Realtime Test of the ElarmS Earthquake Early Warning Methodology

Richard M Allen, Holly Brown, Margaret Hellweg, Alexei Kireev, Douglas Neuhauser


Modern seismic networks and telemetry systems make it possible to rapidly detect the beginnings of earthquakes. Combining these technical capabilities with recently developed methodologies to estimate earthquake magnitude using just a few seconds of P-wave data (e.g. Allen and Kanamori, 2003) makes it possible to provide a few seconds warning prior to damaging ground shaking. ElarmS is the methodology under development and testing at the Berkeley Seismological Laboratory. It is designed to provide earthquake early warnings by compiling data from multiple seismic stations, and updating those warnings every second as additional data becomes available (Wurman et al., 2007).

Realtime state-wide testing

The Berkeley Seismological Laboratory is a partner in the state-wide effort to test algorithms for the purpose of earthquake early warning. This is an effort by CISN partners (UC Berkeley, Caltech, USC/SCEC, USGS and ETH Zurich) and funded by the USGS. The goal of the project is to implement early warning algorithms within the realtime processing system of the CISN to evaluate algorithm performance in terms of warning accuracy and timeliness.

The ElarmS algorithms have now been ported to continuous realtime processing in California. The system, referred to as ElarmS-RT, provides automated detection, location, magnitude estimation, and ground shaking prediction. As with the previous non-realtime implementations of ElarmS, the output can be displayed as an "AlertMap" which maps the predicted peak ground motion starting one second after the first seismic station to trigger and updates every second. In addition to these early warning products, ElarmS-RT also provides very rapid post-event information. One such product is the ElarmS-ShakeMap. It is available immediately after peak ground shaking has been observed within  100 km of the epicenter (the usual frame size of a ShakeMap) and is the same as the CISN ShakeMap except that the ElarmS magnitude estimate is used and the map is therefore available more rapidly.

Figure 2.59: AlertMap: Ground motion prediction generated by ElarmS-RT using the data available 1 sec after the Alum Rock earthquake was detected.
\epsfig{file=allen08_1.eps, width=8cm}\end{center}\end{figure}

Figure 2.60: ElarmS-ShakeMap using observations of peak ground motion and ElarmS magnitude for the Alum Rock earthquake and typically available within 1.5 minutes. This is nearly identical to the CISN-ShakeMap.
\epsfig{file=allen08_2.eps, width=8cm}\end{center}\end{figure}

Earthquake detections

ElarmS-RT was operational at the time of the October 30th, 2007, $M_w$ 5.4 Alum Rock earthquake near San Jose, the largest earthquake to strike the San Francisco Bay Area since the 1989 Loma Prieta earthquake. The system first detected the earthquake when two seismic stations triggered simultaneously in the epicentral region. One second later, the first magnitude estimate of 5.2 was available and the first AlertMap (Figure 2.59) was calculated. The error in the MMI predictions at this time was 0.1 $\pm$ 0.6 (in MMI units). These are small errors, as can be seen by comparing Figure 2.59 with Figure 2.60 which shows the ElarmS-ShakeMap for the event. As time proceeded, additional data was incorporated and the errors in the ground motion prediction decreased further.

ElarmS-RT currently processes data 15 sec behind realtime. The current telemetry system in Northern California provides most data with 5 to 10 sec delay. By waiting 15 sec $\sim$90% of seismic stations have reported before a particular time increment is processed. Even with this 15 sec delay, ElarmS-RT computers had assessed the hazard posed by the Alum Rock earthquake before ground shaking was felt in San Francisco.

The ElarmS-ShakeMap is generated using the data available 30 sec after the origin time of all detected earthquakes. This time interval allows peak ground shaking to be observed within $\sim$100 km of the event. The ElarmS-ShakeMaps are usually available within 1.5 minutes of the origin time. The ElarmS-ShakeMap for the Alum Rock earthquake is shown in Figure 2.60 and is nearly identical to the CISN ShakeMap. CISN ShakeMaps are typically available in 7 to 10 minutes.


The Alum Rock earthquake was a successful illustration of ElarmS-RT performance. While the realtime testing of early warning in California has only just begun, some of the requirements for a fully implemented system are becoming clear. ElarmS performs well for all events M $>$ 3 in the greater Bay Area where instrumentation is dense and typical station spacing is $\sim$20 km. To the north and south of the Bay Area along the San Andreas Fault System, instrument density drops and typical stations spacing is $\sim$100 km. In these regions, the system works, but is slower, as it takes longer for the radiating seismic waves to be sampled. East of the Mendocino Triple Junction, in the Sierras, and in the Central Valley of Northern California, current instrumentation is not sufficient for ElarmS operation.


Support for this project is provided by the USGS NEHRP program (06HQAG0147).


Allen, R. M., and H. Kanamori, The potential for earthquake early warning in southern California, Science, 300, 786-789, 2003.

Wurman, G., R. M. Allen, and P. Lombard, Toward earthquake early warning in northern California, J. Geophys. Res., 112 (B08311), 2007.

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail:
© 2007, The Regents of the University of California