Exploring Earthquake Depth variability through Between‑Event Residuals: Insights from Italy

Abstract

Despite the crucial importance of accurate event localizations, focal depths are generally poorly constrained, especially for very shallow events (depth <10 km), which are often mislocated. In this study, we investigate whether the event‑specific ground‑motion residuals, 𝛿𝐵𝑒, are correlated with focal depth. In particular, we examine whether and to what extent the focal depth influences
the spectral shape of the 𝛿𝐵𝑒 curves.
The residual investigated in this study are derived from the ITACAext 1.0 flatfile, which is related to the ITalian ACcelerometric Archive (i.e., ITACA database) and collects waveforms of seismic events with magnitude greater than 3.0, recorded in Italy since 1972. The residual analysis is conducted using the latest predictive ground-motion model for shallow crustal earthquakes in Italy as a reference. To emphasize the 𝛿𝐵𝑒 differences for events with varying focal depths, the attenuation in the groundmotion
model is measured using the Joyner-Boore distance.
A cluster analysis is applied on the 𝛿𝐵𝑒 spectral curves, revealing systematic trends that can be related to different focal depths. In particular, a cluster of very shallow earthquakes (<10 km; VSE) is identified. This cluster is characterized by 𝛿𝐵𝑒 amplitudes that are systematically positive at long periods (>1 s) and negative at short periods (<1 s). Opposite features are observed for deeper earthquakes (>10 km), with 𝛿𝐵𝑒 amplitudes that are systematically negative at long periods (>1 s) and positive at short periods (<1 s). As expected, the overall 𝛿𝐵𝑒 amplitudes of deep events decrease as depth increases.
The proposed method appears to be a promising tool for identifying mislocated events or detecting earthquakes erroneously classified as shallow crustal seismicity (i.e., very shallow earthquakes of volcanic origin or deep earthquakes in subduction areas). Improving the estimates of focal depths would lead to a reduction in the variability of the ground motion models in epicentral area, with significant implications on hazard estimates.