Multi-phase Fluid Circulation and Ground Deformation: a New Perspective on Bradyseismic Activity at the Phlegrean Fields (italy)

Like many caldera structures around the world, Phlegrean Fields caldera (Italy) periodically undergoes volcanic unrest, characterized by seismic activity and vertical ground displacement (bradyseism). These bradyseismic crises are usually interpreted as the product of pressure increment at the magma chamber level, but the existence of an important hydrothermal system in the area suggests that the migration of hot hydrothermal fluids should also play a role in the process. A complete description of the processes requires therefore a coupled analysis, accounting for both the flow of hot, multi-phase fluids and for the deformation of the porous medium. Here we applied the coupled thermo-hydro-mechanical model (TOUGH-FLAC) to study the role of hydrothermal circulation in ground deformation episodes at the Phlegrean Fields. Based on previous modeling results, we calculated the effects of an increased magmatic degassing on the deformation of a shallow elastic porous medium. Our results show that a short period of increased injection of deep fluid into the shallow hydrothermal system can represent a potential trigger for bradyseismic events, even in absence of a new magmatic recharge. Results also provided interesting insights on the complex interaction between fluid migration and ground deformation when phase changes are allowed. The comparison between these preliminary results and field data shows that the model captures both the temporal evolution of ground displacement and the compositional variations of fumarolic gas recently observed at the Phlegrean Fields. RECENT EVOLUTION AND MODELING EFFORTS AT PHLEGREAN FIELDS The Phlegrean Fields is a large and widely urbanized volcanic district in southern Italy (Fig. 1). In historical times, this caldera structure has undergone remarkable episodes of ground deformation, accompanied by seismic activity. Secular subsidence is periodically interrupted by short-lasting uplift phases (Parascandola, 1947; Dvorak e Berrino, 1991; Dvorak e Gasparini, 1991; Dvorak e Mastrolorenzo, 1991). A renewal of eruptive activity may follow the maximum uplift, as in the year 1538, when 7 m of vertical displacement preceded the Monte Nuovo eruption, the last recorded in the caldera (Dvorak e Gasparini, 1991). Fig. 1. The Phlegrean Fields caldera and its major collapsing structures: Campanian Ignimbrite (solid lines) and Neapolitan Yellow Tuff (dashed lines). Other times, however, unrest crises do not culminate with eruptive activity, and terminate with a slow aseismic subsidence, as recently happened, in 1969-72 and 1982-84. Both episodes were confined within the central portion of the caldera, and were characterized by ca. 2 m of vertical displacement at the caldera centre (Caputo, 1979; Barberi et al., 1984; Berrino et al., 1984; Corrado et al., 1984). Since 1985, three minor uplift episodes interrupted a general trend of slow subsidence (Fig. 2). The region is also affected by intense hydrothermal activity, with surface manifestations concentrated within the ancient crater of Solfatara. Recent measurements of diffuse carbon dioxide emission through the soil revealed the impressive magnitude of this phenomenon. The energy budget associated with fluid ascent and shallow condensation was shown to be higher than the energy involved in recent seismic activity and ground deformation (Chiodini et al., 2001). Fumaroles, with a maximum discharge temperature of 164°C, are also present at La Solfatara and have been sampled during the last 20 years. Fluid composition (H2O, with CO2, H2S, N2, H2, and CH4 as minor components) is typical of the hydrothermal environment. The important contribution of magmatic components is apparent from the isotopic composition of H2O, CO2, and He (Tedesco et al., 1990; Allard et al., 1991; Panichi and Volpi, 1999; Tedesco and Scarsi, 1999). Fumarolic gases underwent remarkable chemical changes during and after recent bradyseismic events (Barberi et al., 1984; Cioni et al., 1984; Martini, 1986; Martini et al., 1991; Tedesco and Scarsi, 1999). These variations were traditionally interpreted in terms of variable degree of boiling that would affect the hydrothermal system when heated by a new magma supply reaching the magma chamber. Fig. 2. Vertical ground displacement measured in Pozzuoli (Benchmark 25) since 1969. The recent unrest phenomena have prompted new research efforts to unravel the mechanism driving bradyseismic crises. Several models have been proposed to explain the observed ground deformation, commonly in terms of pressure build-up inside a magma chamber (Caputo, 1976; Casertano et al., 1976; Corrado et al., 1976; Berrino et al., 1984; Bonafede et al., 1986; Bianchi et al., 1987). Most of these models describe the effect of a pressure source embedded in a homogeneous half space, with elastic or viscoelastic properties. Further models introduced the effects of structural discontinuities in controlling both the magnitude and the areal extent of ground deformation (De Natale and Pingue, 1993; De Natale et al., 1997; Orsi et al., 1999). The importance of fluids in bradyseismic crises was first recognized by Olivieri del Castillo and Quagliariello (1969), and later on by Casertano et al. (1976). More recently, other authors have recognized that purely mechanical models cannot fully explain the observed ground deformation and began to address the effects that heating and expansion of hydrothermal fluids can produce during a bradyseismic event (Bonafede, 1991; De Natale et al., 1991; 2001; Gaeta et al., 1998; Castagnolo et al., 2001). These models provided important insights but they are based on simplified descriptions of the fluid dynamics, usually accounting for a steady, single phase fluid of constant properties. On the other hand, a more realistic modeling of twophase, two component hydrothermal fluid circulation at La Solfatara was performed recently (Todesco et al. in press; Chiodini et al., subm.), with the TOUGH2 geothermal simulator (Pruess, 1991). These works could capture most of the main features characterizing the hydrothermal system at La Solfatara (Todesco et al., in press) and its recent compositional variations (Chiodini et al., subm.). However, they did not account for the deformation of the solid porous matrix. The present work represents a natural evolution of the previous modeling, and it is aimed at coupling the fluid-dynamics of hydrothermal circulation to a mechanical analysis of the porous media. The simulations presented here are based on the results described in Todesco et al. (in press), that we shall briefly recall hereafter. The work was focused on the shallower portion of the hydrothermal system of La Solfatara, from the surface to a depth of 1500 m. This shallow system was heated by a continued inflow of a hot mixture of water and carbon dioxide, representing the product of deep magmatic degassing. Upon prolonged injection, a large plume of hot fluids develops and a single-phase gas region forms at shallow depths, as predicted by the independent geochemical model (Chiodini and Martini, 1998). The composition and physical condition of this single-phase gas region correspond to those inferred from geochemical data for the source region feeding the fumarolic activity (Todesco et al., in press). The same model was then applied to simulate subsequent unrest crises at La Solfatara (Chiodini et al., subm), by means of discrete periods during which the injection rate of the deep fluid at the source was increased. The model successfully reproduced the observed chemical variation, all preceded by an increase of pore pressure and temperature. MODELING OF FLUID INJECTION AND GROUND DEFORMATION Further insights into the phenomena driving bradyseismic events may be achieved with a more complex approach, in which both the dynamics of multi-component hydrothermal circulation and the mechanical aspects are considered. The coupled TOUGH-FLAC model applied here links the capabilities of the TOUGH2 geothermal simulator (Pruess, 1991) with the geotechnical analysis of rock and soil performed with FLAC3D, a commercial code for rock mechanics (Itasca Consulting Group Inc., 1997). Both codes are fully described elsewhere, as is their coupling (Rutqvist et al., 2002). Briefly, TOUGH2 describes the coupled transport of heat and multi-phase, multi-component fluids through a porous matrix, accounting for phase changes and associated 0 0.5 1 1.5 2 2.5 3 3.5 1/69 12/73 12/78 12/83 12/88 12/93 12/98 Date (month/year) G ro un d up lif t (m ) latent heat effects. The fluid components considered here are water and carbon dioxide. FLAC3D is an explicit finite difference program applied here to describe the coupled thermomechanical behavior of a continuous elastic medium, as it reaches equilibrium. The coupled TOUGH-FLAC model was applied to study the mechanical effects potentially induced by an intensified magma degassing. Starting from the present system conditions, calculated following Todesco et al. (in press), bradyseismic crises are here simulated by temporarily increasing the injection rate at the fluid source. Composition and temperature of the injected fluid mixture were unchanged. The coupled model allowed us to first calculate the effects of increased injection rate on overall fluid circulation and phase distribution, and then to estimate the corresponding rock deformation, arising from temperature and pore pressure changes. At this stage, we did not take advantage of the full (two-way) THM coupling allowed by the code. In other words, we did not calculate the effects that changes in stress distribution would have on rock porosity and permeability. This decision was based on methodological considerations, suggesting to address a simplified problem before dealing with more complex interactions, but was also due to the lack of necessary information on the relationship linking stress distribution, and rock porosity and permeability for the rock of the Phlegrean Fields. To allow for the coupling with FLAC3D, all simulations were run in a three-dimensional domain, whose geometry and dimensions were chosen to match previous models. We here consider the inner portion of the Phlegrean Fields caldera, bounded by the Neapolitan Yellow Tuff collapsing structures (Fig.1). Thanks to the problem’s symmetry, the computational domain represents 1/4th of the caldera (Fig. 3). As in Todesco et al. (subm), only the shallower portion of the hydrothermal system was considered. This was necessary to keep the simulation within the modeling assumptions and, as a consequence, thermal effects associated with the deeper portion of the system are neglected. At this stage, we therefore expect to reproduce only a fraction of the observed deformation. In all simulations we performed, the fluid is a mixture of water and carbon dioxide and rock physical properties are considered to be homogeneous and constant, and are reported in Table 1. Two values for the rock elastic properties were chosen based on literature data. NUMERICAL SIMULATION Initial conditions were calculated with TOUGH2, as in Todesco et al. (in press), simulating a prolonged (4000 years) injection of water and carbon dioxide from a source area (150x150 m) at the base of the domain (Fig. 4). The composition of the fluid mixture (CO2/H2O=0.2 vol%) was taken to be representative of recent fumarolic emissions. Table 1. Physical and elastic properties of the porous