Faults strengthening and seismicity induced by geothermal exploitation on a spreading volcano, Mt. Amiata, Italia

Journal of Volcanology and Geothermal Research 301 (2015) 159–168

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research journal homepage: www.elsevier.com/locate/jvolgeores

Faults strengthening and seismicity induced by geothermal exploitation on a spreading volcano, Mt. Amiata, Italia Alberto Mazzoldi a,c,⁎, Andrea Borgia c, Maurizio Ripepe b, Emanuele Marchetti b, Giacomo Ulivieri b, Massimo della Schiava b, Carmine Allocca b a b c

UMSNH-IIM, Instituto de Investigaciones en Ciencias de la Tierra, Edif. “U” Ciudad Universitaria, CP 58060 Morelia, Mich., Mexico Dipartimento di Scienze della Terra, Università degli Studi di Firenze, via La Pira 4, 50121 Firenze, Italy EDRA, via di Fioranello 31, 00134 Roma, Italy

a r t i c l e

i n f o

Article history: Received 30 December 2014 Accepted 20 May 2015 Available online 29 May 2015 Keywords: Induced seismicity Geothermal exploitation Critically-stressed faults Fault strengthening Tectonic vs hydro-fracturing events Amiata volcano

a b s t r a c t Seismogenic structures such as faults play a primary role in geothermal system generation, recharge and output. They are also the most susceptible to release seismic energy over fluid injection/extraction operations during anthropic exploitation. We describe the microseismic activity recorded in 2000–2001 in the Piancastagnaio geothermal field, on the SE flank of Mt. Amiata volcano, southern Tuscany, Italy. From our field observations we find that a relatively high percentage (i.e. about 5%) of the recorded events are of hydro-fracturing origin and have a distinct waveform seismic signature when compared to the recorded events of tectonic shear-fracturing origin. While hydrofracturing events are mostly concentrated around the geothermal fields, the spatial distribution of hypocenters shows a deepening and a density increase of the micro-seismic activity from the volcanic axis toward the exploited geothermal reservoir, suggesting that volcanic spreading at Amiata is still active. The study of different data-sets from different time periods together with the knowledge from Terzaghi's law that production of large quantity of pore-fluid with the associated fluid pressure reduction could augment the stress normal to faults' surfaces (and thus their resistance to slip), make us argue that the process of volcanic spreading affecting the edifice of Amiata may allow augmented accumulation of stresses on faults, eventually leading to the release of higher stress drops, once ruptures occur. The Gutenberg–Richter magnitude–frequency distribution shows that the strongest events on record have a local magnitude in the 5–5.5 ML range, for 100-year recurrence time. In conclusions, we infer that geothermal exploitation at Mt. Amiata should be closely monitored in order to understand how fluid injection/production is responsible for the hydrofracturing seismic activity and affects stress accumulation on and rupture of faults within and in the neighborhood of the geothermal fields. This understanding may allow a geothermal field management that will hopefully reduce the risk for inducing larger seismic events in the area. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Due do public concern, induced seismicity is becoming an issue in almost every subsurface technology: from hydrocarbon exploitation (Dost and Haak, 2007; Suckale, 2010) and hydraulic fracturing (Davies et al., 2013) to Carbon Capture and Storage (Rutqvist, 2012; Zoback and Gorelick, 2012), through Enhanced Geothermal Systems (Majer et al., 2007; Sharma et al., 2013), waste-water management (Baisch and Harjes, 2003; Ake et al., 2005), reservoir impoundment (Stabile et al., 2014), nuclear underground explosions (Engdahl, 1972) and mining (McGarr, 1976). Seismogenetic structures, such as faults, can be found at any depth in the brittle crust and, with different sizes and frequencies, in every ⁎ Corresponding author. E-mail address: [email protected] (A. Mazzoldi).

http://dx.doi.org/10.1016/j.jvolgeores.2015.05.015 0377-0273/© 2015 Elsevier B.V. All rights reserved.

tectonic environment (Morris et al., 2010; Torabi and Berg, 2011). In the upper crust, at depths accessible to human activity, faults can behave both as conduits for fluid flow, allowing leakage from reservoirs (Sibson, 1990; Mazzoldi et al., 2012), or as flow barriers (Manzocchi et al., 2010; Yielding et al., 2010). In this second case, sealing properties of faults are well known among petroleum geologists (Yielding et al., 1997; Finkbeiner et al., 2001). Barton et al. (1995) found that, in crystalline rocks, faults that are optimally oriented for shear failure and are critically stressed tend to have increased permeability and to conduct fluid along their planes; non-critically stressed faults, in contrast, appear to provide no fluid migration pathways. In general, anthropic fluid injection in the neighborhood of a fault (but also the natural movement of fluids underground) can create conditions for pore-pressure build up on the fault gauge, which opposes the normal stress on the fault plane and induces fault-slip (with associated triggered or induced seismicity, Mazzoldi et al., 2012; Styles et al., 2014)

160

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

– Other similar happenings, with events of ML N 4, were recorded at Cooper Basin (Australia, November 2003), at Berlin (El Salvador, January/February 2001) and at The Geysers (California, various occurrences over the last three decades) geothermal fields (Majer et al., 2007).

Nomenclature C p w L M0 Mw M σ1,2,3 σn τ Δσ

rock cohesion (MPa) pore pressure (MPa) fault' width (along depth) (m) fault' length (trace) (m) seismic moment (Nm) moment magnitude (−) event' magnitude (−) higher, medium, lower principal stresses (MPa) stress normal to a fault, plane (MPa) shear stress, parallel to a fault' plane (MPa) coseismic stress drop (MPa)

when the shear stress exceeds fault-strength (e.g. Barton et al., 1995). Numerically, this can be expressed by Terzaghi's law: τ ¼ C−μðσn −pÞ

ð1Þ

where τ is shear stress, C is cohesion, μ is coefficient of friction, σn is the stress normal to the fault plane, and p is fluid pressure. Given that fault slip can be caused by pore-pressure enhancement near a fault plane, it seams at odds how induced seismicity has also been reported from production operations that effectively decrease fluid pressure on faults (Seagall, 1989; Dost and Haak, 2007; Majer et al., 2012). According to Eq. (1) in fact, lowering of pore pressure should enhance fault' strength and lock fault segments. However, due to the preferential compaction of the reservoir during production, the poroelastic behavior of the rocks can generate additional stresses that accumulate on the locked portions of faults, eventually exceeding their augmented strength and inducing fault slip and seismicity (Suckale, 2010). Stresses can accumulate on faults' locked segments also due to the continued aseismic slip of nearby portions of the faults — or to local tectonic deformation such as volcanic spreading (Borgia et al., 2000). Seismic events induced by geothermal operations have been studied for more than 30 European cases (Evans et al., 2012; Yoon et al., 2013), generally confirming the view that injection in sedimentary rocks tends to be less seismogenic than in crystalline rocks. Thousands of small events, not felt by people, are generated annually, but few earthquakes have magnitudes up to the 4–5 ML range. The presence of a fault in near critically-stressed conditions in the neighborhood of a well is unavoidable for the generation of a sized event. Famous examples of induced seismicity include: – The Rocky Mountains Arsenal project near Denver, CO, USA, where in 1967 the underground disposal of fluids induced the occurrence, several kilometers from the injection well, of an M 5.3 event, more than a year after injection ceased (Ellsworth, 2013). – At Paradox Valley, Colorado, USA, injection is currently active for the bundling of high salinity water from the Dolores river (Ake et al., 2005). The related induced seismicity (with several events of ML N 4) separates into two distinct zones: a principal one (N 95% of events) asymmetrically surrounding the injection well and to a maximum radial distance of ~3 km, and a second zone covering an area of about 10 km2 and centered ~ 8 km northwest of the injection. Highly permeable aseismic faults and fractures at the edges of the valley allow the fluid to reach the secondary seismic zone. – The Oklahoma sequence (November 2011), including events of up to ML 5.7, occurring in the relatively stable continental craton of central USA, which is linked to the long-lasting fluid injection during the past 20 years in the neighborhood of the once-thought non-active fault system (Sumy et al., 2014). – The ML 3.4 event that concurred to the shutdown of the Basel HotDry-Rock geothermal Project, Switzerland, in December 2009 (Dannwolf and Ulmer, 2009).

These examples testify of the importance of local/regional stress fields in the induction of seismicity from underground injections and on the role that fractures and faults play in allowing fluids motion (fracture/fault permeability), and thus stress changes, within the crust. We describe the seismic activity that affected the Piancastagnaio geothermal field on the SE flank of Mount Amiata volcano, southern Tuscany, Italy, in 2000–2001; of which, about 600 seismic events of magnitude −1 b ML b 3 were recorded by a four three-component seismic station arrangement deployed around the epicentral area of the ML 3.9 earthquake, that occurred on 1 April 2000 (Mucciarelli et al., 2001; Fig. 1). We anticipate from the conclusions that geothermal exploitation at Amiata may have augmented faults' segments strength within and around the geothermal fields, while, due to volcanic spreading and regional tectonics, stresses are continuously accumulated on faults and are eventually released through coseismic strain (Borgia et al., 2014). We describe ‘tectonic-fracturing’ events occurring mainly from within the volcanic edifice to below the exploited geothermal field; in addition we identify several ‘hydro-fracturing’ events that correlate with geothermal fluids production/injection from the reservoir. 2. Geologic structure of Amiata volcano Detailed descriptions of the regional geology of the Amiata volcano area may be found in Brogi (2008), in the 1:10,000 Geologic Map of Amiata volcano by Regione Toscana, RT (2006–2009) and in Ferrari et al. (1996), Cadoux and Pinti (2009) and Borgia et al. (2014). In synthesis, in the Amiata volcano area (Table 1; Fig. 2a), during the Apennine convergent orogenic tectonics from Cretaceous to Oligocene, the Ligurian Units were stacked older-on-top-of-younger above the Sub-Ligurian Units; during Oligocene–Early Miocene both of them were subsequently stacked on top of the Tuscan Units. Post-collisional extensional tectonics in the area, from Middle Miocene to Present, produced the collapse and thinning of the lithosphere and of the overthickened crust, with associated renewed sedimentation, emplacement of plutons, uplift, and finally volcanic activity. Borgia et al. (2014) show that, due to gravity and intrusive doming, the volcanic edifice becomes normally faulted while sinking and spreading on its ductile clayey (Ligurian) and evaporitic (Tuscan) substratums (Fig. 1 and 2b). Concentric around the base of Amiata, a set of thrust faults with an appreciable diapiric component, have up-thrown blocks at the distal areas of the volcano; at the surface these thrust faults correspond to topographic ridges that divert the drainage and constitute the external boundary of the exploited geothermal fields. Radially-trending small grabens extend from the volcano into the surrounding peripheral country rocks (Betz, 1962). All these faults, being the result of the spreading of Amiata' edifice, tend to remain through time in a nearly ‘permanent’ critically-stressed condition. That is, after a slip event the stress is recharged relatively fast onto fault planes due to the continuous outward spreading of the edifice (Borgia et al., 2014). 3. Geothermal exploitation Geothermal exploitation at Amiata volcano started in 1959 (Cataldi, 1965), when the uplifted evaporites were first perched by wells at depths generally smaller than 1000 m. For over one year the wells produced a fluid with over 80% of CO2, which subsequently evolved to be H2O dominated. In this process, the fluid pressure within the traps decreased on the SW side (Bagnore geothermal field) of the volcano from about 2 MPa to about 0.5 MPa, while on the SE side (Piancastagnaio

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

161

Fig. 1. In the inbox, red boxes represent the study area. In the main image, the map of Amiata volcano, Italia. Volcanic products are depicted in red and black arrows represent direction of the most recent lava flows. Blue arrows indicate the ongoing spreading movement of the volcanic edifice. The location of the geothermal wells of interest in this paper at the southeastern side of the volcano is represented by a red and black triangle, while the black rings locate the 4 seismic stations we used for our investigation (CLC: Colle Lucciole seismic station). The major yellow dot represents the M 3.9 seismic event of 01.04.2000, with a depth of about 1.5 km, and the smaller dots are the aftershocks (locations from Mucciarelli et al., 2001). Profile A–A′, through the spreading-related structure and the geothermal wells arrangement at Piancastagnaio, is shown in Fig. 2. The large blue square represents the area of Fig. 5, in the map.

geothermal field) from about 4 MPa to 0.7–1.7 MPa. In the years 1980s' exploitation was deepened to 2500–3500 m into a second geothermal reservoir, within which, due to exploitation, the original hydrostatic pressure of 20–35 MPa has been reduced to a pressure generally smaller than 10 MPa. The strong pressure drop within the reservoirs has vaporized the originally water-dominated geothermal field, drastically decreasing the density and viscosity, and increasing the compressibility of the geothermal fluid. All of these parameters have probably contributed to increasing the normal stress on fault planes, “locking” them up for some time. Our conceptual model proposes that volcanic spreading can accumulate sufficient stresses on locked faults, eventually forcing them to rupture at a delayed time compared to natural conditions and with stress drops (and earthquake magnitudes) that will tend to be at the higher end of the historic seismic record.

4. Fault-rupture and fluid-induced seismicity Analysis of historical seismicity reveals that, during the last century, the strongest earthquake occurred in 1919, with epicenter at Piancastagnaio and with an estimated magnitude between 5.1 and 5.4. Two other events of magnitudes 5 and 4.6, with epicenters respectively at Mt. Amiata (1948) and near Radicofani (some 10 km E of the volcano, 1958) are reported. The latter is also the last event of M N 4.5 recorded in the area after the beginning of the geothermal exploitation (1959). From May 2000 to June 2001 we recorded the seismic activity in the Piancastagnaio geothermal field reservoir using four three-component short-period (2s) seismic stations on a rectangle with a center placed just north of the epicentral area of the ML 3.9 earthquake, that occurred on 1 April 2000 (Mucciarelli et al., 2001; Ulivieri, 2001; Fig. 1). The

Table 1 Stratigraphic units of the Amiata volcano area, from the top to the bottom. The Mt. Amiata volcanic complex

VC

The intrusive complex IC Miocene-to-Quaternary marine and continental sediments MPQ The Ligurian and SubLigurian Units LI The Lower Tuscan Units The Triassic evaporates (Anidriti of Burano) The Tuscan-units Metamorphic Complex The Paleozoic Gneiss upper crustal complex

Constituted by lava-flows of dacitic, rhyodacitic and olivine-latitic composition, erupted 300–200 ka B.P.

Intrude the upper part of the GC. Neoautoctonus sequence. The first of Middle Jurassic to Early Cretaceous age, and the second of Eocene to Oligocene age, they are remnants of the oceanic basement and its sedimentary cover. LT-TU2 Related to the Late Triassic–Early Miocene sedimentary cover of the Adria continental paleo-margin. AB-TU1 Formed the detachments above which the Tuscan Units were trusted over the outer paleogeographic domain during Late Oligocene–Early Miocene. TMC Found only in boreholes and in xenoliths within the Mt. Amiata lavas. GC

162

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

Fig. 2. a) Stratigraphy of the area, and b) profile A–A′ from Fig. 1. From a) and b) the variable thickness of the Anidriti di Burano layer (AB-TU1) can be seen, result of the differential weight of the fragile edifice over the ductile anhydrates during the last 0.3 Ma. In b) the location of extensive structures near the top of the volcano and compressive structures at its base can be seen along the SE flank of the edifice. The relative position of the producing wells (not in scale in the image) at Piancastagnaio and the inverse fault at the base of the volcano may be related hydrologically, through the development of a transcurrent stress regime at the slope surface, as discussed in the text.

analysis of historical events and the results of previous seismic monitoring (e.g. Batini et al., 1990; Chiarabba and Amato, 1995) have shown that the SE base of Mt. Amiata has the highest density of epicenters in coincidence with the geothermal field of Piancastagnaio.

earthquake hypocenters are at and a few kilometers below exploitation depths. These observations seem to be consistent with our thesis that geothermal exploitation controls the rupture of faults that have been made critically stressed by volcanic spreading.

4.1. Tectonic seismic events

4.2. Hydraulic fracturing seismic events

The majority of recorded seismic events (about 450 of the over 600 events) are characterized by ML b 1, and have emergent P-wave and S-wave onsets. The time difference (Ts–p) of the S- and P-wave arrival of most (90%) of the earthquakes is between 0.6 and 0.8 s with only a small percentage (10%) that have Ts–p N 2 s, while the frequency spectrum typically peaks at 15–20 Hz (Fig. 3). These features suggest that tectonic stresses produce a large number of low-magnitude seismic events. The daily frequency of these events shows that the normal seismic activity of the area is interrupted by periods of higher activity (seismic sequences) in which one seismogenetic structure seems to produce most of the recorded events. Using a half-space velocity model and assuming a Vp = 3.5 km s−1 and a Vs = 2.0 km s−1 (cf. Chiarabba and Amato, 1995), we localized most of the tectonic events within an area about 5 km in radius (Fig. 5). We observe that at this distance we include both the extensional structures that dissect the volcanic edifice and the compressive structures at the base of the volcano (Fig. 1), all of which are related to the volcanic spreading process (Borgia et al., 2014). To reduce hypocenters localization errors, we use only the time difference (tS–tP) between S-wave (tS) and P-wave (tP) arrival times. We estimated the hypocentral errors by using different seismic-wave velocity models with a difference of up to ~ 30% in the seismic velocity. Our analysis shows that hypocenter coordinates have an error of ±200 m horizontally, and ± 500 m vertically inside our network; these errors become twice as much outside the network at the edge of our seismic survey area. Fig. 5 shows that fault-rupture events recorded during our survey, and in particular the one with higher energies, tend to be located on the volcano side at shallow hypocentral depths and close to the extensive structures of the edifice. Hypocentral depths increase from the volcano toward the geothermal field. Within the geothermal field,

Within the recorded ~600 events, a small but significant percentage (about 5%) of the total seismicity clearly show no-brittle fracture features. Duration of seismic event is longer, in general between 20 and 40 s. The wavefield of these events (Fig. 4) starts with an incipient high-frequency (15–20 Hz) phase 3–5 s long, followed by a harmonic oscillation of longer duration (25–30 s), with a lower frequency content (0.6–4 Hz). Arrival times reveal that the high frequency phase is mainly characterized by a P-wave and the odogram shows no evidence of S-waves, whereas the low-frequency part seems dominated by surface Rayleigh waves. No time delay (tS–tP) could therefore be measured. All together, the wavefields indicate that this seismicity cannot be explained with a pure brittle-fracture mechanism, but rather recalls hybrid events, generally related to fluid-filled fracture dynamics. These seismic events are typical of geothermal fields, volcanic areas and glaciers, in which the dynamics of the fluids is responsible for the fracturing of rocks (Bame and Fehler, 1986; Ulivieri, 2001): the increase in the fluid pressure induces a brittle fracture that is responsible for the first high-frequency phase. Eventually, the propagation of the fracture induces the decompression of the fluid and its resonance within the same fracture, giving rise to the low-frequency long-lasting coda (Aki et al., 1977; Chouet and Julian, 1985). These hydrofracturing events cannot be located with classical time inversion algorithm. Therefore, we used a grid-searching method based on the maximum energy distribution at the 4 stations (Marchetti and Ripepe, 2005). In this method, the rock volume (5 × 5 × 8 km large) is discretized every 100 m with 3D matrix of 50 × 50 × 80 nodes. Assuming that each node could represent the position of the seismic source, we generate a 3D map of the probability that a node would contain the source of the actual seismic event. The sum of all probabilities gives the probability of occurrence of hydrofracturing events at any given point.

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

163

Fig. 3. The waveform of a tectonic microearthquake is depicted. The big majority of events are compatible with local tectonic events recorded at an epicentral distance of less than ~20 Km. The impulsive arrival of P- and S-waves (a), the short duration of the signal (b10 s) and a relatively wide spectral range (2–20 Hz, as seen in b) characterize the seismic energy output for the tectonic seismic events.

The result of this localization procedure indicates that the events are mainly concentrated within the Piancastagnaio geothermal field, on the south-eastern side of the Amiata volcano. The vertical source distribution shows that the hypocentral depths are between 2 and 3 km, actually at the depth of the exploited geothermal field, on the vertical of the PC16 wells (Fig. 5). 5. Discussion Microearthquakes are characteristic of geothermal areas. Event hypocenters are correlated with highly fractured rock volumes interested by hydrothermal circulation (Bachmann et al., 2012) and with surface expression of active faults. Analysis of tectonic setting and of historic records of seismic events in geothermal areas is a useful preliminary investigation for resolving exploitation criteria. If, on one hand, geothermal fields usually cover relatively small areas in the order of 10 km2 on the surface (Ward, 1972), on the other hand faults generating M = 5 earthquakes can interest areas of the order of hundreds of square kilometers (Wyss and Brune, 1968; Kanamori and

Anderson, 1975), suggesting that the area strictly influenced by geothermal activity should only be interested by low magnitude events (M b 3). The large majority of geothermal areas, however, being the result of the emplacement of magmatic bodies (e.g., pluton or dykes) at relatively shallow depth in recent geologic times (e.g., Glen et al., 2013) or the result of regional a-magmatic extensions (Faulds et al., 2004), are bordered by major faults created throughout the emplacement/tectonic processes. This implies that stronger earthquakes have the potential to occur. In the case of geothermal fields associated with spreading volcanoes, as in our conceptual model of Mt. Amiata, major faults include the extensional structures generated within the volcanic edifice and the compressional structures found around their base (Borgia et al., 1992, 2000, 2014). Because of force balance, volcanic spreading also induces radial elastic stresses in the basement rocks that equilibrate the shear stresses of the overlying deforming ductile layer on which spreading occurs. Specifically, at Amiata volcano these basement rocks are the Tuscan Metamophic Complex and the underlying Gneiss Complex (cf. Fig. 2a), that in the Piancastagnaio geothermal field are found between 1 and

164

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

Fig. 4. Characteristic waveform of a hydraulic-fracturing type-B event, which differentiates much from the first: a longer duration (N15 s, as in a) with a higher-frequency first impulse in which it is difficult to distinguish S-waves arrival. The spectral content is narrower (0.6–4 Hz, b) and the signal is concluded by a long-period harmonic oscillation.

5 km below the surface (cf. Fig. 2). Therefore, not only normal faults within and thrust faults around the volcanic edifice can be ruptured by the spreading tectonics, but also the regional basement faults around and below the volcano. We have plotted the Gutenberg–Richter relation of the Piancastagnaio area using historical data from 1287 to 1971 (CTPI04), the data from the Italian parametric seismic catalog (ISIDe) from March 1985 to March 2015, and data from our 14-month seismic survey, specifically from September 2000 to March 2001, during which the seismic record is complete (Fig. 6). While historic and ISIDe events, in an area with a radius of about 30 km, have magnitudes M N 1, our network registered events with M N −1. The Gutenberg–Richter frequency distribution shows a b-value of about 0.88, which is in line with values recorded at other seismically active areas in the world (Wyss et al., 1997; Schorlemmer et al., 2005; Shapiro et al., 2013). Historic data suggests that ML = ~5.5 earthquakes may still have to occur within a 100-year time frame. Although our data covers a limited length of time (only 14 months), the good fit with the seismic catalogs indicates that seismicity recorded by our network can be reasonably

considered representative for the long-term background seismic activity of the area. Therefore, we may propose that the natural seismic potential due to regional tectonics and local volcanic spreading has, in the long term, not substantially changed in the area — even though Batini et al. (1990) suggest a larger number of smaller magnitude events occurring in the early stages of exploitation. Higher b-values in the Gutenberg–Richter diagram are in fact found during hydrofracturing (Shapiro et al., 2013). The stress field generated between the extensional and compressional tectonic environments within a spreading volcanic edifice – and the related normal and inverse faults (Borgia et al., 2000; Fig. 1) – with σ1 radial and σ3 tangential to the edifice (e.g., Acocella and Neri, 2009; Gudmundsson et al., 2014), could create radial permeability, enhancing at production depth the hydrological connection between these structures and geothermal fields. In the case of Amiata, critically-stressed conditions of faults, primarily the extensional faults within the SE side of the volcanic edifice (NW of the geothermal field), maintained by the volcanic spreading process, may sustain the vertical permeability of these structures (Sibson, 1990; Mazzoldi et al.,

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

165

Fig. 5. In this 3D image, the focus is on the SE flank of Mt. Amiata where the location and magnitude of the micro-seismic events recorded in 2000/2001 are depicted through the different colors of the squares representing the hypocenters. Frequency in time and space and depth of events increase toward the Piancastagnaio geothermal field where a consistent part of the events recorded (N5%) is due to hydro-fracturing operations, as explained in the text. Hypocenters for these events could not be located. The colors throughout the trace of the well represent depth. LPC-LI: low permeability cap — Ligurid units; SGR-TU1: superficial geothermal reservoir — Anhydrite formation; LPZ-TMC: low permeability zone — Tuscan metamorphic complex; DGR-TMC: deep geothermal reservoir; GC gneiss complex.

Fig. 6. Gutenberg–Richter magnitude/frequency distribution analysis of the Amiata volcano area, for different periods and different areal expansion. A b-value equal to about 0.88 represents the relation between high and low magnitude events, as for other seismically active areas in the world. The absence of M N 4.5 events after the beginning of the exploitation of the geothermal fluid (1958) and the hydrological contact between the producing wells and the inverse basal fault at the SE side of Mount Amiata can suggest the potential for the happening of a M ≥ 5 in the decades to come.

166

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

2012) and link the exploited reservoir with the shallowest volcanic aquifer, which may indeed recharge the geothermal system. At the same time, geothermal fluid exploitation to the SE of the wells may affect rock-volumes near the spreading-related compressive structures at the base of the volcano, lowering pore-pressure, blocking fault segments within the reservoir and allowing more stress to be loaded on these geologic structures. We estimate the seismic hazard at Mt. Amiata by using fault rupture parameters and relating these to the seismic moment and earthquake magnitude of potential events (Hanks and Kanamori, 1979; Abe, 1981; Wells and Coppersmith, 1994). For a rectangular dip-slip fault, the seismic moment is related to the geometry and size of the fault plane and to the earthquake stress drop by the relation (Kanamori and Anderson, 1975): M0 ¼

3 πΔσw2 L 8

ð2Þ

that has been written for a Poissonian solid and where Δσ is the stress drop accompanying the fault' motion, w is the fault width (along dip) and L is the fault length on the surface. In the case of Mt. Amiata, as for other volcanic environments with sized edifices constructed on weaker materials, the natural average stress drop per event can, in principle, reach a maximum value that is of the same order of magnitude of the pressure at the base of the volcano (Borgia et al., 2000). For Amiata this value is about 25 MPa (Table 2). Common stress drops on faults during earthquakes are probably up to one order of magnitude smaller than this value. In fact, Hanks (1977), Allmann and Shearer (2009) and Shaw (2013) indicate for upper crustal settings stress drops in the order of 1– 10 MPa. The inverse faults SE of the Piancastagnaio geothermal field have lengths in the order of 2–8 km and widths-along-dip in the order of 1–6 km (Calamai et al., 1970; Table 2). Thus, accounting for stress drops given above, the seismic moment calculated through Eq. (2) is in the range M0 = 2.3 ∗ 1015–8.5 ∗ 1018 Nm. To calculate the earthquake magnitude from the seismic moment we may use the relation (Hanks and Kanamori, 1979): MW ¼

2 3½ logðMo Þ−9:1

ð3Þ

where Mw is the moment magnitude and M0 is seismic moment. Substituting the values for seismic moment given above, Eq. (3) estimates values of M in the range 4.1–6.5 (cf. Table 2), which are consistent with the seismicity recorded in the area. We believe that these are, reasonably, the maximum magnitudes of earthquakes that could be expected at Amiata volcano from the volcanic spreading model and from the geometry and strengthening potential of faults. Thus these are also the magnitude of earthquakes that could potentially be triggered by the geothermal exploitation.

6. Conclusions A volcanic edifice spreading onto a ductile substratum (Borgia et al., 2000) uses part of its topography-related gravity potential energy for displacing rock masses radially away from the volcano axis. This process induces relatively large stresses in the volcanic edifice itself, in the viscous ductile layer below, and in the shallower crust between them — stresses that extend well beyond the volcanic edifice. For Amiata volcano these stresses (which could conceivably reach 25 MPa) are, in the far field, ultimately equilibrated by elastic deformation of crustal rocks. Clearly, the gravity-induced stresses cannot be enhanced or diminished by geothermal circulation, which can in fact only control pore-pressure that, in turn, influences rocks' properties and faults' rupture and slip conditions. In accordance with this thesis we observe that during our one-year seismic recording, hypocentral depths at Amiata volcano increase away from the volcanic axis – where hypocenters are within the volcanic edifice – toward the Piancastagnaio geothermal field — where hypocenters become also relatively deeper than exploitation depth (Fig. 5). About 5% of the recorded seismic events are located within the geothermal field and have a hydro-fracturing seismic signature, characteristic of hybrid events where the brittle fracture high-frequency phase is followed by the lower frequency resonance of the fluid (Chouet and Julian, 1985). These seismicity is different from the tectonic shear-fracturing related seismicity from regional (or local) sources and is substantially located in proximity of the PC16 geothermal wells. We infer that this seismicity could be the result of the reservoir-rock fracturing produced by geothermal fluid injection/production. A portion of the recorded tectonic events in proximity of the geothermal field may have also been triggered by the fluid injection/production. We have no access to production/injection data for this period but, following our model, it seams reasonable to infer that injection operations should prevail in stimulating faults' reactivation, because they tend to increase pore pressure on critically stressed faults (Mazzoldi et al., 2012). Our study shows that, in the long term, the seismic rate (Fig. 6) has not been substantially changed by geothermal exploitation, implying that magnitude 5–5.5 ML events can occur in the area, with recurrence time in the order of 100 years. This is relevant, because historic earthquakes of such magnitudes have occurred almost 100 years ago in the study area. The potential for lowering pore pressure near fault planes with the consequent augmented faults' strength, if on one hand would delay the occurrence of fault slip, on the other may circumscribe the maximum expectable magnitude of events to the higher side of the range given above. We consider continuous seismic monitoring at the SE flank of Mt. Amiata vital for understanding how the anthropic geothermal activity affects the natural volcanic spreading process and the generation of seismicity. In the long term, geothermal exploitation of reservoirs affected by a near-field differential gravitational-stress component - or by tectonic loading - should aim at maintaining a steady pore pressure during operations in order to inhibit fault strengthening with augmented local seismic risk.

Table 2 Calculations of moment magnitudes (MW). Parameter

Dimensions

Value

Reference

Fault length

min (m) max (m) min (m) max (m) min (Pa) max (Pa) min (Nm) max (Nm) min max

2.00E+03 8.00E+03 1.00E+03 6.00E+03 1.00E+06 2.50E+07 2.36E+15 8.48E+18 4.14E+00 6.49E+00

Calamai et al. (1970) Calamai et al. (1970) Calamai et al. (1970) Calamai et al. (1970) Shaw (2013) Borgia et al. (2000) Kanamori and Anderson (1975) Kanamori and Anderson (1975) Hanks and Kanamori (1979) Hanks and Kanamori (1979)

Fault width-along-dip Stress drop M0 MW

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

Acknowledgments We thank Carlo Alberto Brunori for helping with figures, and the people of Amiata, particularly Fabrizio Tondi and Marcello Perugini for providing support during field work. Financial contribution was provided by the Comunità Montana Senese, Comune di Piancastagnaio (delibera n. 87 del 16/8/01) and Comitato per la tutela e la difesa della Val d'Elsa. We feel in debt with Dr. Carmen Lopez and an anonymous reviewer whose suggestions improved the manuscript in different points.

References Abe, K., 1981. Magnitudes of large shallow earthquakes from 1904–1980. Phys. Earth Planet. Inter. 27, 72–92. Acocella, V., Neri, M., 2009. Dike propagation in volcanic edifices: overview and possible developments. Tectonophysics 471 (2009), 67–77. http://dx.doi.org/10.1016/j.tecto. 2008.10.002. Ake, J., Mahrer, K., O'Connell, D., Block, L., 2005. Deep-injection and closely monitored induced seismicity at Paradox Valley, Colorado. Bull. Seismol. Soc. Am. 95 (2), 664–683. http://dx.doi.org/10.1785/0120040072 (April 2005). Aki, K., Fehler, M., Das, S., 1977. Source mechanism of volcanic tremor: fluid-driven crack models and their application to the 1963 Kilauea eruption. J. Volcanol. Geotherm. Res. 2, 259–287. Allmann, B.P., Shearer, P.M., 2009. Global variations of stress drop for moderate to large earthquakes. J. Geophys. Res. 114, B01310. http://dx.doi.org/10.1029/2008JB005821, 2009. Bachmann, C.E., Wiemer, S., Goerts-Allmann, B.P., Mean, B., Galli, F., 2012. Why geothermal energy research needs statistical seismology. Proceedings of the 37th Workshop on Geothermal Reservoir Engineering. Standford University, Standford, CA (January 30–February 1, 2012. SGP-TR-194). Baisch, S., Harjes, H.P., 2003. A model for fluid-injection-induced seismicity at the KTB, Germany. Geophys. J. Int. 152, 160–170. Bame, D., Fehler, M., 1986. Observation of long period earthquakes accompanying hydraulic fracturing. Geophys. Res. Lett. 13–1, 149–152. Barton, C.A., Zoback, M.D., Moos, D., 1995. Fluid flow along potentially active fault in crystalline rock. Geology 23, 683–686. Batini, F., Fiordelisi, A., Moia, F., 1990. Main features of the seismicity in the Monte Amiata and Latera geothermal areas (Italy). XII General Assembly, European Seismological Commission, Barcelona, pp. 17–22 (September 1990). Betz, A., 1962. Osservazioni geologiche sulle ricerche minerarie per cinabro effettuate nel permesso “Le Bagnore” negli anni 1959–61. Archivio Rimin, Siena, Italia T-587 p. 34. Borgia, A., Ferrari, L., Pasquare, G., 1992. Importance of gravitational spreading in the tectonic and volcanic evolution of Mount Etna. Lett. Nat. 357 (1992), 231–235. Borgia, A., Delaney, P.T., Denlinger, R.P., 2000. Spreading volcanoes. Annu. Rev. Earth Planet. Sci. 28 (2000), 539–570. Borgia, A., Mazzoldi, A., Brunori, C.A., Allocca, C., Delcroix, C., Micheli, L., Vercellino, A., Grieco, G., 2014. Volcanic spreading forcing and feedback in geothermal reservoir development, Amiata volcano, Italia. J. Volcanol. Geotherm. Res. 284 (2014), 16–31. Brogi, A., 2008. The structure of the Monte Amiata volcano-geothermal area (Northern Apennines, Italy): Neogene–Quaternary compression versus extension. Int. J. Earth Sci. (Geol. Rundsch.) http://dx.doi.org/10.1007/s00531-007-0191-1. Cadoux, A., Pinti, D., 2009. Hybrid character and pre-eruptive events of Mt. Amiata volcano (Italy) inferred from geochronological, petro-geochemical and isotopic data. J. Volcanol. Geoth. Res. 179 (2009), 169–190. Calamai, A., Cataldi, R., Squarci, P., Taffi, L., 1970. Geology, geophysics and hydrogeology of the Monte Amiata geothermal fields. Geothermics 1 (150), 1–9. Cataldi, R., 1965. Remarks on the geothermal research in the region of Monte Amiata (Tuscany, Italy). Paper Presented at the IAV International Symposium on Volcanology. New Zealand. Chiarabba, C., Amato, A., 1995. Upper crustal tomographic images of the Amiata–Vulsini geothermal region, central Italy. J. Geophys. Res. 100–B3, 4053–4066. Chouet, B., Julian, B.R., 1985. Dynamics of an expanding fluid-filled crack. J. Geophys. Res. 90–B13, 11,187–11,198. Dannwolf, U.S., Ulmer, F., 2009. A P6000 Report: Technology Risk Comparison of the Geothermal DHM Project in Basel, Switzerland — Risk Appraisal Including Social Aspects. p. RC006. Davies, R., Foulger, G., Bindley, A., Styles, P., 2013. Induced seismicity and hydraulic fracturing for the recovery of hydrocarbon. Mar. Pet. Geol. 45, 171–185 (August 2013). Dost, B., Haak, H.W., 2007. Natural and induced seismicity. In: Wong, T.E., Batjes, D.A.J., de Jager, J. (Eds.), Geology of the Netherlands. Ellsworth, L.W., 2013. Injection-induced earthquakes. Science 341, 1225942. http://dx. doi.org/10.1126/science.1225942 (2013). Engdahl, E.R., 1972. Seismic effects of the MILROW and CANNIKIN nuclear explosions. Bull. Seismol. Soc. Am. 62 (6), 1411–1423. Evans, K.F., Zappone, A., Kraft, T., Deichmann, N., Moia, F., 2012. A survey of the induced seismic responses to fluid injection in geothermal and CO2 reservoirs in Europe. Geothermics 41, 30–54. Faulds, J.E., Coolbaugh, M., Blewitt, G., Henry, C.D., 2004. Why is Nevada in hot water? Structural controls and tectonic model of geothermal systems in the Northwestern Great Basin. Geothermal Resources Council Transactions vol. 28 pp. 649–654 (August 29–September 1, 2004).

167

Ferrari, L., Conticelli, S., Burlamacchi, L., Manetti, P., 1996. Volcanological evolution of the Monte Amiata, Southern Tuscany: new geological and petrochemical data. Acta Vulcanol. 8 (1), 41–56. Finkbeiner, T., Zoback, M.D., Flemings, P., 2001. Stress, pore pressure and dynamically constrained hydrocarbon columns in the South Eugene Island 330 field, northern Gulf of Mexico. AAPG Bull. 85, 1007–1031. Glen, J.M.G., Egger, A.E., Ippolito, C., Athens, N., 2013. Correlation of geothermal spring with sub-surface fault terminations revealed by high-resolution, UAV-acquired magnetic data. Proceedings of the 38th Workshop on Geothermal Reservoir Engineering. Standford University, Standford, CA (SGP-TR-198, February 11–13, 2013). Gudmundsson, A., Lecoeur, N., Mohajeri, N., Thordarson, T., 2014. Dike emplacement at Bardarbunga, Iceland, induces unusual stress changes, caldera deformation and earthquakes. Bull. Volcanol. 76, 869. http://dx.doi.org/10.1007/s00445-014-0869-8. Hanks, T.C., 1977. Earthquake stress-drops, ambient tectonic stress and the stresses that drive plates. Pure Appl. Geophys. 115 (1977), 441–458. Hanks, T.C., Kanamori, H., 1979. A moment-magnitude scale. J. Geophys. Res. 84, 2348–2350. Kanamori, H., Anderson, L.D., 1975. Theoretical basis of some empirical relations in seismology. Bull. Seismol. Soc. Am. 65 (5), 1073–1095. Majer, E.L., Baria, R., Stark, M., Oates, S., Bommer, J., Smith, B., Asanuma, H., 2007. Induced seismicity associated with enhanced geothermal systems. Geothermics 36, 185–222. Majer, E.L., Nelson, J., Robertson-Tait, A., Savy, J., Wong, I., 2012. Protocol for addressing induced seismicity associated with enhanced geothermal systems. Energy Efficiency and Renewable Energy, Report. DOE/EE-0662. Manzocchi, T., Childs, C., Walsh, J.J., 2010. Faults and fault properties in hydrocarbon flow models. Geofluids 10, 94–113. http://dx.doi.org/10.1111/j.1468-8123.2010.00283.x. Marchetti, E., Ripepe, M., 2005. Stability of the seismic source during effusive and explosive activity at Stromboli Volcano. Geophys. Res. Lett. 32 (3), 1–5. http://dx.doi.org/ 10.1029/2004GL021406. Mazzoldi, A., Rinaldi, A.P., Borgia, A., Rutqvist, J., 2012. Induced seismicity within geological carbon sequestration projects: maximum earthquake magnitude and leakage potential from undetected faults. Int. J. Greenh. Gas Control. 10 (2012), 434–442. http:// dx.doi.org/10.1016/j.ijggc.2012.07.012. McGarr, A., 1976. Seismic moments and volume changes. J. Geophys. Res. 81, 1487–1494. Morris, A.P., Ferril, D.A., McGinnis, R.N., 2010. Fault frequency and strain. Litosphere 1 (2009), 105–109. http://dx.doi.org/10.1130/L23.1. Mucciarelli, M., Gallipoli, M., Fiaschi, A., Pratesi, G., 2001. Osservazioni su danneggiamento nella zona del Monte Amiata a seguito dell'evento del 1 Aprile 2000. X Congresso Nazionale “L'Ingegneria Sismica in Italia”, Potenza-Matera 9–13 Settembre 2001. Rutqvist, J., 2012. The geomechanics of CO2 storage in deep sedimentary formations. I.J. Geotech. Geol. Eng. http://dx.doi.org/10.1007/s10706-011-9491-0. Schorlemmer, D., Wiemer, S., Wyss, M., 2005. Variations in earthquake-size distribution across different stress regime. Nature 437, 539–542. http://dx.doi.org/10.1038/ nature04094. Seagall, P., 1989. Earthquake triggered by fluid extraction. Geology 17, 942–946. Shapiro, S.A., Krüger, O.S., Dinske, C., 2013. Probability of induced given-magnitude earthquakes by perturbing finite volumes of rocks. J. Geophys. Res. Solid Earth 118, 3557–3575. http://dx.doi.org/10.1002/jgrb.50264. Sharma, N., Convertito, V., Maercklin, N., Zollo, A., 2013. Ground-motion prediction equations for the Geysers geothermal area based on induced seismicity records. Bull. Seismol. Soc. Am. 103 (1), 117–130. http://dx.doi.org/10.1785/0120120138. Shaw, B.E., 2013. Earthquake surface slip-length data is fit by constant stress drop and is useful for seismic hazard analysis. Bull. Seismol. Soc. Am. 2A, 876–893. http://dx.doi. org/10.1785/0120110258. Sibson, R.H., 1990. Conditions for fault-valve behavior. Geol. Soc. Lond. Spec. Publ. 54, 15–28. Stabile, T.A., Giocoli, A., Lapenna, V., Perrone, A., Piscitelli, S., Telesca, L., 2014. Evidence of low-magnitude continued reservoir induced seismicity associated with the Pertusillo artificial lake (southern Italy). Bull. Seismol. Soc. Am. 104 (4), 1820–1828. http://dx. doi.org/10.1785/0120130333. Styles, P., Gasparini, P., Hunges, E., Scandone, P., Lesocki, S., Terlizzese, F., 2014. Report on the hydrocarbon exploration and seismicity in Emilia region. International Commission on Hydrocarbon Exploration and Seismicity in the Emilia Region. ICHESE Report (February). Suckale, J., 2010. Moderate-to-large seismicity induced by hydrocarbon production. Lead. Edge (March 2010) http://dx.doi.org/10.1190/1.3353728 29 [3] 310-319. Sumy, F.D., Cochran, E.S., Keranen, K.M., Wei, M., Abers, G.A., 2014. Observations of static Coulomb stress triggering of the November 2011 M5.7 Oklahoma earthquake sequence. J. Geophys. Res. http://dx.doi.org/10.1002/2013JB010612. Torabi, A., Berg, S.S., 2011. Scaling of fault attribute: a review. Mar. Pet. Geol. 28, 1444–1460. http://dx.doi.org/10.1016/j.marpetgeo.2011.04.003. Regione Toscana, 2006–2009. Carta geologica regionale, scale 1:10.000, http://www. regione.toscana.it/-/carta-geologica. Ulivieri, G., 2001. Analisi della sismicita' nel campo geptermico di Piancastagnaio (M.S. Thesis), University of Firenze, Department of Earth Sciences (unpublished work). Ward, P.L., 1972. Microearthquakes: prospecting tool and possible hazard in the development of geothermal resources. Geothermics 1 (1), 1–12. Wells, D.L., Coppersmith, K.J., 1994. New Empirical Relationships Among Magnitude, Rupture Length, Rupture Width, Rupture Area and Surface Displacement. Bull. Seismol. Soc. Am. 84 (4), pp. 974–1002. Wyss, M., Brune, J.N., 1968. Seismic moment, stress and source dimensions for earthquakes in the California–Nevada region. J. Geophys. Res. 73, 4681. Wyss, M., Shimazaki, K., Wiemer, S., 1997. Mapping active magma chambers by b values beneath the off-Ito volcano, Japan. J. Geophys. Res. 102 (B9), 20,413–20,422.

168

A. Mazzoldi et al. / Journal of Volcanology and Geothermal Research 301 (2015) 159–168

Yielding, G., Freeman, B., Needham, D.T., 1997. Quantitative fault seal prediction. AAPG Bull. 81 (6), 897–917. http://dx.doi.org/10.1306/522B498D-1727-11D78645000102C1865D. Yielding, G., Bretan, P., Freeman, B., 2010. Fault seal calibration: a brief review. Geol. Soc. Lond. Spec. Publ. 347, 243–255. http://dx.doi.org/10.1144/SP347.14.

Yoon, J.S., Bruhn, D., Zang, A., 2013. Overview of geothermal energy projects in Europe and the GEISER Project on induced seismicity. Tunn. Undergr. Space 23 (6), 581–592. http://dx.doi.org/10.7474/TUS.2013.23.6.581. Zoback, M.D., Gorelick, S.M., 2012. Earthquake Triggering and Large-scale Geologic Storage of Carbon Dioxide. PNAS. http://dx.doi.org/10.1073/pnas.1202473109.