Research to improve ShakeAlert earthquake early warning products and their utility
Authors:
- Elizabeth S. CochranECElizabeth S. Cochran
- Brad T. AagaardBABrad T. Aagaard
- Richard M. AllenRARichard M. Allen
- Jennifer AndrewsJAJennifer Andrews
- Annemarie S. BaltayABAnnemarie S. Baltay
- Andrew J. BarbourABAndrew J. Barbour
- Paul BodinPBPaul Bodin
- Benjamin A. BrooksBBBenjamin A. Brooks
- Angela ChungACAngela Chung
- Brendan W. CrowellBCBrendan W. Crowell
- Douglas D. GivenDGDouglas D. Given
- Thomas C. HanksTHThomas C. Hanks
Open Access
- 1 January 2018
- report
- Published by US Geological Survey
Abstract
Earthquake early warning (EEW) is the rapid detection of an earthquake and issuance of an alert or notification to people and vulnerable systems likely to experience potentially damaging ground shaking. The level of ground shaking that is considered damaging is defined by the specific application; for example, manufacturing equipment may experience damage at a lower intensity ground shaking than would cause damage to a building. Along the West Coast of the United States, the warning times for ground shaking could range as high as tens of seconds for moderate levels of ground shaking, or potentially longer, if a lower ground-shaking threshold is used to issue alerts. However, it is not always possible to provide advance warning of ground shaking, particularly for locations close to an earthquake that are most likely to experience very strong ground shaking. EEW alerts may be useful to individuals who can use a few seconds to move to a safe zone and to electromechanical systems that can take automatic actions to reduce damage and injuries. An EEW system, ShakeAlert, has been under development in the United States since 2006. Federal and State governments, as well as the private sector, are now investing in the ShakeAlert prototype system that will, when completed, become an operational public system for the West Coast of the United States.While the current prototype is delivering alerts to test users, improvements to the accuracy, timeliness, and utility of the alerts are needed. For this reason, it is essential that the ShakeAlert system be continuously improved through targeted research, involving not only the current ShakeAlert partner organizations, but also the broader scientific, engineering, and emergencyresponse communities. To this end, this report describes the opportunities for improvement that can be addressed through research and development over the next 5 years.Our recommendations are organized into four areas: (1) understand EEW capabilities and user needs, (2) make alerts as fast and accurate as possible, (3) ensure reliability when it counts, and (4) explore the use of new instrumentation.The first challenge is to understand EEW capabilities and user needs. EEW must deliver actionable information to people and to automated systems to mitigate short- and longterm impacts of damaging ground shaking, so development of EEW must be motivated by the needs of users. Within this challenge, we must study the technical capabilities and limitations of EEW in general, and the ShakeAlert system specifically. This includes development of performance metrics that assess the timeliness and accuracy of alerts to understand the value and utility of the ShakeAlert EEW product(s) for various user groups, including different industry sectors, emergency-management agencies, and the public. Research is needed to define the alerting choices that maximize the utility of the system for users and to determine what the available communication pathways are for providing timely alert information. Additionally, we engage users to assess how alerts will be used by different sectors to mitigate losses and to inform EEW product design. Further, social-science research is needed to develop alert messaging, including what relevant prior and follow-up information are required, to ensure effective use of alerts. The second challenge is to make alerts as fast and as accurate as possible. The timeliness and accuracy of an EEW alert is important because it will set in motion a series of actions and downstream products. An EEW alert will trigger notification across emergency-alert systems and across multiple communication channels to populations in impacted regions. The EEW alert region may grow as the earthquake fault-rupture length increases, and the EEW system’s characterization of it, evolves. We must continue research into new or improved seismic and geodetic waveform-processing methods necessary to rapidly characterize the expected ground shaking and associated uncertainties. It is important to thoroughly evaluate whether new methods improve alerts through more accurate ground-motion estimates and (or) reduced latencies (that is, longer warning times). New methods could include tracking the extent of a large rupture in real time (known as finite-fault algorithms) and ground-motionbased EEW algorithms. Additionally, ground motion predictions could be optimized for each earthquake as the earthquake fault rupture progresses by using, for example, event terms to shift ground-motion curves for more (or less) energetic ruptures. The third challenge is to ensure reliability when it counts. This challenge requires us to explore approaches that assess the expected performance of ShakeAlert across the range of earthquake magnitudes, locations, and depths that may occur within the alerting region. Large, damaging earthquakes and their associated aftershock sequences matter most for hazard and for EEW, but these large-earthquake sequences occur infrequently. We expect ShakeAlert to respond robustly to these large-earthquake sequences despite potentially long periods of relative seismic quiescence in the intervening years, and in spite of inevitable communication challenges that arise during and after a large earthquake. We must develop methods to utilize the broadest available datasets to test EEW performance, including ground-motion data recorded in other parts of the world. The observational period for large, damaging earthquakes in any particular region has been short in comparison to estimated large-earthquake recurrence times. Ground-motion records for very large, damaging western United States events and major aftershock sequences do not yet exist, nor do data exist for all potential sources of noise and spurious signals that ShakeAlert must be “tuned” to reject. In addition, robust synthetic data could provide the flexibility to test a wider range of earthquake magnitude, tectonic-setting,...Keywords
This publication has 64 references indexed in Scilit:
- Systematic Observations of the Slip Pulse Properties of Large Earthquake RupturesGeophysical Research Letters, 2017
- The value of real‐time GNSS to earthquake early warningGeophysical Research Letters, 2017
- How “good” are real‐time ground motion predictions from Earthquake Early Warning systems?Journal of Geophysical Research: Solid Earth, 2017
- MyShake: Initial observations from a global smartphone seismic networkGeophysical Research Letters, 2016
- Kinematic rupture scenarios and synthetic displacement data: An example application to the Cascadia subduction zoneJournal of Geophysical Research: Solid Earth, 2016
- Rupture characteristics of major and great (Mw ≥ 7.0) megathrust earthquakes from 1990 to 2015: 1. Source parameter scaling relationshipsJournal of Geophysical Research: Solid Earth, 2016
- Operational real‐time GPS‐enhanced earthquake early warningJournal of Geophysical Research: Solid Earth, 2014
- Real‐time inversions for finite fault slip models and rupture geometry based on high‐rate GPS dataJournal of Geophysical Research: Solid Earth, 2014
- On robust and reliable automated baseline corrections for strong motion seismologyJournal of Geophysical Research: Solid Earth, 2013
- Real‐time prediction of ground motion by Kirchhoff‐Fresnel boundary integral equation method: Extended front detection method for Earthquake Early WarningJournal of Geophysical Research: Solid Earth, 2013