What’s Making Earth Hum?

Seven years ago, seismologists discovered ^1–3 that Earth is vibrating in a discrete set of frequencies between 2 and 7 mHz. The peak frequencies coincide with the spheroidal fundamental modes predicted by a model of Earth. To find these fundamental normal modes, which are masked by earthquakes, seismologists had to look exclusively at data from quiet sites. They also needed the broadband, low-noise seismometers available since the 1980s.

Decades ago, seismologists had seen the so-called microseismic noise at frequencies that peak near 0.2 Hz. Microseisms are known to be caused by ocean-wave interactions, which generate pressure fields that do not wane with ocean depth. The same mechanism cannot explain the generation of the much lower frequency normal modes. Some researchers had speculated that these fundamental oscillations were instead caused by atmospheric turbulence interacting with the solid Earth. Recently, Junkee Rhie and Barbara Romanowicz of the University of California, Berkeley, presented evidence that the oceans play a role. ⁴ Their analysis shows that the normal-mode oscillations originate in the Northern Pacific during the boreal winter and in the southern oceans in the austral winter.

The discrete normal modes occur continuously, so they cannot be caused by intermittent large earthquakes. Nor can they result from an accumulation of many small earthquakes; there’s not enough energy to drive the observed oscillations.

The source must be near Earth’s surface because the fundamental modes are all Rayleigh waves, which propagate largely at the surface. Some researchers suggested that the oscillations are caused by random atmospheric pressure fluctuations distributed uniformly over both land and sea surfaces. The normal-mode amplitudes increase in January and July—times of particularly strong atmospheric disturbances. ⁵ Researchers have calculated that random atmospheric fluctuations can generate seismic spectra consistent with those observed. ^6–8

In their recent work, Rhie and Romanowicz used two arrays of seismometers, one in California and the other in Japan. The Berkeley researchers assumed that a Rayleigh wave arrived at each array from some arbitrary direction. They looked for the direction in which the signals recorded on the stations of the array showed the greatest coherence. As seen in the figure, the Rayleigh waves appear to come from the very regions where strong winter storms roil the seas. Rhie and Romanowicz speculate that Earth’s normal-mode oscillations are generated by nonlinear interactions between atmosphere, ocean, and sea-floor, probably through the conversion of storm energy to low-frequency “infragravity” ocean waves9 that interact with sea floor topography. Identifying the exact mechanism is a work in progress.

Toshiro Tanimoto of the University of California, Santa Barbara, has proposed a similar scenario. ¹⁰ Estimating the size of sea-floor pressure from existing measurements, he finds sufficient energy in these perturbations to excite the observed seismic signals.

Kiwamu Nishida and Yoshio Fukao of the University of Tokyo have completed a preliminary analysis that, like the Berkeley study, suggests that normal-mode oscillations are excited in the ocean areas in northern and southern winters. The results do not necessarily mean, says Fukao, that Earth’s oscillations are caused by oceanic sources rather than atmospheric loads on the sea surface. “We do not know which is more efficient as the excitation mechanism,” he adds. Both Fukao and Romanowicz think that the coupling of ocean waves to the sea floor may occur near shore or perhaps at the continental shelves.

Exploiting the noise

By studying Earth’s normal-mode oscillations, researchers can learn more about the coupling of the atmosphere, oceans, and solid Earth. Seismologists are learning to embrace the noise signals they once regarded strictly as an impediment, especially in the microseismic frequency band above 0.1 Hz. They have, for example, learned that they can extract information about Earth’s crust by analyzing the propagation of noise signals between two seismic stations. ¹¹ The noise in historical seismic data is also being used to track long-term variations in wave climate.