Thermo‐Hydro‐Mechanical Model and Caprock Deformation Explain the Onset of an Ongoing Seismo‐Volcanic Unrest

Waheed Gbenga Akande,David G Cornwell,Luca De Siena,Quan Gan

doi:10.1029/2020jb020449

Abstract

AbstractModeling seismicity at volcanoes remains challenging as the processes that control seismic energy release due to fluid transport, heat flow, and rock deformation are firmly coupled in complex geological media. Here, we couple fluid‐flow and mechanical (deformation) simulators (TOUGHREACT–FLAC3D) to reproduce fluid‐induced seismicity at Campi Flegrei caldera (southern Italy) in isothermal (HM) and nonisothermal (THM) conditions. The unique ability of the Campi Flegrei caprock to withstand stress induced by hot‐water injections is included in the model parametrization. After pore pressure accumulation is guided by a combination of thermal and hydromechanical interactions, contrasting compressive and extensional forces act on the basal and top parts of the caprock, respectively. Then, pressure perturbation and caprock deformation induce fractures that allow hot fluids uprising to pressurize the overlying fault, driving it toward failure and triggering seismicity. Under similar mechanical boundary conditions, the induced thermal effects prompt seismic slip earlier but with higher seismic magnitudes when (1) thermal equilibrium is preserved and (2) the thermal contrast is enhanced due to increased fluid injection temperatures. The results indicate that numerical models of volcano seismicity must consider the influence of rock‐sealing formations to obtain more robust, accurate, and realistic seismic predictions at volcanoes. The proposed models satisfactorily reproduce the magnitude–depth distribution of the swarm (October 5, 2019), preceding the two strongest earthquakes recorded in 35 years at the caldera (3.1 and 3.3—on December 6, 2019, and April 26, 2020, respectively) using hot‐water injection from depth.

Highlights

Coupled thermo-hydro-mechanical (THM) modeling has been widely applied in Enhanced Geothermal System sites (e.g., Gan & Lei, 2020; Rinaldi, Rutqvist, et al, 2015; Rutqvist, Dobson, et al, 2013; Vasco et al, 2013; Ziagos et al, 2013; see reviews by Lu, 2018, and references therein) and becoming a reliable tool to quantify the magnitude of injection-induced seismicity in volcanoes
Plain Language Summary Can we model seismicity in a volcanic caldera using tools and methods taken from deformation and fluid-flow modeling in reservoirs? If the target is Campi Flegrei the application must certainly include parameter such as the recently characterized “caprock,” which partially mitigates the stress induced by dyke intrusions and fluid injections during deformation unrests
Like in many active volcanoes worldwide, it is agreed that seismo-volcanic unrests at Campi Flegrei caldera (CFc), southern Italy, are associated with near coeval phenomenal ground deformation, subsidence, and seismicity

Summary

Introduction

Coupled thermo-hydro-mechanical (THM) modeling has been widely applied in Enhanced Geothermal System sites (e.g., Gan & Lei, 2020; Rinaldi, Rutqvist, et al, 2015; Rutqvist, Dobson, et al, 2013; Vasco et al, 2013; Ziagos et al, 2013; see reviews by Lu, 2018, and references therein) and becoming a reliable tool to quantify the magnitude of injection-induced seismicity in volcanoes. Fluid injections drive subsurface processes that manifest as seismicity and ground deformation (Cornet et al, 1997; Ellsworth, 2013; Majer et al, 2007; Nicholson & Wesson, 1990, 1992; Raleigh et al, 1976) and produce geochemical anomalies (Caliro et al, 2007; Chiodini, Caliro, De Martino, et al, 2012) in different environments. Measured seismicity can be the result of injections into shallow fluid reservoirs that perturb the pore pressure, ambient stress regime and enhance hydraulic properties, such as the permeability of the formation (reservoir), thereby reducing the mechanical strength of rocks and triggering fault slips, especially along planes of weakness of optimally oriented faults (Cappa & Rutqvist, 2012; Cornet, 2016; Cornet et al, 1997; Gan & Elsworth, 2014a; Guglielmi et al, 2015; Nicholson & Wesson, 1990; Zoback & Gorelick, 2012)

Results

Discussion

Conclusion