Identify a "real" event

Compared to a classical seismic system (where a spacing of 10 to 100 km between the seismic stations is typical), in a super-dense network the requirements for the identification and qualification of a potentially dangerous event are relaxed.

With sensors distributed more or less regularly in a grid of 1 x 1 km, we can assume that a few of these will be located close to the epicenter, and therefore will be the first to experience really strong shaking. In these lucky prime locations, a dangerous event can be identified quickly—perhaps just with a single seismogram. Hence, a relatively primitive trigger algorithm would be to use a peak acceleration threshold in the order of 3 to10% g (ptp) or a comparable peak velocity, which might be more diagnostic.

Although a simple threshold algorithm would do the job, a little more sophistication (artefact rejection) would certainly improve the system´s credibility, reliability, and flexibility. Still, this would be an easy job compared to the substantial computational effort involved in a more clasical set-up, where timing and phasing issues between neighboring stations have to be resolved by complex mathematics.

Of course, in all types of early warning systems special care must be taken to avoid false alarms. Any outgoing message has to be verified by suitable methods, such as finding synchronous triggers originating from neighboring stations. Additionally, warning messages should contain statements of reliability and likely event size, as well as the shaking centroid. These would be used by recipients to make more reasoned implementing choices. For more details see (LINK to be defined later).

back