← Back to Sessions

Combining Earthquake Early Warning Solutions From Different Algorithms: Application to Switzerland

Description: 

It is increasingly common for seismic networks to operate multiple independent automatic algorithms to characterise earthquakes in real-time. In particular, multiple algorithms are an advantage when operating an earthquake early warning (EEW) system, allowing for redundancy and increasing tolerance to algorithm errors. Today, ‘decision modules’ that select the preferred solutions are rather simple and use ad hoc rules - an efficient real-time method is lacking. The Swiss Seismological Service (SED) currently tests the Virtual Seismologist (VS) and the Finite-Fault Rupture Detector (FinDer) (Massin et al., 2022). We have explored how to combine ongoing solution updates from these algorithms, inspired by the approach outlined in Minson et al. (2017). First, we use the event characterisation (origin time, hypocenter and magnitude) to predict, at each station in our network, the ground motion envelopes that match the duration of data already observed. We provide an absolute measure of how well the event origin matches the observations by the goodness-of-fit between the observed and predicted envelopes. This method provides a measure which can be used to determine when a preferred solution reaches an appropriate confidence level to send an alert, or indeed to compare two (or more) different event characterisations directly. We demonstrate that this approach can also be used to suppress false alarms commonly seen at seismic networks, improving over using traditional quality criteria based on e.g. location pick RMS or azimuthal gap. Initial tests using the 10 largest earthquakes in Switzerland between 2013 and 2020 and events that caused false alarms (quarry blasts, teleseismic earthquakes, etc.) demonstrate that this approach can effectively identify event characterisations with small errors in location and magnitude, and can clearly identify and discard false origins or incorrect magnitudes. The next step will be to implement the algorithm in a real-time workflow.

Session: Earthquake Early Warning Optimization and Efficacy

Type: Oral

Date: 4/20/2023

Presentation Time: 05:30 PM (local time)

Presenting Author: Dario Jozinović

Student Presenter: No

Invited Presentation: 

Authors