What is a hydrophone?

The acoustic recorder that we put into the water works like any other microphone. It converts sound energy into an electrical signal, which is converted into digital samples. 

The particular device we use is designed in our research group at UC San Diego and has specialized capabilities. For example, it works underwater at great depths, will continuously record for a full year at a time, and is also designed to record at frequencies far above what the human ear can detect.  

What we record as “sound” is actually measurements of pressure—sound waves traveling through the water which push and pull on the hydrophone’s sensor. Those vibrations create the electrical signal. Then, the hydrophone’s data logger continuously converts these signals into digital information and stores it on hard drives as sound files.  

Every second, the data logger records 200,000 digital samples—in a year, that’s roughly 6.3 trillion recorded samples

What’s the purpose of using these instruments?

The science of passive acoustic ocean observation—which is what we call this kind of research—allows us to observe when marine life use the environment, what species were there, and what the environment was like at the time.  

We bring these acoustic observations together with environmental data and measurements of underwater noise to learn about animals’ relationships with their environment and how they are responding to underwater noise. 

What’s the next step in the research process?

After deploying the hydrophones, our next job is to go back and recover them from the seafloor, because all that data is stored on hard drives inside these devices. Once we recover those, we process the samples for analysis and long-term archival storage, then use a variety of computer processes to identify sounds of different species and make measurements of underwater sound levels. 

Marine animals live in a world that is illuminated by sound. Using those 6.3 trillion sound samples, we start by observing and measuring the natural sound environment that they live in: the wind blowing over the ocean, for example, and the advance and retreat of sea ice. Next, we measure the sounds that are added by human activities.  

Once we have measurements of the natural sound environment and the noise that’s added, we detect the presence and acoustic behaviour of different marine mammals using their sounds. For example, narwhals and belugas make distinctive echolocation clicks when they are diving and foraging, which can be detected using specialized computer processes. 

The final component of our research process is studying how animals are responding to ships and underwater noise, or what’s called a behavioural response study. For this, we leverage our team’s knowledge of the area’s wildlife and combine our many types of observations and measurements together into statistical models that help us understand how animals are responding to the presence of ships. 

How do you measure the sounds added by humans?

When we want to understand how animals respond to noise added to the natural environment in a particular place, there are two things to consider. First, we have to learn what we can about the noise source itself. In the case of ship noise, we need to get relatively close to each ship to measure its sound signature and how much noise it’s emitting.   

To do that, we place our hydrophones directly under the shipping lanes—we want the ships to go directly overhead—and take the measurements when they’re as near to the device as they can be. 

Sound from ships can travel very far. So, once we have this information, we can analyze the ship’s noise levels at various distances from the hydrophone—even up to 100 kilometres away. We combine this with the overhead recordings to create a model of the actual sound propagation from the ship for every recording location.  

Ultimately, we use these measurement-based acoustic models to make regional noise maps that can run on historical or future traffic scenarios. To do that, we need to know a lot about the ocean environment, which affects how sound travels underwater.  We look at the temperature and salinity of the water column, the shape of the seafloor, and the density and thickness of the sediment layers below the seafloor. Wind over the ocean and the presence of sea ice at the surface also have a strong effect on underwater sound.  

With all of these data, we can produce maps estimating and even predicting the sound footprints of ships as they go through the different regions—areas where local and regional partners are concerned about impacts of underwater noise on marine wildlife.

Is there a way to visualize the sounds that the hydrophones record?

One way to visualize sound is with a spectrogram. A spectrogram turns sound recorded over time into a picture that separates the sound into different frequencies.  

In a spectrogram, time is displayed from left to right, while frequency, or pitch, is displayed from top to bottom. Colour is used to show sound intensity, with brighter or “hotter” colours representing stronger sounds and cooler colours representing quieter ones. 

Once the data is visualized, how do you interpret the spectrogram?

The visual processing of sound is a fundamental part of how we observe patterns in the ocean environment—wind, sea ice, ships, and different marine species all produce distinct sound patterns that can often be recognized in the spectrogram. 

By combining advanced computer analyses with listening to and viewing the recordings as spectrograms, trained analysts can detect and describe the sound signatures of different species and ships. 

For instance, in the spectrogram above, we can easily identify echolocation clicks from a group of narwhals. From the rhythmic patterns of these clicks, we may infer that they are diving and foraging. A ship also appears in the spectrogram—an icebreaker that regularly travels through the region—with a very distinct acoustic signature. 

So much happens in a day in the ocean. These methods, based on recording and analyzing underwater sound, allows us to observe patterns across many time scales within a vast listening area; we can observe an hour, a day, a year, or even a decade of the ocean environment at each location. Combining recordings from multiple locations allows us to observe patterns within larger areas of the ocean.

We can listen to our recordings, and we can also see them visually in the spectrogram. But dealing with the scale of the data we have requires a combination of seeing and listening, paired with the power of computing and detection algorithms. Detection algorithms and processes like machine learning allow us to work through more than 400,000 75-second sound files recorded at each location every year. 

What other information is needed to draw scientific conclusions about the impact of ship noise on marine life?

The first step is finding out what we already know, looking at previous observations and publications. Importantly, Inuit already know Arctic marine wildlife in great detail. Working with our partners, we can start by asking a question like, “what do we know about normal narwhal behaviour?” Generally, when we do this work with marine mammals, we don’t always know what “normal” is, or what things make up the environment that the animals typically responding to, so we want to start with that existing knowledge to guide any new research.  

In our most recent study on narwhal responses to the proximity of ships in the Pond Inlet region, Inuit knowledge and guidance at every stage of the research process was crucial. Inuit have long pointed to the fact that narwhal have exquisite hearing. They can respond to high pitch sounds and underwater noise at very large distances—larger than what previous academic studies suggest. The initial assumption was that narwhal were similar to dolphins and couldn’t hear well at low frequencies.  

This means that, in the past, other studies of narwhal responses to ships in the Pond Inlet region didn’t investigate responses at great distances from ships because they began with the assumption that narwhal can’t hear them until they are very close.  

By contrast, we started with the understanding that narwhal hear ships very well, and from long distances. We assumed that any noise that is even perceptible may cause a behavioural response, which meant observing narwhal responses at various times, as individual ships passed within 40 kilometres of the hydrophone and for periods of up to 8 hours. 

By designing analyses informed by Inuit knowledge, we identified a change in narwhal behaviour in response to the proximity of large ships. The animals were either going quiet or fleeing when ships were within 20 kilometres from the hydrophone. Previous studies in the region hadn’t looked for behavioural impacts that far away.

What is the best part of this research?

Learning to see an ecosystem and to observe animals through listening is quite fun.  

It’s a process, and like all sciences, there’s a lot of continuity with the science that has come before us. For some species, we have the benefit of using previously published observations of sounds, as well as visual observations of species and their behaviour. There is a record to reference and draw from. There are also descriptions in scientific literature of some of the characteristics of sounds—like the detailed acoustic characteristics of different seal or whale species—that we already have access to. 

In the Arctic, though, we find ourselves in the position where our team has to do a lot of new interpretation ourselves, since many of the specific sounds a species makes have not been described in detail. In the areas where we work, we have the benefit of collaborating with people who pretty much know everything there is to know about their environment and animals. In many other parts of the ocean, you don’t get to work with people who know the different sounds, what the animals are, and when and where they can be found.  

That’s one of the unique aspects of this research in the Arctic: it’s made possible by the profound expertise of Inuit who know what the wildlife here does, the sounds they make, what’s normal behaviour, and what’s not. That’s a huge benefit to advancing the science more quickly, asking questions that are most relevant, and targeting the work to meet current gaps for decision-making. 

What is the biggest challenge when doing this kind of research?

The biggest challenge of observing the ocean in the Canadian Arctic is that the water is inaccessible for seven to ten months out of the year, depending on where you are, because of the sea ice cover. We collect our recordings from the sea floor, with the hydrophones sitting on the bottom, because that’s the best place to listen to and measure sound from. We can’t just go back every month to retrieve our recordings, and we can’t transmit the sounds from a buoy at the surface.  

We often work with community partners in places that they have identified as their highest priority, such as important harvesting regions where there are concerns about increasing ships. In some cases, such as with our Foxe Basin study area (a shipping lane leading up to the loading port at Steensby Inlet), the research location itself is very remote and further offshore. Sometimes, the only safe way to access them is using a larger ship and a multi-day cruise to get to that central location. 

Other times, these spots are close enough to the communities and the hunters and trappers’ organizations we work with, making them easier to access.  

Regardless of where they are, guidance from community members helps determine the sites where we study and the questions that need to be answered. We’re lucky, because we’re often able to go to these sites with the people who are already going out there in their boats. Together, we can find and access the locations, deploy and recover the hydrophones together, and do the work. That has been the basis of our success across our study areas.  

Elisabeth Hadjis is Communications Specialist at Oceans North. 

Abigael Kim is Engagement Specialist at Oceans North.