Earthquake Early Warning: Gravity changes beat seismic signals

By Jean Paul Ampuero, Caltech Seismological Laboratory; Université Côte d’Azur, IRD, Géoazur

Click Here for Spanish Version

The devastating 2011 Tohoku earthquake and ensuing tsunami caused billions of dollars of damage and the deaths of thousands. A new study, using data from this quake, suggests that small gravity changes are the earliest earthquake early warning signals. Photo from: SFDEM


This story starts a few years ago, when astrophysicists in search for gravitational waves from the distant universe crossed paths with seismologists starving for new clues about how earthquakes work beneath our feet. Someone’s noise soon became someone else’s signal, indeed a very unique signal: the earliest harbinger of earthquake shaking that nature and physics have to offer.

Earthquakes move mass around, in enormous quantities. This is obvious to anyone who has been mesmerized by the view of fault offsets of several meters left at the Earth’s surface after a large earthquake. But mass is also redistributed temporarily by seismic waves, even before the earthquake is over. For example, P waves compress and dilate the rock they travel through, perturbing the rock’s density momentarily. These static and dynamic mass perturbations are natural sources of gravity changes … and gravity changes travel remotely at the speed of light!

Earthquake early warning (EEW), which aims at alerting people and automated systems seconds before strong shaking arrives, is one of the important contributions of modern seismology to society. But current EEW systems have a fundamental limitation: the natural information carrier they rely on, P waves, travels only about twice as fast as the natural damage carrier they try to anticipate, S waves. Just like lightning warns us of impending thunder, speed-of-light gravity changes are, in principle, the ultimately-fast earthquake information carrier.

Jean Paul Ampuero at the Temblor booth at the annual Great ShakeOut event at the Museum of Natural History in Los Angeles.


Our team, a mix of physicists and seismologists in the US and Europe, used pen-and-paper and supercomputers to make a first theoretical estimation of how large these early gravity signals could be (Harms et al, 2015). The results looked “promising”: observing the phenomenon with current instrumentation promised to be a nice challenge. Our best bet was then to look for recordings of the huge 2011 Tohoku, Japan earthquake by a superconductive gravimeter installed in a quiet underground site, 500 km away from the epicenter, and by nearby broadband seismic stations. A blind statistical analysis of the data (of the type our gravitational-wave astrophysics colleagues are used to) revealed evidence of a signal preceding the P waves (Montagner et al, 2016). But it was not the smoking gun one would have hoped for. Moreover, my Caltech colleague Prof. Tom Heaton pointed out (Heaton, 2017) that our theory did not account for a potentially important feedback of gravity changes on elastic deformation, which I describe below.

The smoking gun and a more complete theory of early elasto-gravitational earthquake signals are finally reported in our paper published this week in Science Magazine (Vallée et al, 2017). We found that broadband seismometers in China located between 1,000 and 2,000 km away from the epicenter recorded, consistently and with high signal-to-noise ratio, an emergent signal that preceded the arrival of P waves from the Tohoku earthquake by more than one minute. These signals are well predicted by the results of a new simulation method we developed to account for the following physical process. The gravity perturbations induced directly by earthquakes (those studied by Harms et al, 2015) also act as distributed forces that deform the crust and produce ground acceleration. Gravimeters and seismometers are inertial sensors coupled to the ground, they actually record the difference between gravitational acceleration and ground acceleration. Sometimes these two accelerations are of comparable amplitude and tend to cancel each other, thus it is important to include both in simulations.

This figure, modified from IPGP, 2017, shows the signal picked up by a seismometer in the time preceding and following the 2011 M=9.1 Tohoku earthquake. What is important to see in this figure is that there is a 45-60 second window from when the prompt signal drops below normal background rates, until a P wave can be felt. This represents the potential earthquake early warning time. (Figure from Vallée et al., 2017)


How can we use these results to improve current EEW systems? Elasto-gravitational signals carry information about earthquake size but are weak and do not have a sharp onset. We had to use very distant seismic stations and wait more than one minute after the Tohoku mega-earthquake started to see its elasto-gravitational signals on conventional seismometers. This seems too long a wait for an EEW system, but it is enough to significantly accelerate current tsunami warning systems. Indeed our simulations show that the Chinese stations could distinguish earthquakes in Japan with Mw

To develop the full potential of elasto-gravity signals for EEW (and, more fundamentally, for earthquake source studies) we need to develop new, more sensitive instruments. We can leverage on technological advances in gravity gradiometry for low-frequency gravitational wave (GW) detection. The GW detections that led to the recent Nobel Prize were achieved at frequencies of about 100 Hz and required huge facilities, but the GW astronomy community is also interested in observing GW signals in the 0.1-1 Hz band with much lighter and smaller (meter scale) instrumentation. The sensor requirements for EEW are much less stringent than those for GW detection, and should be achieved much sooner.

My personal affair with this new field of gravitational seismology started with a scholar chat at the Caltech Seismolab with Jan Harms, who was then a LIGO postdoc, and continued soon after with my old-time friends from IPG Paris. It has been wonderful to experience first-hand that EEW research is not only about operational and engineering aspects, but also about fundamental physics problems. I also find it exciting that the ongoing revolution of gravitational wave astronomy will not only open new windows into the distant Universe but also into our own vulnerable Earth.



J. Harms, J. P. Ampuero, M. Barsuglia, E. Chassande-Mottin, J.-P. Montagner, S. N. Somala and B. F. Whiting (2015), Transient gravity perturbations induced by earthquake rupture, Geophys. J. Int., 201 (3), 1416-1425, doi: 10.1093/gji/ggv090

T. H. Heaton (2017), Correspondence: Response of a gravimeter to an instantaneous step in gravity, Nature Comm., 8 (1), 966, doi: 10.1038/s41467-017-01348-z

J.-P. Montagner, K. Juhel, M. Barsuglia, J. P. Ampuero, E. Chassande-Mottin, J. Harms, B. Whiting, P. Bernard, E. Clévédé, P. Lognonné (2016), Prompt gravity signal induced by the 2011 Tohoku-oki earthquake, Nat. Comm., 7, 13349, doi: 10.1038/ncomms13349

M. Vallée, J. P. Ampuero, K. Juhel, P. Bernard, J.-P. Montagner, M. Barsuglia (December 1st 2017), Observations and modeling of the elastogravity signals preceding direct seismic waves, Science, doi: 10.1126/science.aao0746