Project description

A persistent goal in geoscience is to improve protection of metropolitan regions against seismic risk. Earthquake early-warning systems activate only once an earthquake has started, and within the short time following an alert, only limited actions can be taken. For active faults near urban areas, there may be only few seconds between the start of the earthquake and the arrival of the damaging S- and surface waves. A longstanding and lingering question is whether hazardous earthquakes show any kind of characteristic nucleation process that could be identified in ample time before their onset. If such processes could be reliably observed prior to large earthquakes, then automated near-real-time detection and processing of the geophysical signals (e.g. seismic waveforms or geodetic transients) could provide extended warning and preparation time. This is a key ingredient for activating civil protection protocols to mitigate and eventually to reduce risk. The key unanswered question in this context is: Under what conditions do earthquakes display detectable nucleation processes, what is the time extent of these processes and how can they complement physics-based earthquake forecasting? Within QUAKEHUNTER, we address this question by exploring new methodological approaches based on artificial intelligence to systematically monitor nucleation processes of moderate and large earthquakes at various time scales. The ultimate goal is to test its performance in near-real time for an active fault near a mega-urban area in broader Europe.

Project duration

2023 - 2028

Funding agencies

European Research Council (ERC Starting Grant)