Earthquake sensors are picking up hurricane signals that aircraft are too dangerous to collect, and scientists say it could finally unlock the storm’s most deadly secret
Image created with artificial intelligenceFlying into a hurricane means accepting that the plane might suddenly drop hundreds of meters with no warning. In 2021, one research flight plunged roughly 500 meters in a single downdraft — about 17 times the worst turbulence a commercial passenger typically experiences. And that’s not even the most dangerous part of the storm.
The lowest layer of a hurricane’s atmosphere, a churning zone just one to two kilometers above the surface, remains largely off-limits to crewed aircraft after a 2003 flight dove so deep that salt air destroyed one of the plane’s engines. Scientists know this boundary layer controls whether a storm intensifies — they just can’t safely get close enough to study it.
The most dangerous layer no one can study
That 2003 engine incident didn’t just ground one flight — it effectively ended crewed research into the hurricane boundary layer altogether. This zone, spanning the lowest one to two kilometers of the storm’s atmosphere, is where wind and sea interact most violently, driving the rapid intensification that makes hurricanes so hard to forecast. Atmospheric scientist Ipshita Dey, who worked on the Stanford study, calls it “the most underobserved part of the hurricane” — critical for prediction, yet nearly impossible to access safely.
NOAA researchers deploy parachuted sensors dropped from higher in the storm, remotely operated drones, and surface tools like buoys and wind-measurement towers.
Scientists haven’t abandoned the zone entirely. NOAA researchers deploy parachuted sensors dropped from higher in the storm, remotely operated drones, and surface tools like buoys and wind-measurement towers. Each approach carries real limits. Airborne sensors can’t provide continuous measurements, and surface instruments capture only the very bottom of the atmosphere — often recording only intermittently. None of them can directly observe turbulence at the frequency and resolution that forecasters actually need.
How a 2012 hurricane became an accidental experiment
The opportunity arrived unexpectedly in August 2012, when Hurricane Isaac made landfall along the Louisiana coast — directly over a pre-existing network of seismoacoustic sensors. These instruments were built to detect ground vibrations and infrasound, the low-frequency sound waves below the threshold of human hearing. Volcanic eruptions are a classic source of infrasound. Hurricanes, it turns out, are another.
The physics is relatively straightforward. A hurricane’s turbulent winds create chaotic pressure fluctuations that sweep across the ground surface, and seismoacoustic sensors register that pressure as sound. The Stanford team recognized that a storm passing over an existing sensor network was a rare chance to extract atmospheric data from instruments designed for something else entirely.
Isaac was only a Category 1 storm at landfall — modest by hurricane standards. For the researchers’ purposes, though, that didn’t matter much. “It was a really strong event,” said Qing Ji, who worked on the project as a Stanford student and is now a postdoctoral researcher at the University of Texas at Austin. The turbulence was intense enough to generate clear, usable signals.
A surprising discovery: local signals inside a massive storm
The initial expectation was messy data. A hurricane spans hundreds of kilometers, and as pressure shifts throughout its layers, acoustic waves press against the ground and overlap. The researchers anticipated something closer to noise than signal — a chaotic blur that would be difficult to interpret.
What they actually found was the opposite. The sensors picked up remarkably localized pressure fluctuations, measuring disturbances only tens to hundreds of meters across. That granularity was unexpected for a storm operating at such a massive scale.
“It really shifted our thinking from this very large scale, the whole hurricane scale, down to this few-kilometers-and-less scale, so then we focused on turbulence within that,” said Stanford geophysics professor Eric Dunham. Resolving turbulence at fine spatial scales is exactly what researchers need to understand how the boundary layer drives storm intensification — and it’s precisely the resolution that existing tools struggle to deliver.
A complement, not a replacement, for existing tools
Jun Zhang, the hurricane scientist at the University of Miami who recalled the harrowing 2021 flight, wasn’t involved in the Stanford study. He sees real potential in the findings, though with appropriate caution attached.
“If these relationships they presented in the paper prove robust across many hurricanes…those observations will be useful for model evaluation and physics improvement,” Zhang said. He was careful to frame seismoacoustic sensors as a complement to existing methods, not a substitute. The value is additive: a new, continuous data stream filling gaps that current tools leave open.
The safety advantage is also meaningful. Airborne observations become increasingly dangerous once a hurricane moves over land, while ground-based sensors keep working regardless. The infrastructure barrier is lower than it might seem, too. As Ji noted, many U.S. seismic stations already carry pressure sensors, meaning the network is largely in place — it just hasn’t been used this way before.
What comes next for the research
The immediate priority is validation. The Stanford team plans to analyze seismoacoustic data from multiple hurricanes to confirm that the signatures observed in Isaac reliably reflect atmospheric turbulence — and aren’t an artifact of that particular storm’s behavior or that particular sensor network.
The team also plans to study non-hurricane atmospheric conditions using the same always-on seismic infrastructure. That broader dataset could help establish baseline relationships between ground-level acoustic signals and atmospheric dynamics, strengthening the case for using these sensors in operational forecasting. Both efforts feed into a larger goal: improving the hurricane intensity models that forecasters depend on when a storm is still days offshore.
Whether seismoacoustic sensing becomes a standard part of that toolkit remains an open question. The answer depends on whether the Isaac findings hold up across the full, chaotic variety of real storms — and that work is only beginning.
The complete study can be found here: Qing Ji et al. Turbulent seismoacoustic imprints during a hurricane landfall.Science393,628-632(2026).DOI:10.1126/science.adt7323
Carlos is an engineer with strong expertise in technical and industrial topics. He previously worked at international companies such as Siemens and is multilingual.
