From dense monitoring seismic infrastructures to DAS: bridging detection techniques over different monitoring scales

Conventional catalogs are limited in size, since many small events are hidden in the noise. Therefore, discovering such events requires advances in both earthquake detection techniques and monitoring infrastructures. Dense permanent multidisciplinary observatories have been deploying near active seismogenic areas (Near-Fault Observatories, NFO) collecting seismological, geodetic and geochemical data with the goal of understanding the physical processes governing earthquake rupture. Moreover, temporary integration of dense seismic arrays has been proposed to further decrease the detection threshold. Finally, the progressive adoption of fiber optic systems for earthquake monitoring offers a novel decametric resolution, providing a significantly larger number of observations for earthquake characterization. With the growth of data quality and quantity, advanced machine learning models and similarity-based approaches have been developed to systematically identify low-magnitude earthquakes, reconciling the needing of efficient and reliable strategies.

Here, we apply innovative strategies for generating enhanced microseismic catalogs within the Irpinia Near-Fault Observatory (Southern Italy), which monitors the area struck by the 1980 M6.9 Irpinia earthquake, collecting high-resolution seismic observations from the kilometric-scale of the permanent seismic network (ISNet) to the decametric resolution offered by two DAS systems.

We showcase how the integration of machine learning and similarity-based detection techniques can increase the content of seismic catalogs both for background seismicity and seismic sequences. Characterization of the seismic source reveals the activated fault patches, while constraining evolutive models for the seismic sequences. We confirmed the effectiveness of the integrated detection strategy with the integration of 200 stations in dense arrays for one year. We demonstrated the possibility to consistently detect small magnitude earthquakes with the use of dense arrays lowering the magnitude of completeness of seismic catalogs down to M 0. The new catalog enables to downscale the seismicity characteristics to small, decametric-size events, illuminating active seismogenic structures capable to generate events up to M 7. 

To exploit the novel resolution offered by DAS systems, we exported machine learning models for the identification of P and S waves on native DAS records, showing that existing models can effectively recognize phase arrival times on DAS records, with the integration of 2D cross-correlation techniques identifying lower magnitude earthquakes.  We integrated machine learning models for an automatic characterization of the earthquakes recorded by DAS operating in the Irpinia region, tackling earthquake detection, phase association and local magnitude estimation.