Towed arrays, the long listening cables that ships pull behind them, have so far worked out where a sound comes from by assuming the cable lies in a straight line. A team at Harbin Engineering University built a 770 m fiber-optic towed array that uses its sound-sensing fiber to measure how the cable's depth varies along its length, and published the work in the journal The Innovation in September 2026. In open-sea trials, the array tracked the bearing of a sound source 5.7 km away and picked up waves with a significant wave height (a standard measure of typical wave height) of 1.41 to 2.66 m.
What did the open-sea trials measure, and how well?
In trials at sea in December 2025 and January 2026, the team read three things from a single cable, namely its own depth, the surface waves and underwater sound, and checked each against a value obtained another way.
| What was checked | Compared against | Result |
|---|---|---|
| Cable depth | Pressure sensor | Coefficient of determination 0.99, average deviation 0.80% |
| Wave spectral peaks | Predictions that take towing speed and heading into account | 0.102 Hz and 0.098 Hz (2.86% and 8.41% off) |
| Significant wave height | Copernicus wave model | 1.41–2.66 m (within 4.31–11.03%) |
| Source bearing | GPS reference bearing | Tracked with two signals: vessel noise (120–150 Hz) and 1 kHz pulses from the source 5.7 km away |
| Effect of adding the depth profile | Calculation without it | Beamforming output peak up by as much as 4.10 dB |
Depth measured by the fiber agreed closely with the pressure sensor. Reconstructed from the depth profile, the cable was not a flat line but a bow. In this trial at least, the straight-line premise behind the calculation did not hold in the vertical direction.
In the wave band, the spectral peaks sat at 0.102 Hz and 0.098 Hz, close to the predicted 0.105 Hz and 0.107 Hz. Because those predictions take towing speed and heading into account, the comparison appears to concern the apparent frequency seen from a moving ship.
Significant wave height, derived from these peaks after correcting for element depth, came out at 1.41 to 2.66 m. It was compared with the Copernicus wave model, not with a measuring buoy. The authors estimate a detection floor of 0.15 m and a dynamic range of 180 dB for a tow depth of 50 m, although, because we could not read the paper itself, we could not confirm what quantity that floor refers to.
Why does a towed array need to know its own shape?
A towed array works out the bearing of a sound from tiny differences in arrival time at its hydrophones (underwater microphones), so it needs to know exactly where each hydrophone is. This calculation is called beamforming.
Currents and drag bend a cable in the water, yet conventional towed arrays carry only a few depth sensors and fill in the shape between them by guesswork. Treating a bent cable as straight blurs the focus of the calculation and shifts the bearing. It is a little like trying to focus a camera without knowing how its lens is curved.
What is built into the fiber-optic towed array?
The 770 m cable was towed about 500 m behind the ship during the trials. The release describes two sections: an interferometric section built for sensitive listening and a distributed acoustic sensing (DAS) section that can also measure depth. It does not describe the rest of the cable.
| Section | Length | Count | Main role |
|---|---|---|---|
| Interferometric | 104 m | 128 hydrophones (22 m of fiber each) | Sensitive listening |
| DAS | 160 m | 32 elements (5 m apart) | Depth, waves, sound |
DAS reads changes in the phase of light along an optical fiber. Because water pressure shifts that phase, the DAS elements can act as depth gauges while they pick up sound. Depth was measured by this 160 m DAS section, while the interferometric section handles listening.
The sensing parts of both sections share one design, with fiber wound around an air-backed core 42 mm long and 11.6 mm in outer diameter, and they differ in the length of fiber wound and in how the light is read out. The data are read in the following three frequency bands.
| Frequency band | What it reads |
|---|---|
| Below 0.02 Hz | Depth and shape of the cable |
| 0.02–0.3 Hz | Surface waves |
| 5 Hz–2 kHz | Underwater sound |
Could it track a sound source 5.7 km away?
A cooperating source set up for the trial emitted 1 kHz pulses 5.7 km away, and the bearing obtained by beamforming followed the GPS reference bearing. The team also confirmed that the bearing of vessel noise in a 120–150 Hz band followed the reference in the same way.
Feeding the depth profile measured by the fiber into the array model raised the beamforming output peak by as much as 4.10 dB. The release's subtitle also highlights that the measured array shape was fed back into beamforming.
On noise, the DAS section ran 11.57 dB above its own baseline measured in quiet conditions on land, across the 5–2,000 Hz band. The interferometric hydrophones stayed quieter above about 2 kHz.
How is this different from DAS on Japan's seafloor cables?
DAS itself is already used for ocean observation in Japan. The Japan Agency for Marine-Earth Science and Technology (JAMSTEC), together with NTT Access Network Service Systems Laboratories, connected DAS equipment to a seafloor telecommunication cable between Akusekijima and Takarajima in the Tokara Islands and recorded from 16 September 2025 to 13 January 2026. In results announced on 17 April 2026, JAMSTEC located more than 15,000 earthquakes with this setup, of which 434 had been located by the Japan Meteorological Agency from island stations.
| DAS in the Tokara Islands (JAMSTEC) | Harbin Engineering University's fiber-optic towed array | |
|---|---|---|
| Cable | Existing seafloor telecommunication cable between islands | Dedicated 770 m cable towed by a ship |
| Cable state | Laid on the seafloor and not moving | Moving through the water and bent by currents |
| Main signals | Earthquakes beneath the seafloor | Underwater sound, surface waves, and the cable's own depth |
| Timing | September 2025 to January 2026 | Open-sea trials in December 2025 and January 2026 |
A cable lying on the seafloor does not move, so its depth does not change over time. On a towed cable, depth keeps changing, and measuring it with the same fiber that picks up the sound is what sets this array apart.
What has not been tested yet?
The open-sea trials used one vessel, one towing speed and one sea state. Coastal and shallow waters have not been tested. Because depth and wave estimates rely on water pressure, the release itself notes that accuracy will shift with tow depth and in heavy weather.
We were able to read the EurekAlert! release and the paper's bibliographic record, but not the figures and tables of the paper itself. The release also does not say at what depth the cable was towed during the trials. What was measured is the shape in the vertical direction, and we found no statement that sideways bending was measured.
As future uses, the release lists wave monitoring from vessels of opportunity, meaning ships such as merchant vessels that are at sea for other purposes, along with fisheries and marine-mammal acoustics and sonar that cannot rely on a straight cable. These trials used one survey vessel, and installation on merchant ships and fisheries or marine-mammal work have not yet been tried.
