Seconds Matter

Earthquake Early Warning (EEW) can mitigate losses from earthquakes by providing seconds of advance notice to recipients so they can prepare for imminent damaging ground shaking. EEW is in operation in many countries across the globe, but currently no country in Europe offers a public system despite significant seismic risk. EEW is no longer technologically out-of-reach or exorbitantly expensive, especially if a modern and dense seismic network already exists, which is true in many parts of Europe. Recently, the Swiss Seismological Service (SED) at ETH Zurich has rolled out public EEW systems across Central America using their own EEW algorithms and mobile phone applications. In this proposal, we will replicate and extend this system to key countries in Europe where earthquake hazards range from moderate (Switzerland) to high (Romania and Croatia), and the earthquake risk is high.

In recent years, new approaches have upended seismology - machine learning (ML) is improving automated data processing, and the interrogation of fibre optic cables, or distributed acoustic sensing (DAS), is enabling a transformative sensing technology with unprecedented spatial and temporal resolution. However, it is still unclear if and how these technologies can improve EEW systems.

Our project will, as a baseline, establish and operate end-to-end semi-public EEW in Romania, Croatia and Switzerland using existing algorithms. Our consortium includes the SED at ETH, National Institute for R&D for Earth Physics (INFP) in Romania, and the University of Zagreb (UniZg) in Croatia, each responsible for national seismic monitoring. During these 4 years, we intend to continually improve the existing EEW approaches. This includes scientific tasks such as adaptation for complex regional seismicity such as deep Vrancea events or shallow crustal earthquakes in the Dinarides; improvement in ground motion estimation using synthetic waveforms; integration of ML algorithms for picking and shaking prediction; and integration of DAS data alongside seismic stations. Additionally, we will also focus on social science studies that engage user communities to understand how to best communicate and deliver EEW in the different countries.

In parallel, efforts will be made to future-proof EEW systems in the three countries by leveraging the disruptive technologies of DAS and ML for EEW. A new EEW approach will be developed that by-passes traditional detection and source parameter estimation by focusing directly on ground motion prediction. We will leverage the new ML-generated synthetic seismograms to ensure the method is optimised for rare extreme events and complex sequences that are missing in the instrumental catalogues.