How can I improve dynamic range for towed hydrophones?

Question

Dynamic range is the ratio of the strongest, or loudest part to the weakest, or softest and is measured in dB (as explained in Ch 11 of Applied Underwater Acoustics). This site offers a musician's description of Dynamic Range. The ability to detect quieter sounds will depend on the dynamic range of the recording system.

Towed Hydrophone arrays in the marine environment are notoriously noisy due to the cavitation noise from the towing ship and this negatively impacts the ability to detect sounds with a lower signal to noise ratio. Towed arrays are frequently used during shipboard surveys to detect cetacean species (dolphins & whales). Typical methods include deploying a 2 element array 300 m behind a survey vessel moving at 10knots.

How can filtering improve dynamic range? Are there other tips & tricks to improve dynamic range?

Answer 1

when using towed arrays of hydrophones then I see various types of interfering noise

• flow noise
• tow cable strumming noise
• surface noise
• ship noise

flow noise can be removed by making the array hose larger.

strumming noise is reduced by fairings (caveat: depending on type they can introduce high frequency noise)

surface noise can be reduced by towing the array as deep as possible: with 300 m tow cable and 10 kn, the array will probably around 10 m depth, which is for my taste too shallow. I always insisted to have the array at least at 100 m, which happen to be in the Mediterranean also the lower depth of the thermocline. To get the array deeper, one can use longer tow cable, heavier tow cable, or depressor (V-fin), which however increases tendency of strumming. A lower tow speed (5 kn ) seems excluded by operational protocol.

finally, the OP asks for ideas of ship noise suppression. This is possible, but requires additional information. This old publication describes different methods, and I can say that I used the reference hydrophone method with success. For this an additional hydrophone was towed closer to the ship (in my case it was separated from the towed array) and the algorithm removed the correlated energy from the array measurements.

Caveat: I had a dense hydrophone array available and did time-delay and sum beamforming, so translating this method to the standard 2 element bioacoustic arrays would need some investigation, but it should give an idea what had been done.

Note: increasing the bit depth of the recorder does not help as ship noise masks the environmental signals and additional bits do not remove the masking

Answer 2

I would argue that the issue discussed here is more closely related to the signal-to-noise ratio (SNR) of the acoustic environment of a towed array than to "dynamic range" of the measurement system per se.

In general, "dynamic range" is more often used to discuss the limits of your data acquisition system - i.e. what is the range between the loudest and the quietest sounds your system can acquire/record without distorting. At the quietest end, your DAQ will only "record" its own noise (i.e. electronic noise generated by the measurement gear itself), which is why this limit is often referred to as the noise floor of the system. This noise floor is related to the electronic design of the device, and the bit depth (i.e. how many bits it uses to store each sample of the signal) is one of the factors involved in it. Increasing the bit depth may reduce the noise floor and improve your dynamic range, but not always.

In practice, it is much more common to have the ambient noise levels much higher than your DAQ noise floor, and this is almost sure the case with towed arrays. Some people would call the increased background noise level as the "noise floor" of the recording, but I would suggest the term "signal-to-noise ratio" is more appropriate in this context and avoids confusion.

As suggested by WMXZ, this issue is better treated at the source - i.e. reduce it before it is recorded. If your recording already contains both signal and noise, traditional filtering (high-pass, band-pass, FIR/IIR, etc) will only attenuate the magnitude of the signal+noise mixture, but won't separate them. Sometimes it's possible to improve your overall SNR by filtering out frequencies where there's mostly noise and very little signal (low SNR) and leave/amplify frequencies where there's little noise and more signal (high SNR).

Actually separating the signal from the noise after acquisition is a very complex task and a very active field of research. If you feel like going down a rabbit hole, some common terms used in this area are "blind source separation" and "independent component analysis", for example. Good luck!

Answer 3

The simplest way to increase dynamic range is to select a system with a large bit depth.

Check out this explaination by Mark Harris, What Is Bit Depth? published on Lifewire.com on October 14, 2021.