What if the ocean heard Malaysia Airlines Flight 370 crash—and everyone missed it? An exclusive investigation reveals a long-hidden scientific report that could reshape the hunt for the airliner, 12 years later.
This article orginally ran in Popular Mechanics on July 16, 2026.
JEAN-YVES ROYER COULDN’T STOP thinking about what his underwater microphones might have heard.
It was March 2014, days after Malaysia Airlines Flight 370 took off from Kuala Lumpur for Beijing and mysteriously vanished into the night with 239 people on board. Investigators would soon come to believe the Boeing 777 had flown south and crashed somewhere in the southern Indian Ocean, and searchers were urgently looking for the plane’s black boxes. But Royer, a scientist at the French government research organization CNRS, knew that something else was already waiting in those waters, potentially recording the sound of the airliner’s final moments: his network of autonomous hydrophones.
The network was called OHASISBIO, a French acronym meaning the Hydroacoustic Observatory of Seismicity and Biodiversity. Just weeks before MH370 disappeared on March 8, Royer’s team had moored hydrophones at six sites in the southern Indian Ocean, from the waters near Crozet and Kerguelen to the Madagascar Basin and the Southeast Indian Ridge. They were built to continuously record the ocean’s low-frequency sounds, including submarine earthquakes, volcanic eruptions on the seafloor, the calls of baleen whales, icequakes, ice tremors, and the background noise of a vast ocean in motion.
Now, Royer realized, they might also have captured something they were never designed to detect: the sound of a missing commercial jet hitting the sea.
The catch? He couldn’t listen yet.
Royer’s hydrophones didn’t stream their recordings back to shore. They were autonomous instruments moored far from land, storing recordings internally as they listened through the months after MH370 disappeared. Before Royer could search for the acoustic trace of a possible crash, he had to wait for the buoys to finish their mission, release from their moorings, and return to the surface in January and February 2015. Only then could his team recover them, download the recordings, and search the audio for the hours after MH370’s final satellite transmission for a sound loud enough to have come from a Boeing 777 hitting the Indian Ocean.
When the buoys finally came back, Royer had his sounds. By late 2016, he had turned the recordings into a formal analysis and says he sent it to investigators in France and Australia, two countries by then drawn into the widening search for MH370. But then the trail went quiet. The document never became part of the public search record. For years, whatever Royer’s hydrophones had captured in the southern Indian Ocean stayed locked inside a report that few people outside the investigation have seen.
Until now. When Popular Mechanics asked Royer about his 2016 analysis, he returned to a file that had been sitting out of public view for nearly a decade and sent it over: 18 pages of maps, spectrograms, arrival times, and careful conclusions drawn from the buoys after they came home from the Indian Ocean.
Royer’s report has surfaced at a critical moment. On June 29, Malaysia gave the search for MH370 another year, extending its no-find, no-fee agreement with Ocean Infinity, the U.S.- and U.K.-based marine robotics company scanning the southern Indian Ocean for the plane’s wreckage. The renewed contract gives Ocean Infinity until June 30, 2027, to search the remaining 2,868 square miles of seabed it has yet to inspect, with vessels expected to return between November 2026 and April 2027, when the seas are calmer.
For now, the search is still following MH370’s satellite data because the aircraft’s automated contacts with an Inmarsat ground station—a kind of routine electronic check-in between the plane’s satellite system and the network above it—remain the clearest trail investigators have from the aircraft itself. These contacts sent searchers deep into the southern Indian Ocean, but never gave them a crash site. Royer’s data is different. Instead of another reading of the satellite trail, it’s a separate record from instruments that were already in the water on the night the plane disappeared.
If Ocean Infinity’s search comes back empty yet again, the old question inside Royer’s data becomes much harder to set aside: What did the ocean record that night?
ROYER WASN’T THE ONLY ONE who thought hydrophone data might help solve the mystery. The Australian Transport Safety Bureau (ATSB), which assumed responsibility for conducting the underwater search for MH370 in late April 2014 and remained central to the search effort until it was suspended on January 17, 2017, had long understood that sound could fill in some of the gaps left by the case’s thin evidence.
When MH370 vanished in the early morning hours of March 8, 2014, it first went dark on civilian radar. Its transponder stopped identifying the aircraft to air traffic controllers, and the Boeing 777 disappeared from the radar screens tracking its planned route to Beijing. But for hours afterward, the aircraft continued making automated contacts with an Inmarsat satellite.
Although those contacts didn’t give investigators a GPS position or crash site, they did provide timing and frequency measurements—known as BTO and BFO—that left a faint trail to follow. Analysts inferred that the plane had flown south into the Indian Ocean and that its flight had ended near a curved line of possible positions that became known as the “seventh arc.” When investigators scanned that area of the seabed, however, they found no wreckage.
Investigators hoped sound data could resolve the mystery. At first, they focused on hydrophones in the International Monitoring System of the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO), whose global network listens for signs of nuclear explosions. The hydroacoustic system includes six underwater hydrophone stations and five land-based seismic detectors. Its hydrophones sit near the SOFAR channel—short for “sound fixing and ranging”—a layer where sound can travel efficiently over great distances, making the network capable of monitoring the world’s oceans.
There is precedent for using hydroacoustic data to solve a seabed mystery. In 2017, the Argentine submarine ARA San Juan vanished in the South Atlantic. CTBTO later reported that two hydroacoustic stations—Ascension Island and Crozet—detected an unusual signal from an underwater impulsive event near the submarine’s last known position. In November 2018, after two months of seabed searching, Ocean Infinity found the wreckage about 600 kilometers east of Comodoro Rivadavia, Argentina, in roughly 920 meters of water.
A sinking submarine isn’t a crashing airplane, but a large jet hitting the ocean hard should leave an acoustic trace. In a 2024 paper in Scientific Reports, Cardiff University researcher Usama Kadri found that past airplane crashes at sea produced acoustic signals that traveled 2,000 to 5,000 kilometers through the water. Given that MH370’s presumed crash zone was roughly 1,000 miles from the nearest hydrophone, “one expects a significant signal to appear,” Kadri wrote in the study.
An ATSB analysis of MH370’s final satellite transmissions suggested that the plane had made a steep descent, while Malaysia’s final investigation found that the pieces recovered from the cabin interior indicated the aircraft might have broken apart either in the air or on impact. If MH370 struck the Indian Ocean with that kind of force, hydrophones should have had a chance to hear it.
But when researchers at Curtin University’s Centre for Marine Science and Technology in Perth analyzed CTBTO data, they found no convincing acoustic trace of MH370 near the seventh arc. The most intriguing candidate was a low-frequency rumble recorded west of Rottnest Island just after 1:30 UTC, with a faint match later found in data from CTBTO’s Cape Leeuwin hydroacoustic station.
Because sound takes time to travel through the ocean, the signal reached the hydrophones after the event itself. Back-projecting the signal placed its origin earlier in time and far to the northwest. “When you measure it in multiple places, you can identify the position,” David Dall’Osto, a senior research scientist at the University of Washington Applied Physics Laboratory who has worked on the CTBTO hydroacoustic data, tells Pop Mech. “It’s in the northwest Indian Ocean.”
The source appeared to lie near a geologically active region called the Chagos-Laccadive Ridge, far from the seventh arc. That made it difficult to reconcile with the official search area for MH370. The simplest explanation was geological: an underwater earthquake or another natural event.
The absence of a CTBTO detection near the seventh arc doesn’t, by itself, prove that MH370 somehow wound up outside the search area. Whether the hydrophones would have detected an impact depends on how the plane hit the water. A high-speed impact should have produced sound loud enough to travel through the ocean. But a controlled ditching—a pilot trying to set the aircraft down on the surface rather than plunge into it—would be different. “A ditching like the Miracle on the Hudson is a much quieter event,” says Robert Parker, a former U.S. Navy intelligence officer. “You’re definitely not going to get any sound going into the deep sound channel from that.”
IF THE CTBTO DATA couldn’t settle the question, more hydrophones might at least narrow the possibilities. And that’s what Royer’s data offers.
Royer’s network, OHASISBIO, had been listening to the Indian Ocean since 2010. Its original purpose had nothing to do with airplanes. “We were interested in detecting small earthquakes that tell us about the dynamics of seafloor spreading,” Royer says. “And since we are recording low-frequency sounds, we also record calls from the large whales in different places and then figure out their migration pattern.”
Unlike the global CTBTO hydroacoustic network, OHASISBIO was built for the southern Indian Ocean. In 2014, Royer’s network included eight autonomous hydrophones at six sites, giving his team a more local record of the region’s low-frequency soundscape than the two relevant permanent CTBTO hydroacoustic stations—Cape Leeuwin and Diego Garcia—that Royer later used for comparison. “OHASISBIO data provide useful complementary information, as they improve the geographic coverage,” Royer says.
But OHASISBIO had a major drawback. CTBTO hydrophone stations transmit data continuously by cable and satellite for near-real-time processing at the International Data Centre in Vienna. Royer’s instruments worked differently: They recorded continuously, stored the data internally, and had to be recovered before the recordings could be downloaded and analyzed.
That delay shaped everything. MH370 disappeared on March 8, 2014, only weeks after the OHASISBIO hydrophones had been deployed. By then, there was nothing Royer could do but wait. The instruments weren’t emergency sensors, and they couldn’t be queried from shore. They were built for long, quiet missions, and they wouldn’t be recovered until January and February 2015.
When the research vessel Marion Dufresne finally returned to collect the buoys, bad luck intervened. The Northeast Amsterdam hydrophone—the OHASISBIO site closest to MH370’s presumed crash area—was lost during recovery. It likely would have had the best chance of detecting an impact near the seventh arc.
But the rest of the network survived. From those recovered instruments, Royer built a broader picture of the Indian Ocean’s sound environment than the CTBTO data alone could provide. In all, he tentatively identified five acoustic events in the early hours of March 8, around the time of MH370’s disappearance, two of which had already had been reported in CTBTO-based work.
Royer says this was the analysis he sent to French and Australian officials in late 2016. The next year, the ATSB published its final search report, The Operational Search for MH370. The report included appendices on hydroacoustic evidence, but Royer says his analysis was not among them. Royer believes Australian officials may have viewed his work as too close to earlier CTBTO-based analysis to publish separately. “It didn’t bring any new information relative to the CTBTO data,” he says.
But that may undersell what Royer’s report did. Although his hydrophones didn’t detect an impact near the seventh arc, they independently recorded some of the same events seen in earlier CTBTO-based work, including an ice-related event far to the south. In science, such duplication isn’t always redundant. It can show that two separate networks were hearing the same ocean, and had enough sensitivity and timing accuracy to serve as checks on each other.
The absence of an acoustic trace near the seventh arc doesn’t overturn the satellite analysis. But it belongs with a larger discomfort in the case: The evidence that should have converged, including end-of-flight modeling and debris analysis, has never quite lined up cleanly.
THE PLANE’S DEBRIS HELPED, but it didn’t close the loop. A right flaperon that washed ashore on Réunion Island, east of Madagascar, in 2015 confirmed that pieces of MH370 had reached the western Indian Ocean, and later fragments found along African beaches and nearby islands gave investigators more material to work with. But beach debris is a backward-looking clue.
Drift models could show how wreckage might travel across the ocean, yet they struggled to account cleanly for every fragment, landing site, and timeline. Even the evidence offered by marine life attached to the wreckage was puzzlingly incomplete: While barnacles on the flaperon appeared to preserve chemical traces of the waters they had crossed, offering a possible route back through the Indian Ocean, the specimens studied so far were too small to let researchers reconstruct the full journey, which marine biologists find hard to explain.
The most direct test of all—years of high-resolution sonar and seabed mapping across the priority search areas—found nothing. Between 2014 and 2017, the Australian-led underwater search covered more than 120,000 square kilometers of seafloor in the remote southern Indian Ocean and didn’t find the aircraft. Ocean Infinity’s 2018 search also returned nothing, bringing the total area searched close to 200,000 square kilometers.
The company later went back with more advanced robotics and, by March 2026, said it had spent more than 150 days at sea since 2018 and mapped more than 140,000 square kilometers of seafloor. Still, no wreckage. “If nothing else, we can say with confidence that it isn’t where we looked,” Ocean Infinity CEO Oliver Plunkett said after the latest phase ended.
Could investigators have gotten something wrong? Was there a hidden weakness in the analysis that sent search ships thousands of miles into the southern Indian Ocean? The suspicion isn’t new. When Curtin University researchers discussed the low-frequency signal in 2014, they cautioned that it appeared incompatible with the satellite handshake data, regarded at the time as the most reliable source of information in the case. In plain terms: The satellite-derived search line might be wrong.
Today, Dall’Osto still has doubts. “I agree with Dr. Kadri that if the airplane crashed, it could have made a sound loud enough to be heard,” he says. “And if it was detected, it wasn’t on the seventh arc.”
Dall’Osto thinks the northwest Indian Ocean signal could be the acoustic signature of MH370’s impact. If he’s right, that means searchers have spent the last 12 years looking in the wrong place.
If Ocean Infinity’s next search is also a bust, Dall’Osto says, it may be time to stop treating the seventh arc as the only place MH370 could be. “There was a sound that was made right there,” he says. “It’s not like it was made along some 6,000-mile-long arc. It was made at this spot right here.”