Seismometers in Louisiana read Hurricane Isaac's eyewall, and the wind term was wrong

The interesting part of this work is not that a storm shows up in seismic data. It is which number the ground is actually responding to.

The Stanford group simulated the turbulent pressure field around a station, calibrated the model against seismometer records and other atmospheric sensors, and then computed the Earth's elastic response from the simulated field [s1c6]. That response matched the observed vertical displacement [s1c7]. The simulations then showed the controlling parameter is the velocity at which pressure fluctuations are carried downstream, not the near-surface wind speed that had been assumed [s1c8]. Anyone converting ground motion into a wind number was using the wrong variable in the conversion. The correction is not cosmetic; advection velocity and near-surface wind speed are different quantities, and any retrieval built on the second one inherits that error.

The sensing argument stands on its own arithmetic. There are 1700 stations in the US network the Louisiana instruments belong to [s1c3], each carrying an infrasound sensor and a seismometer [s1c4], each recording continuously and at a high sampling rate relative to other environmental sensors, as Qing Ji notes [s1c11]. Compare that with the current sources for boundary-layer turbulence: reconnaissance aircraft, ocean buoys, onshore towers and radar [s1c9]. Flying into a hurricane is dangerous, onshore towers are sparse, and continuous coverage of a storm's passage is hard to get [s1c10]. The seismoacoustic network's advantage is not resolution. It is that the instruments are already bolted down, already powered, and do not stop recording when the eyewall arrives.

What the Isaac data resolved was structure: the calm eye, the turbulent eyewall, and the circular bands of thunderstorms and winds as the storm passed over the sensors [s1c5]. That maps onto two of the three parameters forecasters use, which are track, intensity and the radii of the wind layers around the eye [s1c12]. Turbulence in the lowest layer, where momentum, heat and moisture mix between the atmosphere and the surface, is what matters most at landfall [s1c13], and measuring it is critical both for forecasting wind intensity and for validating weather models [s1c14].

The honest limit is that this is one storm, one state, one network, from 2012 [s1c2]. And the retrieval leaned on other atmospheric sensors for calibration [s1c6], which means the method has not yet been shown to work where such sensors are absent. That matters because the place Eric Dunham names as wanting coverage is the southeastern US [s1c15], which is exactly where independent calibration data would be thinnest. Seismoacoustic stations have already been read for ocean wave energy as waters warm and for the evolution of tropical storms and their winds [s1c16]; the atmospheric imprints are less studied, partly because the amplitudes are smaller [s1c17]. Small amplitudes plus sparse ground truth is a harder problem than the Louisiana case, not the same one.

The cheapest version of this is not new hardware. It is reprocessing the archive. A continuous, high-rate record already exists across 1700 sites [s1c3][s1c11], and the storms have already crossed them.