GNGTS 2023 - Atti del 41° Convegno Nazionale

Session 3.3 ______ ___ GNGTS 2023 probably related to shallow rock damage, and which is followed by a stable decrease (deeper crack closure). Zooming in on short intervals (0.5-10 days) immediately after mainshocks (i.e., Amatrice, Visso-Norcia, Capitignano), we see a consistent pattern. Each mainshock that initiates a sequence is associated with a sharp increase in followed by a comparatively steep drop. We −1 ,( ) observe coincident distance (d) and time (t) behavior of M ≥ 3.5 earthquakes that is consistent with fluid diffusion, where . A subsequent gradual recovery of persists up until the next ∝ 0.5 −1 ,( ) mainshock (Fig 2). We hypothesize that this recovery is associated with the redistribution of fluids into newly damaged faults and into the shallow crust where bubble production induced by traveling stress-waves may also cause significant attenuation (Tisato et al., 2015). An interesting feature of the quantity: is that its variations are roughly anticorrelated with the −1 , = 2 ( ) instantaneous moment release during the seismic sequence. Our hypothesis is that in the Central Apennines the coseismic permeability changes create fluid diffusion pathways that are at least partly responsible for triggering multiple mainshocks during the same sequence: a behavior that is very common in this area (examples of diffusion-driven seismic sequences in the Central-Northern Apennines is the Umbria-Marche one, Miller et al., 2004, and the 2009 L’Aquila, Malagnini et al., 2012). For extensional crustal environments with compartments of pressurized fluids, the described behavior may be true in general. Ridgecrest The Ridgecrest area seems to be affected by anomalies of lesser amplitudes when compared to the ones observed in the Central Apennines, even though the sequence was characterized by a mainshock substantially larger than the largest Italian event (M7.1 vs. M6.33). In agreement with what found by Lu and BenZion (2022), who observed a quick recovery of the mainshock-induced velocity anomalies in the area, we also observe a quick recovery of the attenuation parameter (Fig 3). In addition, and differently from the Central Apennines, in the Ridgecrest area we observe significant changes in attenuation from local and remote earthquakes. The waveforms analyzed here from the Ridgecrest seismic sequence are from stations to a maximum distance of 30 km from the epicentral location of the mainshock, and the largest allowed hypocentral distance was 80 km (Fig 3). Note that the investigated area at Ridgecrest is relatively compact. We tried to separate data collected in the two sides of the Ridgecrest fault: the Sierra Nevada and the Pacific sides, but our results did not show significant differences between the two, at the expenses of a severe depletion of the number of waveforms that were available for the regressions. We also tried to use only earthquakes and stations very close to the fault trace (within 20 km), but the reduced dataset did not seem to have much resolving capabilities with respect to time-domain attenuation anomalies.