How well will Earthquake Early Warning work? (Reader Poll)

By David Jacobson, and Ross Stein, Temblor

Check your hazard rank

Update: It now appears that the earthquake early warning system did not go off prior to shaking and that the video below has been altered

Wouldn’t it be great to be warned? Shaking from a scenario M=7.8 quake (dashed black line) on the southern San Andreas Fault reaches Los Angeles about 80 seconds after the rupture nucleates in the Salton Sea. Source: Geoffrey Ely, Temblor.


In the M=7.1 Puebla, Mexico earthquake on September 19, which caused significant damage in Mexico City, the city’s earthquake early warning sirens went off approximately 12 seconds before the strongest ground shaking (See video below – sirens go off at the 37 second mark). While the onset of shaking appears to begin only a few seconds after the sirens, that should have been enough for people to take cover underneath tables or desks. Coincidently, this earthquake occurred on the 32nd anniversary of a devastating M=8.0 earthquake, the quake that prompted the creation of the Earthquake Early Warning system. This is the second earthquake early warning that the capital has received in the last two weeks. The September 7 M=8.1 Chiapas earthquake, which caused little to no damage in Mexico City was triggered a 60 second warning.


Although this is the type of earthquake early warning that people desire, Mexico City is unique: The capital city is located on a geological “water bed,” and so gets shaken even by earthquakes 400 km (250 mi) away. In all likelihood, earthquake early warning elsewhere will contend with uncertainties and false alarms, or at least nuisance alerts. Just as important, in most instances that matter, we will only get a few seconds of warning, just enough time to ‘drop, cover, and hold on.’

This Temblor figure highlights how much of Mexico City is susceptible to extreme soil amplification during earthquakes.


Stirring up the hornet’s nest

At the annual meeting of the Southern California Earthquake Center last week, USGS seismologist Sarah Minson took a deeper look at what we can expect in the U.S., focusing on ideal performance under perfect conditions, so that its theoretical limits can be understood. Earthquake Early Warning lies at the unexplored interface of seismology and sociology, and so to make the warnings beneficial, both its capabilities and weaknesses must be probed. Her presentation triggered a vigorous discussion by the proponents and skeptics of this important technology.


Three types of Warning Systems

Around the world, several countries, including the United States, are either in the process of developing, or have already deployed, earthquake early warning systems. While each system has its own unique traits, they all rely on the ability to detect different types of seismic waves, and alert people before strong shaking arrives. The fastest seismic waves are P waves, which are generally undetectable to humans, but can be picked up by high quality seismometers. Next are S waves, which cause strong ground shaking and can result in significant damage to man-made structures. Because P waves travel roughly 60% faster than S waves, if the P wave can be reliably detected, in theory one should be able to alert people that damaging S waves will soon arrive. The farther away from the quake you are, the longer the warning time, but also the weaker the shaking. So, there is a ‘sweet spot’ that optimizes the timeliness and value of the warnings.

In a nutshell, there are three types of earthquake early warning systems. The simplest of these is a stand-alone P-wave detector that alerts users if a P wave is detected (Wu et al., 2013). This approach does not depend on telemetry or fast cloud computing, and can be used anywhere in the world. But rarely, it can be ‘spoofed’ by vibrations when there is no earthquake, and it can give no indication of how strong the shaking will be. In Taiwan, where 500 Palert instruments have been in place for a decade, with warning times of 4-5 seconds for most locations that were strongly shaken by the M=6.5 quake in 2016. That’s enough time to duck for cover, and so they worked.

The second type of system involves a network of sensors that detects S waves that are at least 30 km (20 mi) away. As more seismographs detect these S waves, updated projections of how far the waves will travel and how strong they will be can be given. This is how earthquake early warning will soon be implemented in Japan’s second-generation system (‘PLUM’) (Kodera et al., 2016). The prototype has been successful in warning before subduction zone earthquakes, since they are offshore and so provide long warning times. The last type of system, ShakeAlert, is what the USGS, UC Berkeley, Caltech, and University of Washington and University of Oregon are jointly developing (Strauss and Allen, 2016). ShakeAlert, which will have a limited rollout next year, is designed to not only determine the location and magnitude of the earthquake, but will also alert people how strong the shaking will be like at their location.


The difficulty of warning only for strong shaking

But here’s the nut of Dr. Minson’s message: In order to give sufficient warning, the shaking threshold for the alert has to be set very, very low. So, while earthquake early warning could very well save lives and reduce injuries, getting useful alerts comes at a steep cost—and that’s quite apart from money. Most alerts you receive will not be followed by strong shaking, and some shaking will be imperceptible. In these instances, the alert could be considered a nuisance. While the shaking threshold required to trigger an alert could be raised, doing so would increase the “blind zone,” the area in which no warning could be given. So, if you only want to be warned when there will be strong ground shaking (say, because scramming your factory is costly or disruptive), you will almost never be forewarned.

We are tracking clicks, nothing will change on the screen when you vote. We will place the results here at the end of the week.

65% of people prefer a longer warning


Nuisance alerts or teachable moments?

What all of this means is that for earthquake early warning to work, we have to accept the idea that several times a year, we’ll get an alert when shaking is very light. Some may see these as an inconvenience, and others could lose confidence in the system.

But there is a flip side to this: Perhaps these warnings provide a natural opportunity to prepare for what we should do in the event of an earthquake. In a sense, these might serve as unscheduled and unannounced fire alarms. That could help all of us train for what we’ll need to do with several seconds of warning so we protect ourselves in the event of strong shaking. Don’t run out of the house or office, and don’t call Mom. Instead, dive under a table or desk. This preparedness training could be the real pearl of Earthquake Early Warning: In the end, it’s not the technology but the muscle memory that will count.



Jennifer A. Strauss and Richard M. Allen, Benefits and Costs of Earthquake Early Warning, Seismological Research Letters Volume 87, Number 3, May/June 2016

Yuki Kodera, Jun Saitou, Naoki Hayashimoto, Shimpei Adachi, Masahiko Morimoto, Yuji Nishimae and Mitsuyuki Hoshiba, Earthquake early warning for the 2016 Kumamoto earthquake: performance evaluation of the current system and the next-generation methods of the Japan Meteorological Agency, Earth, Planets and Space (2016) 68:202

Yih-Min Wu, Da-Yi Chen, Ting-Li Lin, Chih-Yih Hsieh, Tai-Lin Chin, Wen-Yen Chang, Wei-Sen Li, and Shaw-Hsung Ker, A High-Density Seismic Network for Earthquake Early Warning in Taiwan Based on Low Cost Sensors, Seismological Research Letters Volume 84, Number 6, November/December 2013

M.-A. Meier, J. P. Ampuero, T. H. Heaton, The hidden simplicity of subduction megathrust earthquakes, Science 357, 1277–1281 (2017) 22 September 2017