Fluid migrations and volcanic earthquakes from depolarized ambient noise

The study of volcanic systems has significantly advanced with the utilization of ambient seismic noise, generated by sources such as human activities, ocean swell, and atmosphere-solid Earth interactions. Analysis of noise data from expanding seismic networks, employing array and interferometric techniques, enables the detection of volcanic processes and hazard forecasting without relying solely on earthquake events. Ambient noise exhibits complex polarization, characterized by preferential oscillation directions and planes, particularly when higher modes of surface and body waves interact with fundamental modes. While noise polarization has been used to study oceanic processes and relate to stress and stiffness anisotropy across faults, the use of depolarization – the loss of polarization – to monitor deep fluid-induced dynamics in stressed volcanic structures has remained unexplored. This study aims to address this gap by investigating the potential of depolarized noise to monitor volcanic activity and associated earthquakes at Campi Flegrei caldera in Southern Italy. Campi Flegrei, located near Naples, is an ideal location due to its inhabited nature, its capped geothermal system where pressurized fluids propagate from a primary deformation source to fumaroles, and its history of unrest episodes resulting in stress accumulation and reservoir expansion. Previous research, including analogue modelling, seismic tomography, and geological fieldwork, has identified NW-SE trending extensional faults and a NE-SW trending transfer structure as crucial for volcanic unrest. This study proposes to utilize depolarized noise as a novel technique to monitor these structures and processes, enhancing real-time volcano monitoring capabilities.

Previous research on Campi Flegrei has utilized various geophysical methods to understand its complex system. Studies have focused on the 1984 unrest, identifying a low-velocity hydrothermal reservoir expansion from an injection point, reaching western caldera-bounding faults and causing seismic swarms. The eastern sector, however, showed limited expansion, seemingly hindered by a high-velocity, high-stress barrier. Various techniques, including InSAR interferometry, shear-wave-splitting anisotropy, gravity gradiometry, and strong seismic velocity contrasts, have been employed to map the structure and stress distribution within the caldera. These studies have provided valuable insights into the location and dynamics of fluid reservoirs and stress accumulation, but the temporal resolution and spatial detail have been limited by the methods employed. This paper builds upon this previous work, leveraging the unique information provided by ambient noise polarization to enhance the understanding of fluid migration and its relation to seismic activity.

This study utilizes ambient noise polarization attributes measured at Campi Flegrei using data recorded between 2009 and 2020. The data comprises recordings from broadband stations of the INGV, Sezione di Napoli-Osservatorio Vesuviano (INGV-OV) seismic network. Data from different periods (2009, 2017, 2018, 2019-2020) were processed to compare polarization patterns during periods of low and high seismic release. The azimuth of the horizontal polarization vector and its resultant length (R) were calculated using the covariance matrix method. Data with rectilinearity less than 0.5 were discarded. The analysis focused on horizontal ground motion polarization due to its sensitivity to medium properties. A threshold incidence angle of 45° was used to select azimuth values associated with high horizontal polarization. The stability of polarization patterns was assessed by comparing results from different time scales (years, months, days, hours) and across different stations, employing bootstrap tests to quantify the variability. Synthetic seismograms were generated through finite-difference simulations to model noise polarization under both isotropic homogeneous and heterogeneous conditions, allowing the researchers to isolate the effects of medium properties and source locations on polarization patterns. The simulations utilized Morlet wavelets, a viscoelastic constant-Q Zener model, and considered various source configurations (line source in the central Tyrrhenian basin and circular source offshore) to assess the influence of source location and medium heterogeneities on the observed polarization patterns.

The key findings of the study reveal a strong correlation between ambient noise polarization and the underlying geological and geophysical structures at Campi Flegrei. During periods of low seismic release, high polarization clearly defined both the NW-SE trending extensional faults and the NE-SW trending transfer structure. This transfer structure, previously hypothesized but not directly imaged, connects the central caldera's deforming center to degassing vents. Crucially, depolarization of the transfer structure coincided with fluid injections and migrations preceding periods of increased seismicity. The high resultant lengths observed in the eastern caldera aligned with the extensional faults, potentially explained by a high-velocity waveguide effect or seismic anisotropy. However, azimuths perpendicular to the transfer structure's primary direction were better explained by near-field sources, suggesting that fluid migration plays a significant role in shaping the polarization patterns. The analysis of data from the 2019-2020 seismic sequence demonstrated a clear link between depolarization of the transfer structure and fluid migration, with fluids moving towards the eastern sector (Solfatara and Pisciarelli vents) and Monte Nuovo. The temporal evolution of polarization patterns before and after the Md 3.1 and Md 3.3 earthquakes further reinforced this connection, with depolarization preceding the earthquakes and repolarization occurring after the stress release. The post-seismic migration of the eastern unpolarized anomaly confirmed the persistent lateral stress driving fluid movement toward the eastern caldera. The study also highlighted the potential of short-term (hours) monitoring of depolarized noise to detect fluid injections and the subsequent increase in compression preceding higher-magnitude earthquakes.

The results of this study significantly advance our understanding of the interplay between fluid migration, stress buildup, and seismicity in the Campi Flegrei caldera. The identification of the transfer structure and its behavior during periods of unrest provides valuable insights into the mechanisms driving volcanic activity. The direct correlation between depolarization and fluid migration offers a novel, real-time monitoring tool for assessing volcanic hazards. This method has the potential to improve early warning systems by providing near-instantaneous information about fluid movement and stress changes. The findings highlight the importance of both the caprock and the high lateral stress in enabling over-pressurization, fluid migration, and deformation. The ability to discriminate depolarization from processing uncertainties is crucial, and this study demonstrates that persistent high polarization in surrounding structures allows for reliable interpretation, particularly in volcanoes with longer repose periods. The success at Campi Flegrei suggests that this method could be adapted to other volcanic systems with similar characteristics. Future research should focus on validating this approach in other volcanic settings and integrating this new data with existing monitoring techniques to provide a more comprehensive understanding of volcanic processes.

This study demonstrates the effectiveness of using depolarized ambient noise to monitor fluid migrations and predict volcanic earthquakes at Campi Flegrei. The identification of a previously unimaged transfer structure and the observation of its depolarization preceding seismic events provide a novel approach to real-time volcano monitoring. The findings suggest that this technique could significantly improve early warning systems by offering near-instantaneous information about fluid movements and stress changes, particularly in volcanoes with extended periods of repose. Future work should focus on applying this method to other volcanoes to assess its wider applicability and on improving the automation of data processing to facilitate real-time monitoring.

While this study provides compelling evidence for the use of depolarized ambient noise in volcano monitoring, several limitations should be acknowledged. The study primarily focused on Campi Flegrei, which possesses a unique geological setting that may not be representative of all volcanic systems. The interpretation of depolarization relies on distinguishing it from processing uncertainties, which could be challenging in volcanoes with frequent seismic activity. The current method requires relatively high amounts of data and processing, although improvements are being made towards real-time applications. The detailed modeling of wave propagation and polarization requires advanced computational resources and expertise.