermometry and explains how low-frequency whale calls cross oceans. It is not universal: the axis shoals at high latitudes and vanishes in shallow shelf seas, which is where most coastal monitoring actually happens.

What frequency ranges matter in underwater acoustics, and what lives in each one?

Ocean sound spans more than five orders of magnitude, and knowing which band your subject occupies determines every hardware choice that follows. Below 20 Hz sit blue and fin whale calls, earthquakes, and the accumulated rumble of distant shipping. From 20 Hz to 1 kHz are most large whale vocalizations, ship machinery and pile driving. Between 1 and 20 kHz are dolphin whistles, fish choruses, snapping shrimp, and wind and rain noise. From 20 to 200 kHz come echolocation clicks and high-frequency sonar. Frequency governs range as strictly as content, and no single instrument covers the span well.

What makes sound in the ocean

What are the main sources of sound in the ocean?

Ocean sound comes from three families of sources that are usually all present at once: biological, physical, and human. Biological includes whale and dolphin calls, fish choruses that swell at dusk, and snapping shrimp, which in warm shallow water can dominate everything. Physical includes wind and breaking waves, rain, ice, earthquakes and volcanic activity. Human sources are led overwhelmingly by commercial shipping, then seismic airgun survey, pile driving, dredging, sonar and small craft. The overlap is what matters. Shipping and baleen whale calls occupy nearly the same low-frequency band, which is why shipping noise is treated as a conservation problem rather than a nuisance: it masks the signals animals depend on rather than merely adding to them.

What is ambient ocean noise, and what sets the noise floor at a site?

Ambient noise is the background level at a site with no single identifiable source. It is what anything you want to detect has to rise above, and different processes dominate different bands. Below 10 Hz it is turbulence and wave interaction. From 10 to 500 Hz, distant shipping dominates across most of the northern hemisphere even far from any visible vessel. Between 500 Hz and 50 kHz, wind-driven surface agitation sets the level, rising with sea state. Above 100 kHz the thermal motion of water molecules imposes an absolute floor. The Wenz curves remain the standard reference. One consequence governs instrument choice: your equipment's own noise must sit below ambient, or you are measuring your electronics.

Can you measure wind and rain from underwater sound?

Yes, a hydrophone can measure wind and rain reasonably well, because both leave a distinctive acoustic signature. Rain produces a characteristic spectral peak between roughly 13 and 25 kHz, generated by tiny bubbles entrained as small drops hit the surface, and the spectrum shape distinguishes drizzle from downpour. Wind noise is broadband from about 500 Hz to 20 kHz and rises steadily with wind speed, which is the basis for acoustic wind retrieval. A hydrophone therefore doubles as a rough weather sensor, useful where no surface station exists. Treat it as a supplement rather than a replacement: the inference is empirical and usually needs local calibration, shipping and biological noise contaminate the same bands, and frozen precipitation behaves differently.

What frequencies do marine mammals vocalize at?

Marine mammals vocalize across more than four orders of magnitude, from around 10 Hz to above 130 kHz. Baleen whales occupy the low end: blue and fin calls between about 10 and 100 Hz, right whale upcalls around 50 to 200 Hz, humpback song reaching several kilohertz. Toothed whales work far higher and use two kinds of sound: social whistles typically 2 to 20 kHz, and echolocation clicks much higher still, with dolphin clicks peaking between 40 and 130 kHz and harbor porpoise producing narrowband clicks near 130 kHz. Seals mostly call below 10 kHz. The consequence is blunt: a system configured for blue whales will not hear a porpoise. Start from the species list.

What does vessel noise look like acoustically?

Vessel noise is broadband, weighted heavily toward low frequencies, and shaped by three mechanisms that together make vessels acoustically identifiable. Propeller cavitation dominates above a few knots, producing broadband noise that swells and fades over minutes as a vessel passes. Machinery produces narrowband tones at blade rate and its harmonics, and those lines are what make acoustic classification possible. Hull flow noise adds energy higher up. On a spectrogram it is unmistakable: a broadband smear crossed by steady horizontal lines. Most energy sits between roughly 10 and 500 Hz, though small craft push into the kilohertz range. What acoustics does not easily give is range, since source level varies enormously and a single hydrophone cannot resolve it.

Do reefs, fish, and seagrass make sound you can actually measure?

Yes,  reefs, fish, and seagrass all produce measurable sound, and in some habitats they set the entire soundscape rather than sitting in the background. Snapping shrimp produce an extraordinarily loud broadband click by collapsing a cavitation bubble, and a colony generates a continuous crackle peaking at a few kilohertz but extending well above 100 kHz. Many fish are soniferous, producing grunts and knocks mostly below 1 kHz, often in synchronized choruses at dawn and dusk. Seagrass releases oxygen bubbles during photosynthesis that are acoustically detectable, proposed as a way to monitor productivity without touching the meadow. This is the basis of soundscape ecology. The catch is bandwidth: much of that energy sits above a general-purpose hydrophone's range.

Hydrophones and how measurement works

What is a hydrophone and how does it work?

A hydrophone is an underwater microphone: a sensor that converts pressure fluctuations in water into an electrical voltage. Almost all are piezoelectric. A ceramic element generates a small voltage when mechanically strained, which a pre-amplifier inside the housing boosts before an analog-to-digital converter samples it. The assembly is encapsulated, usually in polyurethane, both to keep water out and to couple the element acoustically to the water. Most are omnidirectional, and that sets the fundamental limit. A hydrophone measures pressure at one point, so it tells you a sound occurred and how loud it was, but not its direction or range. It also does not measure particle motion, which is what many fish and invertebrates actually sense.

What is hydrophone sensitivity, and how do you read a sensitivity spec?

Sensitivity is how much voltage a hydrophone produces for a given sound pressure, written in dB re 1 V/µPa. The figures are always negative and less negative means more sensitive, which catches out almost everyone reading a datasheet for the first time. The negative sign is arithmetic rather than physical: a micropascal is tiny, the voltage per micropascal is far below one volt, so the ratio's logarithm is negative. A unit at −180 dB re 1 V/µPa is more sensitive than one at −200. On its own the number tells you little, because what sets the quietest resolvable sound is sensitivity, pre-amp gain and self-noise together. Sensitivity also varies with frequency, so ask for the response curve rather than the figure.

What sample rate does underwater acoustic recording need?

Underwater acoustic recording needs a sample rate of at least twice the highest frequency you want to capture, with margin. That is the Nyquist limit, a hard mathematical constraint rather than a guideline. Sampling faithfully represents sound up to half the sample rate; anything above folds back as false low-frequency content, an artifact called aliasing that cannot be removed afterward. So dolphin whistles reaching 20 kHz need at least 40 thousand samples per second, while harbor porpoise clicks near 130 kHz need above 260 ksps and realistically 300 to 500. Rate then drives storage, power and everything you can get ashore. Over-specifying is expensive rather than merely safe, and under-sampling is irreversible.

What are dynamic range, bit depth, and pre-amp gain?

Dynamic range, bit depth, and pre-amp gain are three settings that together decide what a recorder resolves. Bit depth is how many levels the converter uses per sample: 16-bit gives about 96 dB of range, 24-bit about 144 dB. Dynamic range is the span between the quietest sound distinguishable from system noise and the loudest before clipping. Pre-amp gain is the amplification applied before digitizing. The point people miss is that gain moves the window rather than widening it. Turning it up lowers both the quietest sound you can resolve and the level at which loud sounds distort, so a system set for distant whale calls will clip on nearby pile driving. If your deployment must cover both quiet and very loud events, say so before choosing a configuration.

How do you choose a hydrophone for the sounds you want to hear?

Start from the sound, not the instrument. Four things settle it, in order. The frequency range of your target sets bandwidth and sample rate, and a 20 Hz fin whale call and a 130 kHz porpoise click cannot be captured well by one configuration. The expected received level against your site's ambient floor sets required sensitivity and self-noise. Deployment depth sets the pressure rating. Whether you need full recordings or only detections sets storage and telemetry. Check the response curve against your band with margin at both ends. Because no hydrophone covers the whole useful range, a platform that accepts different sensors, including third-party units, lets you match the instrument to each study rather than the reverse.

How deep can a hydrophone be deployed?

Hydrophone deployment depth depends entirely on the pressure rating of the specific unit, which ranges from tens of meters for compact coastal instruments to full ocean depth for deep-sea packages. The limit is mechanical: pressure rises about one atmosphere every 10 meters, and the housing, connector and encapsulation all have to withstand it. Depth also changes what you hear, not just whether the instrument survives. Sound speed varies with depth so ray paths bend, and a sensor near the surface picks up wave and wind noise plus, on a moored system, the platform above it. On a moored system the cable and mooring hardware carry their own depth ratings, which may be shallower than the sensor's. Ask for the working window, not just the maximum.

What is hydrophone self-noise, and why does it set your measurement floor?

Self-noise is the level a hydrophone appears to record when there is no sound at all: the electrical noise of the sensor and pre-amplifier expressed as an equivalent acoustic level. It sets the quietest thing you can honestly measure, and what matters is the comparison against ambient at your site. If self-noise sits well below local ambient you are measuring the ocean; if ambient falls toward it, you are measuring your electronics. One distinction datasheets never make: instrument self-noise is not platform noise. Flow noise, cable strum and motion transmitted from a surface float are covered by no sensor specification and usually dominate in shallow moored deployments. Ask any supplier for both numbers.

How is a hydrophone calibrated, and how often?

Calibration establishes the relationship between the voltage a system produces and the actual sound pressure, which is what turns recorded numbers into physical measurements. Absolute calibration is done in a tank or open-water facility against a reference standard, producing a frequency-dependent sensitivity curve; manufacturers normally supply a certificate. Field calibration is a spot check with a portable calibrator confirming the system has not drifted between deployments. Without it you can detect events and compare levels within one dataset, but you cannot report an absolute sound pressure level, which is what regulators and peer review require. Elements drift slowly and encapsulation ages, so annual recalibration is common practice.

Passive, active, and everything called "acoustic"

What is the difference between passive and active acoustics?

Passive acoustics listens. Active acoustics transmits a sound and interprets the echo. A passive system detects sound something else produced, adds nothing to the water and cannot itself be detected, but observes only things that make noise. An active system such as an echosounder, multibeam sonar or current profiler emits a pulse and measures the return, so it can map the seabed or measure currents regardless of whether anything is vocalizing. Two consequences. To know which animals are present, listen; to map a habitat, transmit. And active systems add sound to the environment, which increasingly needs its own permitting and can disturb the very animals a co-located passive system was deployed to detect.

What is passive acoustic monitoring (PAM)?

Passive acoustic monitoring is listening underwater over extended periods to detect, identify and count sound-producing animals or human activity. One or more hydrophones record continuously or on a schedule, and the recordings are processed by analysts or automated detectors into a time series of detections, yielding presence, seasonal pattern, distribution and, with more work, density. Its advantage over visual survey is availability: it works at night, in heavy weather, and continuously for months, detecting animals that spend almost all their lives submerged. Its central limitation is that it detects animals only when they vocalize, so silence is not absence. Presence data is strong evidence; absence data is much weaker.

How is acoustic telemetry different from passive acoustic monitoring?

Acoustic telemetry tracks tagged animals. Passive acoustic monitoring listens to sounds animals make naturally. Both get called acoustic monitoring and the hardware looks similar, which causes persistent confusion in procurement. In telemetry a small transmitter is attached to an animal and emits a coded ping, commonly around 69 or 180 kHz; moored receivers log the tag identity and time whenever it passes within a few hundred meters, so you learn where known individuals went. In passive monitoring there is no tag: a broadband hydrophone records ambient sound and you detect whatever vocalized. A telemetry receiver is narrowband and stores detections rather than audio, so neither instrument substitutes for the other.

What are acoustic modems and acoustic releases used for?

Acoustic modems and acoustic releases both use sound to carry information or commands through water where radio cannot reach. An acoustic modem encodes data as a modulated acoustic signal, letting a subsea instrument talk to a vessel, a buoy or another instrument. Rates are low, typically hundreds to a few thousand bits per second over a few kilometers, because the underwater channel suffers severe multipath and Doppler. An acoustic release is simpler: equipment held on the seabed by an anchor lets go on receiving a coded command, floating up for recovery. Releases are standard on deep moorings. Modems are a last resort: slow, power hungry, unreliable in shallow reverberant water, and every transmission adds sound to an environment you may be monitoring.

What makes a modern acoustic system

What's the difference between a recorder you retrieve and a system that reports while it's deployed?

An autonomous recorder stores sound to internal memory and tells you nothing until someone collects it. A reporting system analyzes sound where it sits and sends results ashore over satellite or cellular while the deployment continues. Recorders have been standard for decades and are good at what they do: high sample rates, a complete archive, no power spent transmitting. The cost is time, since the data carries an embargo equal to the deployment plus the wait for a vessel, often months. The two are not substitutes, because no link that can carry a detection can carry the audio behind it. A reporting system is triage, and one that reports but keeps no archive has traded away every question you had not thought of yet. Most current systems do both: Spotter Sound sends detections and band levels ashore over satellite or cellular while the full-rate waveform stays on the instrument.

What can you do with acoustic data in real time that you can't do with data recovered six months later?

Live acoustic data lets you act on anything that requires acting while the sound is still happening, recovered data cannot. Recovered data can tell you what occurred; only live data can change what happens next. Three kinds of work depend on it. Intervention, such as shutting down pile driving when a marine mammal enters an exclusion zone, where a recorder retrieved in autumn can only prove a shutdown should have happened in spring. Alerting, such as vessel intrusion or whale presence warnings, where the entire value is latency. And knowing the instrument is alive, since a recorder that floods in week two costs the season and you find out at recovery. Retrospective science needs none of this and is often better served by a recorder.

If a system records two minutes an hour, what does it miss?

A two-minutes-per-hour duty cycle misses ninety-seven percent of the time. Two minutes in sixty is a 3.3% duty cycle, and nothing in the other 58 minutes exists in your dataset. Whether that matters depends on the target. For persistent sound such as shipping or ambient noise, a duty cycle is a sound sampling strategy, because you are estimating a distribution. For discrete events the arithmetic turns against you: a brief isolated call has roughly the duty cycle as its chance of being recorded at all. Because the bias depends on each species' vocal behavior it cannot be corrected generically, and duty cycling cannot capture the onset of an event, which is exactly what compliance monitoring needs.

How long can a passive acoustic system listen before someone has to go service it?

How long a passive acoustic system can listen unattended depends on whether power is the constraint, and on many modern systems it no longer is. Battery-powered recorders last weeks to about a year depending on pack size and duty cycle. Solar-recharged platforms take power off the table. The arithmetic is worth seeing. A hydrophone node drawing 150 mW while recording uses about 3.6 Wh a day. A surface buoy carrying roughly 10 W of solar panel and 50 Wh of battery holds about two weeks of listening with no sun at all, and the panels need to average under two percent of peak output to cover the draw indefinitely. What actually ends deployments is mechanical and biological: biofouling, chafe, storage filling, permits and vessel schedules. Power stops being the limiting factor, which is not the same as unlimited. On the Spotter hydrophone the node runs off the buoy's solar-recharged battery, so what ends a deployment is fouling and servicing rather than a battery budget.

Why process acoustic data on the device instead of sending it all ashore?

On-device processing exists because underwater audio is thousands of times larger than any link available to a buoy at sea. Continuous recording at 62.5 thousand samples per second, logged at 16 bits, produces about 450 MB an hour, close to 11 GB a day. An Iridium satellite link carries a few hundred bytes per message, so even at a message a minute it moves roughly 20 kB an hour: about twenty thousand times less. Transmitting bits also costs far more power than computing on them. A detection is a timestamp and a label; a set of band levels is a few kilobytes. Those travel, the waveform does not. The limitation is that edge processing returns only what you asked for, so keep the archive and make sure you can change the question. Spotter Sound works this way, running detection models on the node while events and band levels are what cross the link.

What does an "AI-enabled" hydrophone actually do?

An AI-enabled hydrophone runs a detection or classification model on a small processor inside the instrument, so instead of storing or sending sound it reports that something happened at a particular time. The pipeline is similar regardless of vendor: audio becomes a compact representation, a detector flags candidates, a classifier assigns a label, and a timestamped event leaves the instrument. How well it works deserves skepticism, and any single accuracy figure should be pressed. Performance is site-specific, and a model trained in one environment often degrades in another with different noise and species. Thresholds trade false positives against missed detections, so decide in advance which you can afford. And no model finds what it was never trained on.

Can you update detection models or change settings after a system is already in the water?

On some systems, yes, detection models and settings can be updated over the same link the data comes back on. On most traditional recorders, no: every setting is fixed when the instrument is sealed. Two things get updated with very different costs. Configuration changes such as duty cycle, sample rate and thresholds are a few bytes and travel over almost any link. Model updates are much larger and generally need cellular or better. This matters because it decouples the deployment from the decision. Without it, every parameter must be right before the instrument enters the water and any change costs a vessel trip. It is also the answer to the previous limitation, since it lets you change what you are asking for without recovery.

Why use a network of listening stations instead of one recorder?

One hydrophone tells you a sound happened. Several separated ones tell you where it came from, cover an area rather than a point, and keep the dataset intact when one fails. Localization is the clearest gain: a single omnidirectional hydrophone has no sense of direction, while three or more at known positions with synchronized clocks locate a source from arrival-time differences. Coverage is less intuitive, because detection range varies enormously with target. A low-frequency whale call may carry tens of kilometers while a high-frequency click fades within hundreds of meters, so station spacing follows the sound rather than the map. This differs from an array, which is closely spaced elements on one platform used for bearing.

Operating an acoustic system in the real ocean

How do you choose a duty cycle?

Work out how many recording hours your power and storage allow across the deployment, then decide how to spend them. For persistent sources such as ambient noise or shipping, spread the sample evenly and a low duty cycle serves well. For episodic targets the reasoning inverts. Concentrate hours on when the sound occurs if you know, and if you do not, many short periods across the day sample the diel cycle far better than one long block, though very short periods truncate long calls and bias anything calculated from duration. Check first whether you need to duty cycle at all: continuous data can always be subsampled afterward, while a duty-cycled recording can never be made continuous.

What actually ends an acoustic deployment?

What actually ends a deployment is rarely what the plan assumed. On battery-powered recorders it is usually power. On solar-recharged platforms it is physical, biological and logistical. Biofouling accumulates on the sensor and mooring line. Hardware chafes and corrodes, with shackles and cable jacketing the usual failure points. Storage fills if you archive at high sample rate. Vessel availability, permit windows and seasonal weather often decide the recovery date before any hardware limit is reached, and sometimes the answer is loss: entanglement, vessel strike, storm damage or theft. Plan against the shortest of these, and when evaluating a system ask what typically ends deployments rather than what the maximum endurance is.

Do you need one hydrophone or an array?

An array is several hydrophone elements on one platform, spaced from centimeters to meters apart. What it buys is direction, which one hydrophone cannot provide at any price. Because sound reaches each element at a slightly different time, comparing arrival times gives the bearing to a source, a technique called beamforming. Spacing sets where it works: elements need to sit roughly half a wavelength apart at the target frequency, so a high-frequency array is compact and a low-frequency one is long. An array gives bearing, not position, and locating a source needs a second array or a network of separated receivers. For a fixed station that only needs presence, one hydrophone is the simpler right answer.

How do you work out where a sound came from?

Working out where a sound came from takes multiple separated receivers and precise timing. A single omnidirectional hydrophone cannot do it, and no processing changes that. The standard method is time difference of arrival: the same sound reaches each receiver at a slightly different moment, each pair narrows the source to a hyperbolic surface, and three or more intersect to a point. Three receivers give a two-dimensional position inside the array footprint; four give depth. Accuracy depends on clock synchronization, since sound covers about 1.5 millimeters per microsecond, on knowing where the receivers actually are, and on the sound speed profile. It works far better for impulsive signals such as clicks than for continuous sound, and poorly for sources outside the footprint.

How do you keep a mooring quiet enough to measure on?

Isolate the sensor mechanically and reduce flow past it. Most of what looks like ocean noise in a moored recording is the mooring itself. Three paths contribute. Flow noise is turbulent pressure fluctuation as water passes the hydrophone, not sound at all but indistinguishable in the recording, reduced by shrouding the element and avoiding high-current placements. Strum is vortex-induced vibration of the mooring line, addressed with cable design, fairing and tension. Mechanical transmission carries wave-driven float motion down the line, reduced with compliant elements and depth. No mooring is silent, so the honest comparison is against the ambient level you need to measure rather than against zero. Ask for measured platform noise, not just electrical self-noise. The Spotter hydrophone addresses the latter two paths with a suspension chain and a cover around the sensor, plus a specified minimum standoff below the surface.

Where on a mooring should a hydrophone sit?

A hydrophone on a mooring should sit deep enough to escape surface noise, shallow enough to stay inside the pressure rating, and positioned for the depth your target occupies. Near the surface a hydrophone picks up breaking waves, wind, rain and, on a moored system, the float above it, so several meters of standoff is a common minimum. Going deeper reduces that but changes what you hear, since sound speed structure refracts ray paths and a sensor can sit in a shadow zone. Position also matters relative to other instruments: an active device such as a current profiler on the same line will be clearly audible, so separate them or offset their duty cycles.

How do you service a deployed hydrophone or recover its data?

Servicing a deployed hydrophone depends on the system, and the honest default is that you will be recovering the instrument. Three models exist. Full recovery brings the mooring up to swap or download storage on deck: simple and reliable, but it needs a vessel, a weather window and crew. In-water servicing, where a diver swaps a card, is less common than people hope, because any underwater connector operation risks flooding. Telemetered offload sends summary products ashore continuously while the archive stays aboard as a backup. What telemetry changes is urgency, not necessity. Ask specifically whether storage is field-swappable and whether connectors are rated for wet mating, and assume no unless stated. The Spotter hydrophone combines the last two models: summary products stream ashore continuously while swappable cards hold the full-rate archive.

How does biofouling affect acoustic measurements?

Biofouling degrades the measurement before it damages the hardware, and it is the most underestimated constraint on long deployments in productive coastal water. Growth on the encapsulation changes the acoustic coupling between water and sensing element, attenuating and distorting the response, worst at high frequencies where the layer becomes significant relative to wavelength. A fouled instrument can look like it is working while quietly losing its upper band. On the mooring, growth adds weight and drag, increasing motion noise and accelerating wear. Rate varies enormously with temperature, nutrients and season. Coatings, copper components and wipers help, none permanently. Treat fouling as a line item in the deployment plan rather than as something that goes wrong.

Reading and analyzing acoustic data

What is a spectrogram, and how do you read one?

A spectrogram is a picture of sound over time: time on one axis, frequency on the other, color showing how much energy is present at each frequency at each moment. It is made by cutting the recording into short overlapping windows and running a Fourier transform on each. Once you know the vocabulary the patterns are unmistakable. Horizontal lines are steady tones from machinery. Vertical lines are impulsive sounds such as clicks or a hammer strike. Sweeping curves are frequency-modulated calls such as whistles or whale song. A broad brightening that swells and fades over minutes is almost always a vessel passing. Window length trades time resolution against frequency resolution, so the settings are part of the result and two spectrograms of the same file can look quite different.

What is sound pressure level, and how is it calculated?

Sound pressure level is loudness expressed logarithmically against a reference pressure, written dB re 1 µPa underwater. It is the fundamental quantity in nearly all reporting and regulation. From a calibrated recording you square the pressure time series, average over a stated window, take the root, and express the result against one micropascal in decibels. Two details decide whether one figure can be compared with another, and both must be stated: the averaging window, since an RMS level over a second and over a minute differ for the same event, and the frequency band. Peak sound pressure level, the largest instantaneous pressure, is a separate quantity used by impulsive noise criteria. A figure without these is not interpretable.

What is sound exposure level, and when do you report it instead of SPL?

Sound exposure level measures the total acoustic energy an event delivers rather than how loud it was at any moment, reported in dB re 1 µPa²s. Where sound pressure level averages pressure over a window, sound exposure level integrates squared pressure across the whole event, so a loud brief sound and a quieter sustained one can carry identical values. That makes it the right measure where risk accumulates, which is why it sits at the center of marine mammal injury criteria. For pile driving or seismic survey, regulators typically want peak SPL, RMS SPL and cumulative SEL together, because each answers a different question. Cumulative SEL is partly modeled, since it assumes how long an animal stayed and whether it moved away.

What are decidecade bands, and why are they the reporting standard?

Decidecade bands, long known as third-octave bands, divide the frequency axis into slices that get proportionally wider as frequency rises. Each band's upper edge sits at a fixed ratio above its lower edge, so bands are narrow at low frequency and wide at high. This roughly matches how animal hearing resolves frequency and reduces thousands of spectral values to a few dozen numbers that can be tabulated and compared. Decidecade is now preferred in standards because it has one unambiguous definition where third-octave has two. ISO 18405 fixes the terminology for underwater acoustics, and using its terms exactly is what makes a measurement comparable between programs. Band levels discard detail, so they supplement spectrograms rather than replacing them.

How much data does continuous acoustic recording produce?

Enough to shape the entire system design: continuous recording at 62.5 thousand samples per second, logged at 16 bits, produces about 450 MB an hour and just under 11 GB a day per channel. The arithmetic is simple: sample rate times bytes per sample times channels. Halving the rate to 31.25 ksps gives about 225 MB an hour, or 5.4 GB a day, so a one-terabyte card runs roughly six months instead of three. This is why sample rate is an economic decision as much as a scientific one, and why detections and band levels are what travel over a telemetry link while audio stays local. Compression helps less than people hope, because ambient ocean noise barely compresses losslessly.

How accurate is automated species detection, and how would you know?

Automated species detection accuracy varies enormously by species, site and conditions, and any single accuracy figure quoted without those attached should be treated with suspicion. Performance needs two numbers that trade against each other through the detection threshold: precision, the fraction of reported detections that are real, and recall, the fraction of real events found. Loud, stereotyped signals do well, and mature detectors exist for harbor porpoise clicks and right whale upcalls. Species with variable calls or sites with unusual noise do worse, and detectors commonly degrade when moved between environments. Decide in advance which error costs you more, validate against human-verified data from your own site, and keep the recordings so results can be re-derived as detectors improve.

What software do people use to analyze underwater acoustic data?

People analyzing underwater acoustic data use a mixture of research tools and commercial packages, most of the widely used ones free. Raven, from the Cornell Lab of Ornithology, is the most common general-purpose viewer and annotation tool. PAMGuard is the open-source standard for detection and localization, including real-time mitigation work. Triton and MATLAB toolkits are common in academic labs, and Audacity handles basic inspection. For standards-compliant soundscape metrics, tools such as MANTA and PyPAM implement the ISO-defined quantities consistently. No single tool covers everything, so most programs run two or three. Before buying any instrument, confirm what file format it writes: a proprietary format with no open reader is a long-term risk to your archive.

How underwater acoustics is used

How is passive acoustic monitoring used to study and protect marine mammals?

Passive acoustic monitoring is now a primary method rather than a supplement for studying and protecting marine mammals, establishing which species are present, when they arrive and leave, how they use an area, and increasingly triggering action while they are there. Long recordings reveal patterns no visual survey could resolve, because listening works at night, through heavy weather, and continuously for months. In offshore construction it supports exclusion zone mitigation. In shipping lanes, real-time whale detection feeds strike-reduction warnings to mariners. In protected areas it establishes baseline occupancy and tracks change over years. The constraint carries through every application: acoustics detects animals only when they vocalize, so it is strong evidence of presence and weak evidence of absence. Spotter Sound is built for this work, with marine mammal detection models running on the instrument.

How is underwater acoustics used for vessel detection and maritime domain awareness?

Underwater acoustics is used for vessel detection and maritime domain awareness to detect and characterize vessels that other systems miss, particularly those that do not want to be seen. AIS transponders can be switched off and radar has horizon limits, but a vessel's propeller and machinery noise cannot be turned off while under way, so a hydrophone detects transits regardless of what is being broadcast. Because passive listening emits nothing, the sensor cannot itself be detected. Applications include marine protected area enforcement, illegal fishing monitoring, port security and persistent monitoring of chokepoints, with machinery tonals supporting classification as well as presence. Acoustics gives detection and timing, not identity, so its power comes from fusion with AIS and radar, where a detection with no matching AIS track is itself the finding. Unauthorized vessel detection is one of the cases Spotter Sound is designed around.

How is underwater noise monitored during offshore construction and wind development?

Underwater noise is monitored during offshore construction and wind development to demonstrate compliance and to make mitigation decisions while work is happening. Pile driving is among the loudest human activities in the ocean, and consent conditions increasingly require prediction beforehand and measurement during. Developers model expected levels and the ranges at which thresholds are exceeded; hydrophones then measure what actually occurs, and the comparison is typically what a regulator wants to see. Separately, monitoring the exclusion zone supports the decision to delay or halt piling when a marine mammal is detected. The mitigation half only works live: a recorder retrieved afterward can prove a shutdown should have happened but cannot cause one. That is the argument for a system that reports while the work is happening, such as Spotter Sound.

How do ports and harbors use acoustic monitoring?

Ports and harbors use acoustic monitoring to understand and manage the noise their own operations produce, to track marine life in approaches, and increasingly to meet noise reduction commitments. Several major ports run incentive programs rewarding quieter vessels, and those require measurement to administer. Acoustic monitoring establishes the baseline soundscape, quantifies what different vessel classes contribute, and verifies whether interventions such as speed reduction actually change levels. The same instruments support marine mammal detection in approach channels. Ports are acoustically among the hardest environments there are: shallow, reverberant and dominated by intense variable noise, so a design that works offshore usually will not transfer unchanged. A moored system reporting continuously, such as Spotter Sound, suits measurement that has to run through changing traffic rather than sample it.

How is underwater acoustics used in fisheries and aquaculture?

Underwater acoustics is used in fisheries and aquaculture in several distinct ways that are easy to conflate, because one word covers all of them. Passive listening detects soniferous species directly, and because many fish chorus while spawning, acoustic detection locates and times spawning aggregations without capture or handling. Acoustic telemetry tracks individual tagged fish through a study area. Active fisheries acoustics is a different discipline again, using calibrated echosounders to estimate biomass, and it underpins many stock assessments. In aquaculture, passive monitoring characterizes farm noise, detects predators approaching pens and in some operations monitors feeding. The confusion is practical rather than semantic: a telemetry receiver cannot hear a chorus and a broadband hydrophone cannot decode a tag. Whichever you need, check the band and sample rates against your target first; the Spotter hydrophone specifications list both.

‍

‍