Virtual Seismologist Algorithm
The Virtual Seismologist (VS) method is a Bayesian approach to earthquake early warning (EEW) that estimates earthquake magnitude, location, and the distribution of peak ground shaking using observed picks and ground motion amplitudes, predefined prior information, and envelope attenuation relationships (Cua, 2005; Cua and Heaton, 2007; Cua et al, 2009). The application of Bayes’ theorem in EEW (Cua, 2005) states that the most probable source estimate at any given time is a combination of contributions from prior information (candidate priors include network topology or station health status, regional hazard maps, earthquake forecasts, and the Gutenberg-Richter magnitude-frequency relationship) and constraints from the available real-time ground motion and arrival observations. VS is envisioned as an intelligent, automated system capable of mimicking how human seismologists can make quick, relatively accurate “back-of-the-envelope” interpretations of real-time (and at times, incomplete) earthquake information, using a mix of experience, background information, and real-time data.
The formulation of the VS Bayesian methodology, including the development of the underlying relationships describing the dependence of various channels of ground motion envelopes on magnitude and distance, and how these pieces come together in EEW source estimation, was the result of the PhD research of Georgia Cua with Prof. Thomas Heaton at Caltech, from 1998 through 2004.
Implementation of the VS algorithm into real-time EEW codes began in 2006, and is a continuing effort at the Swiss Seismological Service (SED) at ETH Zürich (Swiss Federal Institute of Technology Zürich). VS implementation efforts at the SED received support from the following sources:
- Swiss Seismological Service (SED), from 2006 – present
- ETH Zürich, from 2006 - present
- EU project SAFER (Seismic Early Warning for Europe, 2006 – 2009)
- United States Geological Survey (CISN ShakeAlert Phase II, Prototype Development of the new CISN Earthquake Alert System (EAS): Collaborative Research with California Institute of Technology, University of California Berkeley, University of Southern California, and the Swiss Federal Institute of Technology Zürich, 2009-2012)
- EU project NERA (Network of European Research Infrastructures for Earthquake Risk Assessment and Mitigation, 2010-2014)
- EU project REAKT (Strategies and Tools for Real-Time Earthquake Risk Reduction, 2011-2014)
- United States Geological Survey (CISN ShakeAlert Phase III, Prototype Development of the new CISN ShakeAlert: Collaborative Research with California Institute of Technology, University of California Berkeley, University of Southern California, and the Swiss Federal Institute of Technology Zürich, 2012 – 2015)
Conceptually, VS maps incoming phase arrival and ground motion envelope amplitude information into continuously updated estimates of earthquake magnitude, location, depth, origin time, and likelihood. Within the VS (and ShakeAlert) context, the likelihood parameter expresses the degree of belief that the incoming data come from a real earthquake, as opposed to non-earthquake related signals. The higher the likelihood value, the more consistent the incoming data are with a regional earthquake. Figure 1 shows the system diagram of the real-time VS codes. Phase picks from a short-term average/long-term average picker are first sent to a Pick Filter. If the picks satisfy certain thresholds, they are sent to the Binder Earthworm phase associator module (Dietz et al, 2002), which provides estimates of event origin time, latitude, longitude, and depth. The location estimate is passed to the Virtual Seismologist module, which provides a magnitude estimate, given the Binder location estimate, and the available ground motion envelope values. The QuakeFilter module then takes the candidate event information (location estimate from the Binder module, magnitude estimate from Virtual Seismologist module), checks the consistency of these estimates with the observed phase picks and ground motion envelope values, and outputs a likelihood value.
The real-time VS codes are currently configured to require a minimum of 4 stations to initiate event declarations. In southern California, on average, the initial VS estimates are available approximately 20 seconds after the earthquake origin time, corresponding to a blind zone of approximately 75 km. (The time to the initial estimate is strongly driven by the station density in a particular region.) Producing faster estimates and hence increasing the available warning time is a high priority of on-going VS research (Meier et al, 2011). An experimental code version of VS, referred to as VS-MTED (Virtual Seismologist Multiple Threshold Event Detection, Fischer et al 2009), employs multiple instances of Binder to allow VS to initiate event declarations based on data from a single station, if the observed amplitudes are high enough. VS-MTED potentially increases the available warning time for large events, where EEW information is relevant. Testing and optimal configuration of VS-MTED is planned to continue.
Figure 1: System architecture of real-time VS codes (graphic from Michael Fischer).
Thus far, VS implementation efforts have focused on real-time processing of incoming waveform data. Inclusion of prior information in real-time processing is still forthcoming. The VS codes have been running in real-time in southern California 2008, in northern California since 2009, and in Switzerland since 2010. The VS software currently running in California and Europe was written by Dr. Michael Fischer during his term at the Swiss Seismological Service (2006-2012).
From 2006 through 2012, VS developments at the SED were carried out under the Real-Time Seismology group, led by Georgia Cua. As VS is transitioned from a research product into standard operations at the seismic network, VS activities at the SED (from July 2012 onwards) will be done through the National Networks Group, led by John Clinton.
VS in Europe
The VS codes have been running in real-time test mode in Switzerland since 2010. Within the EU FP7 REAKT project (2011-2014), VS testing in Europe will potentially be extended to southern Italy (in collaboration with the AMRA and Irpinia Seismic Network group in Naples), western Greece (in collaboration with the University of Patras Seismic Network), Istanbul (in collaboration with the Kandilli Observatory), Romania (in collaboration with the National Institute of Earth Physics), and Iceland (in collaboration with the Icelandic Meteorological Office).
Distribution of VS within the European seismological community will be based on a SeisComP3 platform. The implementation of the VS algorithms within the SeisComP3 framework is supported by the EU FP7 project NERA.
VS in California
VS is among the 3 EEW algorithms contributing to the California Integrated Seismic Network (CISN) ShakeAlert system, the prototype California-wide EEW system being funded by the US Geological Survey (USGS). From 2006 – 2012, ETH has developed, maintained, and operated the VS codes in California. The goal of ETH involvement in Phase III of the CISN ShakeAlert effort (2012-2015) is to transition the operation and maintenance of VS in California to Caltech. Future VS software developments are foreseen as coordinated and collaborative efforts between ETH and Caltech, with the intention that the European and California VS operations are able to leverage each other’s efforts.
Figure 2: Summary of VS real-time detections in northern and southern California from 2009 through 2012.