The Preparation Phase of the 2009 L'Aquila and 2016 Norcia Earthquake: Insights from the Covariance Matrix Analysis of Continuous Seismic Data

Abstract:

Recent studies using laboratory seismic data identify clear precursors preceding rock failure and show the predictability of labquakes. Contextualizing the laboratory results into real Earth suggests the existence of signals beyond earthquakes in continuous seismic data, which are rich with information about fault physics and may not be easily summarized into a seismic catalog. We explore continuous array seismic data using covariance matrix analysis and extract coherent signals to investigate the preparation phase of the 2009 L'Aquila earthquake and the 2016 Norcia earthquake. These two normal fault earthquakes exhibit different preparation patterns: the L'Aquila earthquake was preceded by a significant increase of foreshock activity, while the Norcia earthquake shows no obvious increase of seismicity rate before the mainshock. We calculate the wavefield coherence of continuous seismic data as seen by an array of stations. By defining a coherence threshold, signals which are coherent across the whole array deployed in the vicinity of fault systems can be retrieved for studying earthquake preparation processes. In addition to signals in the existing earthquake catalog, we also detect many new coherent signals from the continuous data. The location of sources generating these signals are revealed by analyzing covariance matrix eigenvectors. Clustering analysis based on spatial and spectra information classifies these signals and the related sources into different groups, e.g. anthropogenic signals and fault-related signals. The newly discovered signals serve as a supplement to the earthquake catalog, and together provide insights into earthquake preparation studies.