CISN ShakeAlert: Next Generation ElarmS
Abstract
Earthquake Early Warning (EEW) is a method for recognizing and characterizing earthquakes in progress and sending immediate alerts to surrounding population centers, ideally seconds before damaging ground shaking begins. Magnitude, location, and origin time are estimated from P-wave arrivals at seismic stations close to the hypocenter. ElarmS is an EEW algorithm developed at the University of California, Berkeley. It is part of CISN ShakeAlert, a state-wide EEW system operating within the California Integrated Seismic Network (CISN). ElarmS consists of two primary modules: (1) continuous waveform processing of real time seismic data, which runs in parallel at UC Berkeley, Caltech, and USGS Menlo Park, and (2) a single state-wide event detection algorithm at UC Berkeley. The event detection module analyzes incoming data from the three waveform processing streams and identifies earthquakes in progress. In March 2011, ElarmS began sending event messages to the ShakeAlert Decision Module for events in the San Francisco Bay Area and Central Coast. At the same time, we have been developing ElarmS-2 or E2, updated waveform processing and event detection modules, based on C++ for speed and adaptability. New, more flexible communication software connects the remote waveform processing modules to the event detector. The new event detection module (E2) uses the existing location and magnitude relations, but has an updated method of associating triggers together to form events. Now E2 is the operational system and has been processing and publishing statewide real-time data since April 2012. Its new capabilities include a) A split event check to prevent double event alerts for a single event b) Linear teleseismic filtering to reduce triggers on teleseismic arrivals c) Use of 1-second data package from BK network to increase speed d) Replay capability for past earthquake events e) Improved magnitude determination for southern California f) Improved estimation of event location g) Specific algorithms for offshore events h) Dynamic integration of station data from throughout California We inspect and assess system performance on a weekly basis. The assessment software automatically evaluates station latencies, promptness of alerts, accuracy of magnitude, location, and ground-shaking estimates, and number of successful event detections, false alarms, and missed events. This information is evaluated and E2 is being optimized based on performance. In offline tests of performance using the latest version (E2, v2.2) applied to the data from a period of ~4 months (December 6, 2011 to April 20, 2012) we find that E2 detected 164 events of magnitude 2.5 or greater in California and sent 4 false alerts, 3 of which are due to a single teleseism. In this period 22 events are declared with magnitude greater than 3.5 with 2 false and only one missed earthquake. For the greater Bay Area, 11 events were detected with magnitude greater than 3.5 with no false and no missed events. On average, E2 declared events 15 seconds after the origin time with a 5 second standard deviation. Initial P-waves take an average of 3 seconds to reach the first station, meaning that the communications, processing and algorithms require an average of 11 seconds to declare an earthquake.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFM.S52B..02K
- Keywords:
-
- 7215 SEISMOLOGY / Earthquake source observations;
- 7223 SEISMOLOGY / Earthquake interaction;
- forecasting;
- and prediction;
- 4315 NATURAL HAZARDS / Monitoring;
- forecasting;
- prediction;
- 4341 NATURAL HAZARDS / Early warning systems