California Earthquake Early Warning tested by moderate Bay Area earthquakes

By Alka Tripathy-Lang, Ph.D. (@DrAlkaTrip)
 
Warnings were issued for the magnitude-4.5 and magnitude-4.7 quakes 5-6 seconds after they ruptured via ShakeAlert, and alerts reached at least one beta tester of the MyShake app about 2 seconds later. While important progress, these intrinsic delays mean that those who experienced strong shaking received no warning. Alert delays may be longer as the number of alert recipients grows.
 
Citation:Tripathy-Lang, Alka (2019), California Earthquake Early Warning tested by moderate Bay Area earthquakes, Temblor, http://doi.org/10.32858/temblor.55
 

Earlier this month, the Bay Area was gently reminded, twice, of the seismic hazards posed by the numerous faults lacing California. A magnitude-4.5 earthquake struck near Pleasant Hill on 14 October 2019, and a magnitude-4.7 temblor shook Hollister on 15 October 2019, with both felt throughout the greater Bay Area.

Two days after the second quake, on the day of the Great ShakeOut and marking the 30th anniversary of the Loma Prieta quake, the California Governor’s Office of Emergency Services publicly released the state’s Earthquake Early Warning system. The early warning system consists of two components. The first is ShakeAlert, which is managed by the U.S. Geological Survey in collaboration with State and university partners and is designed to forecast shaking as soon as possible after a quake is detected. The second component comprises separate alert distribution systems. 

 

 

These alert distribution systems include the Wireless Emergency Alert system (think AMBER alert) and the newly updated MyShake App, which is now California’s official earthquake early warning app. Both will deliver shake warnings directly to Californians’ cell phones.

 
ShakeAlert provides the forecast …

According to Sarah Minson, a research geophysicist at the U.S. Geological Survey, ShakeAlert works by detecting an earthquake, estimating location and initial magnitude, predicting the ground motion, and producing a shaking forecast, or “ShakeAlert”. All of this is accomplished within seconds.

Minson explains that after the initial alert is released to the alert distribution system, ShakeAlert continuously updates the forecast as data continue to come from seismic stations farther from the epicenter. If additional regions require shake warnings, new alerts are generated and sent to the alert distribution systems. “Not everyone can be warned by the first alert,” she says, “because big or small, earthquakes start the same, and most small earthquakes don’t grow up to be big ones.” 


For the magntiude-4.5 and magntiude-4.7 earthquakes that jostled the Bay Area in mid-October, Minson says ShakeAlert took 5.6 and 5.1 seconds, respectively, to issue a forecast after these earthquakes nucleated underground. ShakeAlert then sent the forecast to the alert systems for public notification.

 
… and MyShake distributes the alert

A beta version of the early warning-enabled MyShake app was put to the test prior to its pre-planned public release, alerting beta users of shaking for both the magnitude-4.5 and magnitude-4.7 events. Jennifer Strauss, MyShake project manager at the Berkeley Seismology Lab says that “up until release [last Thursday], we had a little over 100 beta testers.” One of the testers was Peggy Hellweg, operations manager for BSL, a co-PI for ShakeAlert and former commissioner of the California Seismic Safety Commission.

 

 

She happened to be one of two MyShake beta testers to receive a push notification via the app for the magnitude-4.5 Pleasant Hill earthquake. Hellweg was in downtown Orinda, east of Berkeley, and says it took 2.16 seconds for her phone to receive the alert, which arrived about 2 seconds after shaking. She emphasizes “For this earthquake no beta tester who experienced moderate or higher shaking was alerted BEFORE the shaking arrived.”

 
The late alert zone

Hellweg was at the edge of what seismologists call the ‘late alert zone’. This is the area around the epicenter that is too close to receive an alert before shaking begins.

 

From the epicenter of the 2014 magnitude 6.0 Napa earthquake (red star), the P-wave (yellow circle) raced ahead of the more damaging S-wave (red circle). The late alert zone, formerly referred to as the blind zone, for which no warning was possible, is shown by the solid gray circle with a red slash. Small green circles show seismic stations. Credit: Incorporated Research Institutions for Seismology (IRIS).
 

There are two components to the late alert zone. The first is the time it takes for the first P-wave to be detected after an earthquake nucleates. The second is the time it takes to calculate the forecast and distribute the alert.

Strauss says that “if you’re right on top of the earthquake, you will not receive an alert [before shaking starts],” because the quake’s waves have yet to be detected by the nearest seismometer. For Hellweg’s experience in the magnitude-4.5 Pleasant Hill earthquake, which had a late alert zone with a radius of around 14.5 miles (24 km) according to Minson, the total time from earthquake origin to the arrival of the push notification on her phone was about 7.8 seconds—5.6 seconds from origin to ShakeAlert forecast, and 2.16 seconds for MyShake to push the alert to Hellweg’s phone— about 2 seconds after she felt the shaking.

A recent review of Earthquake Early Warning systems by Allen and Melgar [2019] says “with a sparse [seismic] network, the [late alert] zone is larger.” It follows that with a denser seismic network, the late alert zone should decrease. But how might scientists increase the number of seismometers in a region already populated with many seismic stations?

 
Citizen science project

Prior to the new release with earthquake early warning capability, the MyShake app was primarily a citizen science project that harnessed the technology that orients a smartphone screen, creating many thousands of mobile earthquake monitoring stations [Allen et al., 2019]. The app collects data only when it distinguishes earthquake ground motion from everyday movement. MyShake’s distribution platforms, Google Play and Apple iTunes, have upward of 3 billion users between them. If every user downloaded the app, the potential reach is staggering.

“If you can combine MyShake phone data with that from traditional seismic instruments, you can increase the confidence of the forecasts, and potentially decrease the alert time,” says Strauss of MyShake’s potential to create an incredibly dense seismic network in urban areas by turning users into walking seismic stations. In other words, MyShake could help reduce the size of the late alert zone by detecting an earthquake before traditional seismic stations can.

 
Exploring crucial bottlenecks

An often ignored point, says Ross Stein, a geophysicist and CEO of Temblor, is that “no one has figured out how to simultaneously alert millions of people.” Aside from message delivery problems related to cell carrier, mobile device and cell tower (according to the FCC), the WEA system takes seconds to minutes to deliver alerts, according to the USGS.

In the Pleasant Hill earthquake, a push notification via MyShake was sent to Hellweg in 2.16 seconds, which is encouraging compared to the WEA system.

 

Video clip showing the current MyShake push notification that Hellweg received during the magnitude-4.5 Pleasant Hill earthquake. It includes a verbalized warning from the app that, Hellweg explains, is triggered as soon as the push notification is received. Credit: MyShake: https://myshake.berkeley.edu/
 

However, scaling from a single user notification to millions of people is an area of active exploration, according to Strauss. To tackle this, she says that “every step of the process in MyShake is timestamped. Are our servers the bottleneck? Or the cell carriers? Or cloud services? ShakeAlert [and MyShake] is at the point where we have to start testing at scale. We can’t sit back and wait for the system to be perfect.”

 
What people want

“Now that ShakeAlert is live, people can give feedback because we need to know what people want, and people may not know what they want until an earthquake happens,” says Minson, in part referencing ShakeAlert’s decision to warn the public of felt shaking, instead of strong shaking after complaints about the Ridgecrest earthquake alert. She continues, “if we practice a better-safe-than-sorry approach, we can warn the public of a small, distant earthquake that has yet to grow into a big one.”

In the end, ShakeAlert’s goal “is to determine the intensity of shaking where you are, and whether that will be damaging to you,” says Wendy Bohon, an earthquake geologist at IRIS who was not involved in developing ShakeAlert. She continues, “ShakeAlert is not an earthquake prediction. It is an alert issued after an earthquake has already begun. If you feel shaking or receive an alert, immediately take protective action. Drop. Cover. Hold On.”

 
 
References
Allen, R. M. and Melgar, D., 2019, Earthquake Early Warning: Advances, scientific challenges and societal needs, Annual Review of Earth and Planetary Sciences, 47, 361-388. doi: 10.1146/annurev-earth-053018-060457

Allen, R. M., Q. Kong, R. Martin-Short (2019). The MyShake Platform: A Global Vision for Earthquake Early Warning. Pure and Applied Geophysics: doi: 10.1007/s00024-019-02337-7