Key points are not available for this paper at this time.
The 2016 Central Italy seismic sequence occurred within an area dominated by normal-fault systems present along the Apennines. The sequence began with the Mw 6.0 Amatrice event on the 24 August 2016, followed by the Mw 5.9 Visso event on the 26 October 2016 and then, four days later, the Mw 6.5 Norcia event. In this study, we aim at modeling the seismicity of this complex earthquake sequence in order to determine the location of highly-pressurized fluids under the studied area through swarms occurring during the sequence. To do so, we take advantage of a high-resolution earthquake catalog based on arrival times derived using a deep-neural-network-based picker. As a first step, we apply a density-based clustering approach to group earthquakes into dense clusters. The majority of the resulting clusters highlight distinct fault planes which indicates an activation of a complex fault network. We further define a 4-dimensional seismicity model based on the Epidemic-Type-Aftershock- Sequence (ETAS) model, in which we introduced an earthquake detection probability to accommodate observed rapid fluctuations in earthquake detection throughout the sequence. By computing the ratio between the observed and ETAS-modeled rates of high-density clusters, we can identify candidate seismic swarms. To evaluate their consistency, we compute the weighted index of the largest seismic event and the magnitude difference between the largest and the 4th-largest earthquakes occurring within a target candidate. Furthermore, to analyze their migration behavior, eigenvectors are computed to identify primary and secondary directions, and swarm earthquakes are projected onto these directions, in which we fit a linear regression model. Observed slope values are compared to those from a distribution of simulated slopes generated by 1000 random shuffling, and statistical significance is assessed. The results reveal among the 40 seismic swarms, 29 of them show significant migration with a significance level of 95%. Migration velocity is determined by components representing velocities along primary and secondary directions, measured in kilometers per day based on slope and earthquake count. Our swarms exhibiting significant migration suggest the implications of variations in pore fluid pressure influenced by structural complexity and intense faulting in the region. Before Norcia mainshock, we observe that swarms are detected in shallow depth with z = 0;5 km, whereas after the mainshock, swarms are found deeper with z > 8 km where the detachment plane is located.
Xiang et al. (Mon,) studied this question.