Mathematical Model finds Acoustic Signal that May Predict Earthquakes

  • Thread starter jedishrfu
  • Start date
  • Featured

"Previous machine-learning studies found that the acoustic signals detected from an earthquake fault can be used to predict when the next earthquake will occur," Ke Gao, a computational geophysicist at Los Alamos National Laboratory, ... "This new modeling work shows us that the collapse of stress chains inside the earthquake gouge emits that signal in the lab, pointing to mechanisms that may also be important in Earth."


Sounds promising, thanks for posting it. From the summary article:
Thanks to the simulations, scientists are beginning to understand how acoustic signals can reveal the evolution of stress within a fault structure. Eventually, these acoustic signals could be used to predict earthquakes many hours, perhaps days, in advance.
Paging @davenn


Science Advisor
Gold Member
Thanks for the link, @jedishrfu , an interesting read

Thanks @berkeman for the heads up :smile:

Looks to be a promising area of research. I'm not one to say that short term quake prediction will
never happen, just that it's going to be very difficult
The high-tech version of an ominous creak in the timberwork.

It might be interesting to listen to an audio version of the signal, with the playback speed suitably adjusted for human hearing.

They may have provided the raw data under "supplemental material"
(I don't have access.)

Random thought: can pigeons be trained for this task?
Last edited:
I like the one they created for the black hole merger, It sounded like a drop of water falling into a hole. Bloop!

There's some interesting underwater sounds that have baffled scientists. A few have been tracked down but others are still a mystery.

and this one with recordings

Re. the singing sand dunes (previous post)...
Some key inputs from the article:
It's not necessarily the motion of the sandy ocean that determines the pitch of the note—it's the size of the grains, though why the size matters is still unknown. ... The one-note Moroccan sand grains are almost entirely the same size—160 millionths of a meter, or microns across—but the noisy Omani sands run the gamut, from 150 to 300 microns.

But when the messy sands were sieved down to just the 200-to-250 micron particles, the tone cleared into a single tone. "The size of the grain controls the actual sound," Dagois-Bohy concluded.
Why exactly this happens, and how the sound itself is created, is still uncertain.
Can we apply dimensional analysis to connect the ~ 300 micron sand size to the ~100 Hz frequency, and thus get some clue about the kinds of things that may be going on? How can we get the latter order of magnitude from the former?


Science Advisor
Can we apply dimensional analysis to connect the ~ 300 micron sand size to the ~100 Hz frequency, and thus get some clue about the kinds of things that may be going on? How can we get the latter order of magnitude from the former?
The speed of sound in grains that size puts their acoustic resonant frequency in the megahertz range. So that's one possibility eliminated.

Let's try the time between impacts between the grains, and assume the average spacing is one grain.
  • grain size 3E-4 meter; 100Hz = 10mS between impacts
  • in 1 sec. a grain would move 3E-2 meter, or 1.2in.
  • in 1 min. it would move 1.8 meter, or just under 6ft.
Does this seem a ballpark estimate for the speed of sand movement?

If not, how about if a grain impacts another grain at two to three times the grain spacing? That would double or triple the travel speed. My wild guess is they would hit at the most every other grain, traveling at least 3.6M/min, 11.5ft/min.

Hmm... still seems slow. Hope it helps someone else come up with a conjecture.



Science Advisor
I expect the sound is produced from movement within the dune, not from the exposed surface. The fact that it has a clear tone suggests it is composed of many small identical movements that are coupled in some way.

If you take a bag of glass marbles and wring it between your hands it will produce a distinctive growl as the marbles slip and stick against each other.

Consider three interlocked gear wheels, they cannot roll. To roll they must be separated far enough for one gear pair to disengage. Likewise, close-packed spheres cannot roll against each other because every group of three in contact is locked by friction. Each grain contacts up to 12 others which effectively locks a dune in place.

In order that a shear zone form within a dune, a slip of a plane must take place. That is unlikely as adjacent planes are well interlocked. So, imagine one grain somewhere separates from one third of it's neighbours, immediately that would unlock a group and a rotation could take place that might then propagate as a wave along a shear zone through the body of the dune. Each rotation frees the way for the next, and so on.

The question then becomes, what is the minimum movement that will self propagate and how much time does it take each movement to propagate? The reciprocal of that is the audible frequency.

Maybe it can be modelled as a propagation of holes into each of which a grain falls to create another hole, to be modelled as a gravity wave. So how long does it take a grain to free-fall its own radius?
r = ½ ⋅ g ⋅ t2;
t = √( 2 ⋅ r / g )
For 0.2 mm diameter grain, r = 1e-4 metre; g = 9.8;
T = 4.5 millisec; f = 220 Hz.

Want to reply to this thread?

"Mathematical Model finds Acoustic Signal that May Predict Earthquakes" You must log in or register to reply here.

Related Threads for: Mathematical Model finds Acoustic Signal that May Predict Earthquakes

  • Posted
  • Posted
  • Posted

Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving

Hot Threads