1
Subsidence due to crack closure and depressurization of hydrothermal systems: a case study from Mt. Epomeo (Ischia Island, Italy)
Vincenzo Sepe 1,2 , Simone Atzori 1 , Guido Ventura 3
1 Istituto Nazionale di Geofisica e Vulcanologia, Earthquake National Center, Roma, Italy
2 Istituto Nazionale di Geofisica e Vulcanologia -Osservatorio Vesuviano, Napoli, Italy
3 Istituto Nazionale di Geofisica e Vulcanologia, Department of Seismology and Tectonophysics,
Roma, Italy
Corresponding author: Guido Ventura, Istituto Nazionale di Geofisica e Vulcanologia, Department of
Seismology and Tectonophysics, Via di Vigna Murata 605 - 00143 Roma, Italy
Phone (+39) 06-51860221; Fax (+39) 06-5041303; e-mail: ventura@ingv.it
Abstract
Leveling surveys carried out between 1990 and 2003 on the Mt. Epomeo resurgent block (Ischia Island) record negative dislocations on its northern and southern flanks with a maximum subsidence rate of 1.27 cm/yr. This deformation is not associated to the cooling, crystallization or lateral drainage of magma and cannot be explained by a pressure point or prolate ellipsoid source. Results from dislocation models and the available structural and geochemical information indicate that the subsidence is due to crack closure processes along two main ENE-WSW and E-W pre-existing faults, which represent the preferred pathway of
CO
2
degassing from the hydrothermal system located beneath Mt. Epomeo. The monitoring of the dislocations and CO
2
flux along these faults could give useful information on the dynamics of the hydrothermal system.
1. Introduction
The subsidence of active volcanic areas is a relatively poorly studied process because it is not associated to pre-eruptive or eruptive dynamics, which usually produce uplift. The mechanisms proposed to explain the subsidence of volcanoes include (a) post-eruptive magma withdrawal (Dvorak and Dzurisin, 1997), (b) magma cooling and crystallization
(Sturkell and Sigmundsson, 2000), (c) lateral magma drainage from a reservoir (De Zeeuwvan Dalfsen et al., 2004), and (d) passive degassing from calderas (Oh-Ig Kwoun et al., 2006).

2
These mechanisms produce negative, bell-shaped deformations, with maximum dislocations located in the central sector of volcanoes or calderas. Here, we report leveling data collected on the flanks of Mt. Epomeo (786 m a.s.l., Ischia Island, Italy) between 1990 to 2003. These data show that Mt. Epomeo is subsiding. However, this subsidence cannot be explained by pressure point or prolate/oblate ellipsoid sources. We inverted the leveling data using the available geological and geochemical information as constraints, and propose that the recorded subsidence is related to crack closures induced by a decrease in the fluid pressure within the hydrothermal system. The present-day dynamics of the hydrothermal system and the implications for the monitoring strategies are briefly discussed.
2. Geological setting
Ischia is the westernmost, active volcanic complex of the Campanian Plain (Italy), which also includes Campi Flegrei, Procida and Vesuvius (Fig. 1), and it is composed of trachybasalts to phonolites and alkali-trachytes (Civetta et al., 1991), marine and landslide deposits (Fig. 1). The oldest volcanics (130-150 ka) crop out along the coast (Vezzoli, 1988).
The ‘Green Tuff’ ignimbrite emplaced at about 55 ka forming a caldera depression (Gillot et al., 1982) in which the marine Tuffite and Colle Jetto formations emplaced (Fig. 1). The
‘Green Tuff’ lithology is that of a welded tuff. Citara Tuff emplaced between 44 and 33 ka, and, around 33 ka, a 800 to 1100 m uplift affected the Mt. Epomeo resurgent block (Fig. 1;
Gillot et al., 1982). Successive volcanism concentrated in the last 3 ka, in the eastern part of the island (Orsi et al., 1991). The most recent eruption (Arso lava flow) occurred in 1302 AD.
Four main fault systems affect Ischia (Vezzoli et al., 1988; Molin et al., 2003): a ENE-
WSW to E-W striking system cuts the northern and southern flanks of Mt. Epomeo, while
NNW-SSE to N-S faults delimit its eastern and western sectors; NE-SW faults affect the eastern sector of Ischia; NW-SE faults outcrop in the southwestern corner of the island. The southern and northwestern flanks of Mt. Epomeo are affected by active landslides (Fig. 1).

3
The slide of the northwestern sector developed after the 1883 Casamicciola earthquake
( I o
=IX X MCS; Postpischl, 1985). Historical earthquakes occurred between 1762 and 1883 and were related to the E-W striking faults (Alessio et al., 1996; Fig. 1). However, the
‘shallow’ source of the 1883 earthquake could also be due to the dynamics of the Ischia hydrothermal system, which consists of two main reservoirs: a shallower reservoir
(temperature ~250°C) located at 100-300 m of depth, and a deeper one (temperature ~300°C) at about 900 m of depth (Chiodini et al., 2004). According to Penta and Conforto (1951), the hydrothermal circulation concentrates in the highly fractured lavas underlying the Green Tuff
(Fig. 1; Penta and Conforto, 1951). The hydrothermal fluids rise to the surface along the main faults and cracks delimiting the northern and western flanks of Mt. Epomeo, and along the fault affecting its southern, upper slope (see Fig. 1; Molin et al., 2003; Pecoraino et al., 2005).
Subsidence at Ischia is testified by Greek and Roman ruins at about 2 m b.s.l.
(Friedlander, 1938). Precision leveling surveys were done by the Italian Army Geographic
Institute from 1913 and by the Italian Geodetic Society in 1967. This latter survey shows that, relatively to 1913, the central and southern sectors of the island subsided with a maximum dislocation of 31 cm, and a little uplift affected the Ischia northern sector (Maino and
Tribalto, 1971). Using spaceborne Synthetic Aperture Radar images collected between 1993 and 2003, Manzo et al. (2006) suggest that the vertical deformation of Ischia is due to the combined effects of active landsliding and fault activity. However, these data cover the zones of the island located along the coastline, where the presence of buildings allows interferometric processing; on the contrary, dense vegetation prevents from obtaining a good result on Mt. Epomeo.
3. Levelling surveys
The INGV-Osservatorio Vesuviano leveling network of Ischia is shown in Fig. 2a.
Leveling surveys were carried out in 1990, 1994, 1999, 1997, 2001 and 2003. The reference

4 benchmark (bm 1 in Fig. 2a) is located in the northeaster sector of the island, which is considered a stable area (INGV-OV, 2001). The network consists of 230 benchmarks with a mean distance of 300 m on about 100 km of nine circuits besides the coast line. Auto levels equipped with plain-parallel lamina micrometers and couples of rods with invar ribbon have been used in all surveys. Measurements are submitted to least squares adjustment with the method of indirect observations, with a final value of the standard deviation per unit of weight of 0.965 mm. Here, we report results from leveling surveys carried out between 1990 and
June 2003 on the ‘Coastline’, ‘Borbonic line’ and ‘Casamicciola-Chiummano line’, which bound and cross Mt. Epomeo (Fig. 2a). The measured error on the single circuits are less than the expected errors (Table 1).
The vertical deformation recorded in the 1990-2003 period is reported in Fig. 2b. A maximum value of -16.5 cm occurs in the Fango area of the ‘Borbonic line’ at bm 100. Here, the maximum subsidence rate is 1.27 cm/year. The data along this line (Fig. 2b) show two relatively stable areas located in the westernmost and northeasternmost sectors of Ischia (bm
63-65 and bm 86). The subsidence concentrates between bm 93 and 103. However, this subsidence is not constant and increases between bm 97 and 103, where the leveling survey partly crosses the active landslide associated to the 1883 earthquake.
A significant subsidence of -8.6 cm occurs in the Serrara area of the ‘Coastline’, along the southern flank of Mt. Epomeo (Fig. 2b). The maximum subsidence rate is 0.65 cm/year at bm 35. The data of the ‘Coastline’ reveal that the northern sector of Ischia is not affected by significant vertical deformations (bm 1-15 and bm 69-90). Subsidence increases moving from these outer sectors toward the southern flank of Mt. Epomeo. An abrupt increase in subsidence occurs between bm 24 and 43 (Fig. 2b), where the ‘Coastline’ partly crosses active landslides and the ENE-WSW fault affecting the southern flank of Mt. Epomeo (see Fig. 1).
The leveling survey carried out between 1994 and 2003 along the ‘Casamicciola-
Chiummano line’, which crosses the sector of the island where the more recent volcanic activity concentrated (Fig. 1 and 2a), shows a maximum vertical displacement of -3.6 cm in 9

5 years, i.e., a subsidence rate of 0.4 cm/yr. The subsidence rate along the ‘Casamicciola-
Chiummano line’ is roughly constant. The leveling data show that the subsidence rate at bm
35 (Coastline) and bm 100 (Borbonic line), which are the benchmarks where the maximum values of subsidence are recorded, is constant with time (Fig. 3).
As concern the top of Mt. Epomeo, where leveling lines are lacking (Fig. 2a), the data collected during 1999, 2001, and 2003 GPS surveys (INGV-Osservatorio Vesuviano, 1999,
2001, 2003) show that this is a stable area, being the recorded displacement within the measurement error (Pingue et al., 2003).
4. Discussion
According to the displacement pattern inferred from the leveling surveys, two distinct areas are affected by a evident subsidence. These are located on the northern and southern flanks of Mt. Epomeo. The largest values of vertical displacement recorded along the
‘Coastline’ and ‘Borbonic’ line partly overlap the flanks of Mt. Epomeo affected by active landslides. However, the subsidence pattern along these lines (Fig. 2b) largely exceeds the linear extent of the slides. In addition, the nearly constant 1990-2003 subsidence rate on the benchmarks where the maximum displacements occur (Fig. 3) cannot be interpreted as the result of slope instability processes because the Ischia slides are characterized by intermittent, discontinuous movements (Del Prete and Mele, 1999). Therefore, even if a local contribution of landsliding to the whole deformation field of Mt. Epomeo cannot be completely ruled out, the recorded subsidence is not associated to exogenous processes alone. We also exclude that the subsidence is due to the contraction induced by cooling and crystallization of magma at depth because the recent (<3 ka) volcanic activity developed in the eastern sector of the island. Structural data (Vezzoli, 1988) show two main faults striking N80°E and N60°E located in correspondence of the largest displacements (see source 1 and 2 in Fig. 1 and 2), and the geochemical data (Pecoraino et al. 2005) identify the occurrence of CO
2
-rich (70>CO
2

6 flux>18 g m -2 d -1 ) degassing areas along these faults, This calls for a mechanism of subsidence that involves tectonics and/or degassing. Leveling data (Fig. 2a,b) also suggest that the displacement is related to two different sources. These sources are characterized by distinct dislocation mechanisms such to have displacement patterns spreading toward the northernmost and the southern sectors of Mt. Epomeo, without affecting its top. Such a displacement field cannot be due to a simple pressure point source (Mogi, 1958) or to a finite prolate/oblate ellipsoid source (Yang et al., 1988); both of them produce a circular or pearshaped displacement pattern that is incompatible with the observed data. We also exclude that the recorded deformation is related to the gravitational collapse of Mt. Epomeo because the available interferometric data on the horizontal displacements reveal that Ischia and Mt.
Epomeo are affected by a general contraction (Manzo et al., 2006).
We apply more realistic shear dislocation and crack closure (Okada, 1985) models.
We use a non-linear inversion algorithm based on Simulated Annealing (Kirkpatrick et al.,
1983) and a downhill refinement based on Simplex method (Nelder et al., 1965). Due to the low number of observations, we constrained some of geometric source parameters by mean of geological and morphological data (strike, dip and length of the faults). We inverted for shear dislocation and crack closure the faults affecting the northern and southern flanks of Mt.
Epomeo (sources 1 and 2 in Fig. 1). The fit of the observed data retrieved for the shear model is not as good as that obtained by a crack closing model (11.3 mm of rms and 9.7 mm, respectively). Shape and intensity of the two models are similar, but shear models imply the presence of a consistent uplift of Mt. Epomeo. This uplift is 47 mm, a value higher than that obtained by a tensile model, which is 19 mm. This latter value is within the error of the GPS vertical component (~20 mm; INGV-OV, 2003). As a result, we propose that the displacements measured on the Mt. Epomeo flanks are due to crack closure and not to shear dislocations. The parameter of the modelled sources are summarized in Table 2. The displacement field corresponding to the models is shown in Fig. 4a, and the differences between the measured (leveling data) and the calculated values is reported in Fig. 4b. This

7 latter plot shows that the majority of the calculated vertical displacements are within the error bars of the measured values, thus supporting the goodness of the crack closure model. Few points outside the error bars could be due to some intrinsic limits of the model, which assumes planar surfaces and does not consider the possible layering and/or fracturing of the crust and its effect on the surface deformations (e.g., Gudmundsson, 2003). In addition to these limits, the partial superimposition of local effects related to sliding phenomena on the crack closure can also explain the small differences between the calculated and measured values reported in Fig. 4b. Nevertheless, the crack closure model fits well the vertical deformations measured on the Mt. Epomeo flanks. The crack model is supported by the structural features of the faults of Mt. Epomeo. The fault scarps of the northern and southern flanks of Mt. Epomeo (sources 1 and 2 in Fig. 1) are about 10 m high; the damage zones are 1 to 4 m wide. These zones consists of interconnected to parallel veins and cracks with a maximum thickness of 23 cm. The veins are filled of hydrothermal minerals. According to
Chiodini et al. (2004), shear displacements along these faults mainly occurred during the resurgence of Mt. Epomeo. After the 33 ka old resurgence episode, the cracks and veins within the damage zone of these faults controlled the flow of hydrothermal fluids.
5. Concluding remarks
The data presented here show that, between 1990 and 2003, Mt. Epomeo has been affected by a subsidence with two maxima located on its northern and southern sectors. These displacements are due to the closure of cracks associated to ENE-WSW to E-W pre-existing faults along which degassing processes occur. We propose that the recorded dislocations reflect a decrease in the fluid pressure within these cracks. To have a semi-quantitative estimate of the decrease in fluid pressure

P within the modelled crack closures, we use the following relation (Gudmudsson, 1999 and his references):

P


WE
2 L ( 1
 v
2
)

8 where

W is the variation of the vein width (negative for closure), L is the length of the crack, E is the Young’s modulus and

is Poisson’s ratio. Assuming E =15 GPa,

=0.25
(Gudmudsson, 1999) and using the crack parameters L and

W listed in Table 2 we obtain a decrease in fluid pressure of 2.14 MPa for the source 1 and 0.85 MPa for the source 2. This rough estimates suggest that the depressurization of the Ischia hydrothermal system is mainly controlled by source 1, which is located on the northern flank of Mt. Epomeo (Fig. 1). We exclude that the progressive crack closure is due to self-sealing processes because self-sealing induces an increment in the fluid pressure with consequent, possible local uplifts. The decrease in the fluid pressure within the ENE-WSW to E-W striking cracks at Mt. Epomeo, could reflect a decrease in the fluid pressure within the larger scale, deep hydrothermal system, as also suggested by (Manzo et al., 2006) on the basis of radar interferometric data, and by the nearly constant subsidence rate recorded along the ‘Casamicciola-Chiummano line’ (Fig. 2b). The results of this study indicate that changes, i.e. uplift or decrease in the subsidence rate, in the dislocation pattern recognized here could reflect changes, i.e., pressurization, in the deep hydrothermal system. As a conclusion, the monitoring of the dislocations and of the CO
2
flux along the cracks could give useful information on the dynamics of the Mt. Epomeo hydrothermal system.
Acknowledgments This study was supported by INGV-GNV funds. We thank Peter Cervelli and Maurizio Battaglia and the Associate Editor of Terra Nova for the perceptive comments and suggestions. Original data were acquired by the Department of Geodesy of INGV-
Osservatorio Vesuviano from 2001.
References
Alessio, G., Esposito, E., Ferranti, L., Mastrolorenzo, G., Porfido, S., 1996. Correlazione tra sismicita` ed elementi strutturali nell’isola di Ischia. II Quaternario, 9(1) , 303–308.
Chiodini, G., Avino, R., Brombach, T., Caliro, S., Cardellini, C., De Vita, S., Frondini, F.,
Granirei, D., Marotta, E., Ventura, G., 2004. Fumarolic and diffuse soil degassing west of Mount Epomeo, Ischia, Italy. J. Volcanol. Geotherm. Res, 133 , 291–309.

9
Civetta, L., Gallo, G., Orsi, G., 1991. Sr- and Nd-isotope and trace element constraints on the chemical evolution of the magmatic system of Ischia (Italy) in the last 55 ka. J.
Volcanol. Geotherm. Res, 46 , 213–230. de Zeeuw-van Dalfsen, E., Rymer, H., Sigmundsson, F., Sturkell, E, 2004. Net gravity decrease at Askja volcano, Iceland: constraints on processes responsible for continuous caldera deflation, 1988–2003. J. Volcanol. Geotherm. Res, 139 , 227–239.
Del Prete, S., and Mele, R. 1999. L'influenza dei fenomeni d'instabilita di versante nel quadro morfoevolutivo della costa dell'isola d'Ischia. Boll. Soc. Geol. It., 118 , 339-361.
Dvorak, J.J., and Dzurisin, D., 1997. Volcano geodesy: The search for magma reservoirs and the formation of eruptive vents. Reviews of Geophysics , 35 , 343–384.
Friedlander, L., 1938. Sui bradisismi dell’isola d’Ischia e sulla Grotta del Mago.
Boll. Soc.
Geogr. It., 3(1) , 4–54.
Gillot, P.Y., Chiesa, S., Pasquare, G., Vezzoli, L., 1982. 33.000 yr K/Ar dating of the volcano-tectonic horst of the island of Ischia, Gulf of Naples. Nature , 229 , 242–245.
Gudmundsson, A., 1999, Fluid overpressure and stress drop in fault zones. Geophysical
Research Letters, 26 , 115–118.
Gudmundsson, A., 2003. Surface stresses associated with arrested dykes in rift zones. Bulletin of Volcanology , 65 , 606-619.
INGV-Osservatorio Vesuviano, 2001.
Rendiconto sull’attivita` di sorveglianza
.
INGV-Osservatorio Vesuviano, 2002.
Rendiconto sull’attivita` di sorveglianza
.
INGV-Osservatorio Vesuviano, 2003.
Rendiconto sull’attivita` di sorveglianza
.
Kirkpatrick, S., Gelatt, C. D., Vecchi, M. P., 1983. Optimization by Simulated Annealing.
Science, 220 , 671-680.
Maino, A., and Tribalto, G., 1971. Rilevamento gravimetrico di dettaglio dell’isola d’Ischia
(Napoli). Boll. Serv. Geol. It.
, 92 , 109–123.
Manzo, M., Ricciardi, G. P., Casu, F., Ventura, G., Zeni, G., Borgström, S., Berardino, P., Del
Gaudio, C., Lanari, R., 2006. Surface deformation analysis in the Ischia Island (Italy) based on spaceborne radar interferometry. J. Volcanol. Geotherm. Res.
, doi:10.1016/j.jvolgeores.2005.09.010.
Mogi, K., 1958. Relation between the eruptions of various volcanoes and deformations of the ground surfaces around them. Bull. Earth Res. Inst.
, 36 , 99–134.
Molin, P., Acocella V., Funiciello R., 2003. Structural, seismic and hydrothermal features at the border of an active intermittent resurgent block: Ischia island (Italy). J. Volcanol.
Geotherm. Res.
, 121 , 65-81.
Nelder, J.A. and Mead, R., 1965. A Simplex Method for Function Minimization. Computer
Journal , 7 , 308-313.

10
Oh-Ig Kwoun, Lu Z., Neal C., Wicks Jr., C., 2006. Quiescent deformation of the Aniakchak
Caldera, Alaska, mapped by InSAR. Geology , 34 , 5–8.
Okada, Y., 1985. Surface deformation due to shear and tensile faults in a half-space. Bull.
Seism. Soc. Am ., 75 , 1135-1154.
Orsi, G., Gallo, G., Zanchi, A., 1991. Simple-shearing block resurgence in caldera depressions. A model from Pantelleria and Ischia. J. Volcanol. Geotherm. Res.
, 47 , 1–
11.
Pecoraino, G., Brusca, L., D'Alessandro, W., Giammanco, S., Inguaggiato, S., Longo, M.,
2005. Total CO
2
output from Ischia Island volcano (Italy). Geochemical Journal , 39 ,
451-458.
Penta, F., Conforto, B., 1951. Risultati di sondaggi e di ricerche geominerarie nell’isola d’Ischia dal 1939 al 1943, nel campo del vapore, delle acque termali e delle ‘forze endogene’ in generale. Ann. Geoph., 4 , 159-191.
Pingue, F., Del Gaudio, C., De Natale, G., Obrizzo, F., Sepe, V., Cecere, G., De Martino, P.,
Malaspina, S., Serio, C., Siniscalchi, V., Tammaro, U., 2003. Monitoring and deformation analysis in neapolitan volcanic area (southern italy). Proceedings 11th FIG
Symposium on Deformation Measurements , Santorini, Greece, 2003.
Postpischl, D., 1985. Catalogo dei terremoti italiani dall’anno 1000 al 1980.
Quaderni della
Ricerca Scientifica 114(2B), C.N.R, Roma.
Sturkell, E., and Sigmundsson, F., 2000. Continuous deflation of the Askja caldera, Iceland, during the 1983–1998 noneruptive period. J. Geophys. Res.
, 105 , 25671–25684.
Vezzoli, L., 1988. Island of Ischia. Quaderni Ricerca Scientifica 114, C.N.R, Roma.
Yang, X., Davis, P.M., Dieterich, J., 1988. Deformation from inflation of a dipping, finite, prolate ellipsoid in an elastic half-space, J. Geophys. Res.
, 93 , 4249–4257.

Figure captions
11
Fig. 1 Location and geological map of Ischia island, simplified from Vezzoli (1988).
Epicenters of historical earthquakes are from Alessio et al. (1996). Hydrothermal manifestations are from Chiodini et al. (2004) and Pecoraino et al. (2005). Source 1 and
Source 2 are the Mt. Epomeo faults selected for the dislocation models reported in Fig. 4a and
Table 2.
Fig. 2 (a) INGV-Osservatorio Vesuviano leveling network at Ischia Island. The ‘Coastline’
(bleu), ‘Borbonic’ line (green), and ‘Casamicciola-Chiummano’ line (black) are shown with thicker lines. The other leveling lines are in red. Numbers are the benchmarks (b) Vertical displacements recorded along the ‘Coastline’ (bleu), ‘Borbonic’ line (green), and
‘Casamicciola-Chiummano’ line (black) from 1990 to 2003. Numbers are the benchmarks.
Fig. 3 Time (years) vs. vertical displacement at benchmarks (Bm) 35 of ‘Coastline’ and 100 of ‘Borbonic’ line. The linear interpolations suggest constant subsidence rates (1.27 cm/yr at bm 100 and 0.65 at bm 35).
Fig. 4 (a) 1990-2003 vertical displacement field modeled according to the source parameters listed in Table 2. Boxes represent the plane view of the crack planes, while the dashed lines are the projection of the planes on the surface. Observed displacement for the leveling benchmarks are also shown. (b) Comparison between the calculated (red dots) and measured
(black dots) displacement. Error bars of the measured vertical displacements are also reported.
Lines between points show the dislocation trends.

Table 1.
Error percentages on the single circuits ‘Coastline’ and ‘Borbonic line’.
Circuit
Length
(km)
Coastline 32.4
Borbonic line 11.9
Measured error
(MEE; mm)
-3.2
-2.3
Maximum expected error
(MAE; mm)
11.4
6.9
MEE/MAE %
28.1
33.4
12
Table 2 . Source parameters for the tensile (crack closure) model. The location of the source is reported in Fig. 4a.
Length
[m]
Width
[m]
Depth
[m] (a)
Dip
[°]
50
Strike
[°E]
80
East
[m] (b)
North
[m] (b)
Displacement
[m]
Source 1 2050 670 600 406250 4510850 -0.55
Source 2 2800 1800 600 70 60 407050 4508960 -0.30
(a)
Depth of the top edge of the dislocation.
(b)
UTM coordinates, WGS84 system, zone 33.