Ž .Journal of Volcanology and Geothermal Research 82 1998 219–238

Electric and electromagnetic outline of the MountSomma–Vesuvius structural setting

R. Di Maio a, P. Mauriello a, D. Patella a,) , Z. Petrillo a, S. Piscitelli a,A. Siniscalchi b

a Department of Geophysics and Volcanology, UniÕersity Federico II, Largo S. Marcellino 10, 80138 Naples, Italyb Geomare Sud, CNR Institute of Marine Geology, Naples, Italy

Received 8 October 1996; accepted 3 February 1997

Abstract

We present and discuss the results of an integrated electrical and electromagnetic survey in the active volcanic area ofŽ . Ž . Ž . Ž .Mount Somma–Vesuvius Naples, Italy . Dipolar geoelectrics DG , self-potential SP and magnetotellurics MT were

used to investigate the shallow and deep regions of the volcanic area. The DG apparent resistivity pseudosection along aN–S profile across the Vesuvius cone showed the existence of a largely extended conductive zone, closely in correspon-dence to the Somma caldera, including in the middle the top terminal part of the Vesuvius main plumbing system. The SPdata, collected over the whole volcanic area, showed the existence of a W–E-directed wide band of weak positive anomalies,indicating again a conductive zone, not only including the whole Somma caldera but also extending towards the Tyrrheniansea. A roughly N–S-trending narrow fracture system, cutting the lowest Mount Somma eastern slopes, was further evidentfrom the SP data. A new SP tomographic inversion procedure allowed to detect a large positively charged nucleus in thedepth range 600–2200 m b.g.l., located beneath the westernmost portion of a former caldera, related to the Avellino plinianeruption. The geophysical interpretation of this large positive anomaly was made using Onsager’s theory of coupledelectrokinetic and thermoelectric flows. The final interpretation was that the shallow, conductive central zone is very likelymade up of an intensively altered and mineralised block of cemented volcanic breccia. Finally, the MT data, distributedalong two perpendicular profiles, enabled us to obtain the first significant picture of the deep electrical structure of thevolcano. The Bostick inversion revealed the existence of a conductive intracrustal layer, including a perched moreconductive zone located roughly beneath the central-western sector of the Vesuvius apparatus. q 1998 Elsevier Science B.V.All rights reserved.

Keywords: physical volcanology; Mt. Somma–Vesuvius; geophysics; electric methods; electromagnetic methods

1. Introduction

Mount Somma–Vesuvius is considered a high-riskactive volcanic system, because of its history of

) Corresponding author. Tel.: q39-81-5803108; fax: q39-81-5527631; e-mail: patella@dgvna.dgv.unina.it

recurrent devastating manifestations in the last 2000Ž .years Scandone et al., 1993 and for its being

nowadays surrounded by a very high density ofpopulation as far as the southeastern Neapolitan sub-urbs. Although Mount Somma–Vesuvius is one ofthe most studied volcanoes in the world, integratedgeophysical data, sufficient enough to enable the

0377-0273r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved.Ž .PII S0377-0273 97 00066-8

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238220

shaping of a reliable structural model, are still lack-ing. About five years ago, wide-range geophysicalinvestigations in the Vesuvian area started to bepromoted within a renewed research strategy brought

Ž .forward by the Italian Group of Volcanology GNVŽ .of the National Research Council CNR . Since the

beginning we were involved in dipolar geoelectricŽ . Ž . Ž .DG , self-potential SP and magnetotelluric MThigh-resolution survey planning and execution, whichhad never been taken before into consideration.

Since electric parameters are strongly dependenton presence and abundance of mineral particles andfluids, as well as on temperature and pressure, arelevant role is played, in principle, by electric andelectromagnetic geophysical methods in investigat-ing volcanic environments. Despite such a basictenet, to our knowledge no systematic electric androrelectromagnetic study was previously carried out inthe Vesuvian area, except an extended conventionalshallow-depth geoelectrical survey, performed by

Ž .Cassano and La Torre 1987 for geothermal pur-poses, but outside and around the volcanic complex.Our present contribution is thus an effort trying tofill this gap, in the awareness that much more workis yet to be done in order to reach an exhaustiveanswer to the many problems still standing partiallyor totally unsolved.

SP and geoelectrical experiments are by long timeperformed in volcanological and geothermal re-

Žsearch, like, e.g., in the Yellowstone Zohdy et al.,. Ž .1973 , Kilauea Zablocki, 1976 , Long Valley

Ž . ŽAnderson and Johnson, 1976 , Stromboli Bal-. Ž .lestracci, 1982 , Balcova Ercan et al., 1986 , Usu

Ž . ŽNishida and Tomiya, 1987 , Etna Massenet andPham, 1985; Patella et al., 1990; Di Maio and Patella,

. Ž1994 and Vulcano Island Di Maio et al., 1996,

.1997 volcano-geothermal areas. Space and timevariations of SP and geoelectrical anomalies seemcapable of revealing the most important shallow-sitedelectrical charge polarisation and resistivity varia-tions, most probably caused by rock–water–magmainteractions in active areas, and, more generally, byinvasion and circulation of hot fluids in porous me-dia andror along permeable fracture systems.

Accordingly, MT data are also currently collectedŽin volcanic areas see, e.g., Bablo and Bjornsson,¨ ¨

1978; Hersir et al., 1984; Park and Torres-Verdin,1988; Gough et al., 1989; Wannamaker et al.,

.1989a,b, 1991; Skokan, 1993 . The main objective isthe localisation and electrical parametrisation of themagma chamber, which in the case of the Somma–Vesuvius volcano is still an open question, as theonly data at present available are essentially geo-chemical and petrologic inferences. Newman et al.Ž .1985 explored the theoretical possibility of detect-ing silicic magma chambers by a 3D MT modellinganalysis. They found that the direct detection ofconductive magma bodies is not as straightforwardas one may expect, for it strongly depends on theelongation of the target geometry, on the strike-con-cordance of the electric field vector polarisation andmainly on the existence of electric linking with anoverlying or underlying conductive layer. Along this

Ž .line of thought, Mauriello et al. 1996 recentlyshowed that detectability increases if Cole–Cole re-sistivity frequency-dispersion is hypothesised to existwithin the conductive overlying layer, accounting forinduced polarisation phenomena due to the likelydiffuse presence of hydrothermal alteration products.This approach needs the execution of combined MTand deep dipolar geoelectric soundings in the same

Ž .stations Patella, 1987 . However, the execution of adeep geoelectrical sounding, as it requires dipole-to-

Fig. 1. Geophysical survey location map in the Mount Somma–Vesuvius volcanic area. Dots, full line and triangles indicate theSP mapping circuits, the DG pseudosection profile and the MTsounding stations, respectively.

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 221

dipole spacings as high as 10–15 km, is a very hardtask in a densely inhabited and strongly noise-af-flicted area as is the Vesuvius district.

ŽAs pointed out in a very recent paper Di Maio et.al., 1997 , only the integrated use of different elec-

tric and electromagnetic geophysical methods mayprovide the required resolution, enabling one to cor-rectly identify the space distribution of the electricalparameters, in conjunction with the typical volcanicstructural lineaments and activity manifestations.Furthermore, the combined analysis also allows oneto overcome interpretative ambiguities inherent ineach single method.

In the following sections we first review thegeological lineaments of the Somma–Vesuvius com-plex, then give a detailed description of the resultswe have so far achieved. For the location of the DGpseudosection profile, the SP mapping circuits, andthe MT sounding stations we have planned referenceis made to Fig. 1.

2. Geological and volcanological background

About 2 million years ago, several potash-richeruptive centres developed along the Tyrrhenian

Ž .coast, from northern Latium Vulsini area to Cam-Ž .pania Mount Somma–Vesuvius area , forming the

Žso called Roman Comagmatic Province Washing-.ton, 1906 . This volcanism was interpreted in terms

of either a shoshonitic member of the ApennineŽorogenesis related to converging plates motion Di

.Girolamo, 1978 , or as the alkaline products of theŽ .initial stages of a continental rifting Cundari, 1979 .

The Quaternary potassic volcanoes are located atthe intersection between NW–SE-trending Apenninenormal faults and NE–SW-trending anti-Apenninetraverse tectonics. Such a tectonic arrangement seemsto be related, since Miocene times, to the counter-clockwise torsion of the Apennine Range and to the

Žopening of the Tyrrhenian Sea Abyssal Plain Scan-.done, 1978 . The following Plio-Quaternary inten-

sive vertical tectonics produced the general uplift ofthe central part of the Apennine Range and the

Žsinking of its western side Pescatore and Sgrosso,.1973 .

The Somma–Vesuvius volcanic complex is lo-cated in the Campanian plain that structurally corre-

sponds to a sink filled with recent and present allu-vium, colluvium, beach and piroclastic deposits. Theplain is bordered by Tertiary and Mesozoic lime-stones and dolomites, which, lowered by a system oflistric faults, form the sedimentary basement of the

Ž .volcano Principe et al., 1987 .Ž .Marzocchi et al. 1993 maintain that Mount

Somma–Vesuvius is located on an antithetical faultsystem with NE–SW direction bordering the south-eastern edge of the Acerra depression. This faultsystem would be the response to a NW–SE stretch-ing of the crust related to the backward retreat of theItalian peninsula. The two historical flank eruptionsof Vesuvius in 1794 and 1861, near the town ofTorre del Greco, the seismic profiles at sea per-

Ž .formed by Finetti and Morelli 1974 and the gravityŽ .survey on land by Cassano and La Torre 1987

would provide evidence for the presence of theNE–SW-trending antithetical fault system.

The activity at Somma–Vesuvius spans over thelast 25,000 years, ranging from an effusive be-haviour to very violent explosive eruptions. In partic-ular, three distinctive periods can be identified in thevolcano history: an older period, which precedes thefamous A.D. 79 Pompei plinian eruption, a middleperiod, covering the A.D. 79–1631 interval, and a

Ž . Ž .recent period 1631–1944 Arno et al., 1987 . In the´recent period the volcanic activity showed a cyclicbehaviour, alternating an almost continuous summit

Žactivity, ending with an explosive eruption final.eruption , to short quiet periods that never exceeded

Ž .seven years Scandone et al., 1993 . Since 1944Mount Somma–Vesuvius has remained dormant.

3. Dipolar geoelectrics

Fig. 2 shows the DG apparent resistivity pseudo-section across the N–S profile drawn in Fig. 1. TheDG data were obtained with dipoles of 500 m oflength and the measured dipolar apparent resistivitieswere attributed at a pseudodepth equal to half thespacing between the centres of the emitting andreceiving dipoles, along the median axis through theline joining the two dipoles. Continuous displace-ment of the dipoles along the selected profile ofabout 13 km of length provided a very dense net-

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238222

Fig. 2. DG apparent resistivity pseudosection relative to the N–S profile shown in Fig. 1.

work of about 260 experimental data points in thevertical pseudosection.

Ž .According to Worthington 1984 , a DG pseudo-section can be classified as a first-order tomography,for it can provide in a truly objective way symp-tomatic resistivity anomaly images across the inves-tigated section. Indeed, sophisticated 2D and 3Dinversion procedures applied to many field exampleshave shown that the complex geometries of theinterpreted models are closely reflected in the origi-

Žnal pseudosection images see, e.g., Loke and Barker,1995, 1996; Turberg and Barker, 1996; Dahlin,

.1996 .

3.1. Analysis of the dipolar geoelectric data

The DG tomography of Fig. 2 shows that thegeneral apparent resistivity pattern may represent aroughly horizontal alternation of conductive and re-sistive bodies within the investigated depth range.No inverted V-shaped image corrupting effect, char-acteristic of the dipolar nature of the electrode de-vice, appears in the pseudosection, very likely be-cause of the large dipole length unit used.

Ž .Following Di Maio et al. 1997 , the northernmostand southernmost highly conductive shallow regionsmay be ascribed to sea water-invaded volcano-clastic

sediments. The northern highly resistive body, lo-cated nearly beneath the Somma caldera northernrim, may be very likely associated with a slowlycooled magmatic dike. The next internal, notablyless resistive body closely corresponds to the summitpart of the Vesuvius central edifice. The low resistiv-ity can be explained by admitting that the volcanic

Žchimney and associated breccias Principe et al.,.1987 are there most probably filled with highly

conductive fluids andror pore occluding hydrother-mal alteration particles. Finally, the southern, locallydiscontinuous resistive block may be ascribed to athick sequence composed of ancient submarine lavas,locally intercalated with clayey and muddy beds, andcalcareous conglomerates, and to the underlying car-bonate basement, according to the stratigraphy of the

ŽTrecase-1 geothermal well drilled by AGIP Bal-.ducci et al., 1985 .

4. Self-potential

The SP data were obtained by measuring thepotential difference existing between two groundedelectrodes, which were continuously displaced alongthe circuits reported in Fig. 1. The mutual distancebetween every two consecutive electrode positions

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 223

was constantly taken equal to 100 m. A total of 1250SP drop measurements, distributed over an area ofabout 170 km2, were collected. Fig. 3 shows theelaborated SP final map, redrawn after Di Maio et al.Ž .1997 . Separation between negative and positive SPanomalies is marked by the black zero-valued thickline, which was drawn after shifting the SP zero,originally attributed to an arbitrary reference stationfor the calculation of the initial set of SP mappingvalues, to all points corresponding to the average SPvalue of the initial set. For further details about theSP measuring system, survey technique and data

processing the reader is referred to a previous paperŽ .Di Maio et al., 1996 .

4.1. Analysis of the self-potential map

In the SP map of Fig. 3, a long-wavelength,N–S-oriented double polarisation SP field is veryevident, together with the strong SP horizontal gradi-ents from negative to positive SP values, partly inclose correspondence with the Mount Somma calderanorth, northeastern and southwestern boundaries. AnE–W-elongated strip of higher positive SP values,

Fig. 3. The Mount Somma–Vesuvius SP anomaly map. Contour interval is 50 mV. Green and red isolines are referred to negative andpositive values, respectively. The heavy, roughly circular line delineates the Mount Somma caldera rim and, to the west, part of a formercaldera, related to the Avellino plinian eruption, occurred 3800 years before present.

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238224

crossing the whole central volcanic area, apparentlylaterally limited to the east, seems better representingthe general large-scale behaviour of the SP data inthe map. We can tentatively suggest an eastwardsconfined graben-like structural feature, with a nearlyE–W strike direction, as dominating the whole cen-tral sector of the map. A closer inspection seems toreveal, however, that the central-western portion ofthe supposed graben-like structure would perhaps bebetter described as the coalescence of two roughlycircular structures, of which the westernmost is onlymarginally sketched, while the central one almostcompletely corresponds with the Mount Sommacaldera.

A further SP feature is the SP high nearly encir-cling the Vesuvius cone. Both the SP amplitude andhorizontal gradient of this anomaly are notably lessthan those of the companion plume-shaped negativeanomaly, stretching northwards from the northerncaldera rim towards the foothills of Mount Somma.Referring to the SP synthetic responses calculated by

Ž .Fitterman 1983 across a vertical dike for the two-patch-source model with opposite inward polarisa-tions, this feature would be indicative of a sensiblylower resistivity of the volcanic materials right be-neath the SP high, i.e. within the supposed Vesuviusmain conduit, as already outlined by the DG tomog-raphy of Fig. 2.

Finally, it is worth pointing out the appearance ofa nearly N–S alignment, which is signed to the northby a N–S-elongated narrow weakening of the nega-tive isolines between the two strong SP lows, in thecentral part by the W–E-directed steep gradient ofthe positive isolines and to the south by a similar SPgradient, but from positive to negative values. Again,such a feature could be indicative of a polarised,N–S-directed fracture system, wherein a primaryelectrokinetic fluid flow could be the source of theobserved anomaly.

4.2. Analysis of the self-potential tomographies

In order to elicit from the SP survey map of Fig. 3the most of quantitative information on the depthlocation of the polarisation field sources, we adopteda newest tomographic imaging procedure of the SPanomalies. The rationale and physical and mathemat-ical foundations of the SP tomography have been

Ž .very recently developed by Patella 1997a , follow-ing a preliminary idea of cross-correlation filteringof SP data, used for the first time in volcanology tointerpret the SP time variations along a profile at Mt.

Ž .Etna Sicily, Italy during the last 1989–1993 erup-Ž .tion Di Maio and Patella, 1994 . An improvement to

Ž .the method has been then set out by Patella 1997bto include topographic effects, which are quite un-avoidable in volcanic areas.

Referring to the less arduous 2D tomography,which requires SP measurements performed alongprofiles, in a few steps the procedure is as follows.

Taking a horizontal x-axis, representing the pro-Ž .jection of the elevation profile usz x onto a refer-

Ž .ence horizontal x, y -plane, the measured compo-nent of the natural electric field along the u-profile,

w Ž .xsay E x, z x , is cross-correlated with the realuw Ž . xfunction I xyx , z x yz , representing the SPu q q

synthetic response of an elementary positive chargewith unitary intensity located in the subsurface in

Ž .any generic point with coordinates x , y , z , viz.:q q q

I xyx , z x yzŽ .u q q

xyx q z x yz d zrd xŽ . Ž . Ž .q qs 1Ž .3r222 2xyx qy q z x yzŽ . Ž .½ 5q q q

Then, by a normalisation procedure of the cross-correlation operator based on Schwarz’s inequalityŽ .Papoulis, 1962 , a charge occurrence probability

Ž .function h x , y , z is defined as:q q q

h x , y , zŽ .q q q

q`

sC E x , z x IŽ .Hu u uy`

= xyx , z x yz d x 2Ž . Ž .q q

where the elevation profile-dependent normalisationfactor C is given by:u

q`2C s E x , z x d xŽ .Hu u½

y`

y1r2q`2P I x , z x yz d x 3Ž . Ž .H u q 5

y`

Ž .As y1Fh x , y , z Fq1, the above proce-q q q

dure represents a robust method for searching, again

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 225

in a merely objective way, the most probable locali-sation of the electrical polarisation negative and posi-tive source centres in the subsoil. Finally, by asuitable grid distribution of probability values in thevertical section through any measurement profile, thetomographic image can be drawn for inspecting thepattern of the polarisation source centres in a proba-bilistic sense.

Figs. 4 and 5 show the results relative to two setsof tomographies, each made of three parallel pro-files, respectively, in the W–E and S–N directions.The tomographic representations were evaluated in-dependently of one another. In fact, the charge oc-currence probability values over the section throughany profile were estimated only on the basis of theSP observed values that really entered the cross-cor-relation algorithm, i.e. the SP data set belonging tothe selected profile. Hence, as the estimated proba-bilities account only for that particular profile scan-ning axis, in any vertical cross-line between everytwo perpendicular profiles the related probabilityvalues may not necessarily be tightly matching. Only

Žby the global 3D tomography algorithm Patella,.1997b , whose application to Somma–Vesuvius is

still in progress, we will be able to eliminate anyapparent unconformity.

A very complex picture of electric charge accu-mulations underground appears in both sets of pro-files, which is characterised by an irregularly spacedalternate sequence of shallow to moderately deep,positive and negative SP source centres. The highestoccurrence probabilities are those estimated alongthe WE-2 profile, i.e. the traverse running along themedian axis of the SP positive anomaly strip of Fig.3. The depth of residing of the charged nuclei doesnot exceed about 400 m b.g.l., almost everywhere,except in the SN-1 profile, where the central largepositive nucleus appears to be located in the depthrange from 600 to 2200 m b.g.l.

4.3. Analysis of the self-potential sources

The most reliable physical model for interpretingthe above displayed SP source centres is the onepredicted by Onsager’s theory of coupled flowsŽOnsager, 1931; see, also, Landau and Lifsits, 1978,ˇ

.1986 , originally introduced in geophysics byŽ .Nourbehecht 1963 , then thoroughly examined by

Ž . Ž .Mizutani et al. 1976 , Sill 1983 and Di Maio andŽ .Patella 1991 .

We make here the basic assumption that thedisclosed charge accumulation centres, i.e. thesources of steady electric currents through the con-ductive volcanic rocks, causing the detected SP sig-nals, are generated by fluid mass and heat primaryflows, under the action of pressure and temperaturegradients, respectively. The explanation is given herebelow.

Generally speaking, the electric potential functionU in stationary problems is composed of two parts:the primary field due to active sources and theperturbation field due to induced charge distributions

Žover conductivity discontinuity surfaces, viz. Patella,.1997a :

1 =PJ EP=scndUs dVq dV , 4Ž .H H

4p s r s rV V

where J and E are the conduction current densitycnd

and electric field vectors, respectively, s is theelectric conductivity and r is the variable distancefrom a generic SP observation point to any volumeelement dV of the conductive space.

We discuss now the contribution to the potentialfunction related to the divergence of J . Followingcnd

Ž .Sill 1983 we write the electric current flow coupledequation as:

g g12 13J ss Eq I q q 5Ž .tot tot totž / ž /g g22 33

where J is the total electric current density andtot

I sy=P, with P pressure, and q sy=T , withtot tot

T temperature, are fluid mass and heat primaryflows, respectively. Moreover, the generalised con-ductivities g and g are directly proportional12 13

respectively to the product of rock porosity timesŽdielectric constant times z-potential the first two

.quantities being positive-definite and to Thomson’sŽ .coefficient Di Maio and Patella, 1991 . Finally, g22

and g are rock permeability and thermal conduc-33

tivity, respectively, both positive-definite.Ž .It is worth recalling Sill, 1983 that the first

Ž .right-hand term in Eq. 5 is the conduction electriccurrent density J , whereas the two last termscnd

define the convective electrical current density J ,cnv

i.e. J sJ qJ . In absence of external currenttot cnd cnv

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238226

Fig. 4. SP tomographies along three E–W profiles across the SP anomaly map of Fig. 3.

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 227

Fig. 5. SP tomographies along three N–S profiles across the SP anomaly map of Fig. 3.

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238228

sources the steady state vector J is everywheretot

divergence-free, which implies =PJ s=PJ .cnd cnvŽ .Therefore, using Eq. 5 , after a few simple mathe-

matical steps we definitively get:

g g12 13=PJ s g = P= Pqg = P= Tcnd 22 33ž / ž /g g22 33

g g12 13y =PI q =Pq 6Ž .tot totž / ž /g g22 33

The primary sources of the steady SP field can bethus grouped in two classes. The first class, i.e. thefirst sum in square brackets, shows that SP sourcesdevelop where at least one of the gradients of the

Žgeneralised conductivities the g ’s auto and cross-.coupling coefficients is not perpendicular to the

gradient of the corresponding generalised primaryŽ .potential P or T . The second class, i.e. the second

sum in square brackets, means that SP sources gener-ate wherever there are impressed sources of massandror heat primary flows.

Referring to the first class terms, it is interestingŽ .to note that a positive negative scalar product

Ž .produces a positive negative electric current source.Of course, the sign of the final result depends onhow the two contributions in square brackets dosingularly behave. A similar reasoning can be madein relation to the second class. Indeed, a negativeŽ . Ž .positive divergence generates a positive negativeelectric current source. Again, the final result de-pends on how the two terms algebraically add to-gether.

Turning back to the Vesuvius SP case-history wetry now to interpret the most intriguing electriccharge accumulation, i.e. the positive probability nu-cleus in the 2D tomography SN-1 of Fig. 5, locatedin the depth range 600–2200 m b.g.l. and about 1.5km wide.

We first admit that only the primary sourcesŽ .included in the first integral of Eq. 4 should be

responsible of the observed nucleus, as it seems tooccur within a much larger space of nearly constantlow resistivity, as shown in the nearby DG pseudo-section. We also admit that the two last terms in Eq.Ž .6 , related to the primary flow divergences, may besafely considered ineffective. In fact, in a steady-statecondition the two divergences are different from zero

only in the zone where mass and heat flows arecontinuously generated. A likely hypothesis is toadmit that this zone be located within a magmachamber. As it will be shown later by the analysis ofthe MT data, in the same area of the SP anomalysuch a primary source, supposed to correspond to ahigh-conductivity region, may be found at somegreat-depth in the range 6–9 km b.s.l., below sound-

Ž .ing station MT8 see Fig. 8b .With the assumptions so far made, it follows that

the SP positive source nucleus under discussionwould be essentially created by a variation of thecoupling coefficients within the same volume occu-pied by the SP anomaly. In fact, admitting as itseems logic that both =P and =T are positive down-wards, in order to have the maximum positive valuefor the scalar products in the first square brackets of

Ž . Ž . Ž .Eq. 6 , we need both = g rg and = g rg12 22 13 33

be also positive downwards. We can now suggestour last sheerest speculation about such an intrinsi-cally non-unique interpretation as follows.

ŽConcerning the electrokinetic effect the firstŽ ..scalar product in Eq. 6 , we imagine that at the

depth of about 2200 m b.g.l., just at the bottom ofthe SP positive nucleus, there would exist a laterallylimited, roughly horizontal electrokinetic discontinu-

Ž .ity see Fig. 6 , across which, going upwards to thedepth of about 600 m b.g.l., the system would becharacterised by a steep decrease of the rock bulkporosity, dielectric constant and z-potential. The z-potential is supposed to be positive, which meansthat cations are adsorbed in the fixed layer close to

Ž .the rock pore walls Ward and Fraser, 1967 . Theuppermost block may be also supposed to be later-ally bounded by pervious fracture systems, whichwould confer to the whole block a gradual increaseof permeability upwards.

ŽConcerning the thermoelectric effect the secondŽ ..scalar product in Eq. 6 , we try to interpret Thom-

son’s effect as follows. We imagine that the alteredshallow block be abundantly filled of metallic-typemineral particles within cracks and pores. Theseparticles would drastically reduce the rock bulkporosity. The temperature gradient, positive down-wards, would confer to the free electrons of themetallic particles close to the lower boundary of theblock a major kinetic energy with respect to theelectrons close to the upper boundary. A migration

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 229

Fig. 6. The SP anomaly interpretative sketch model.

of electrons from the hot towards the cold boundaryshould be the initial effect, under the action of the socalled Thomson’s electromotive force. An excesspositive charge thus appears close to the lowerboundary, and, consequently, an electric potentialdifference is set up, which, in absence of currents,counterbalances Thomson’s e.m.f.

At present, without any direct information on themagnitude of the above invoked system parameters,it is very difficult to assess which of the two effectswould really dominate or if both are equally impor-tant. Anyhow, as a matter of fact, such a modelwould represent a limited block of intensively alteredvolcanic breccias and scoriae, filling in the abovehypothesised graben-like structure, roughly beneaththe Vesuvius cone and representing the uppermostbowl-shaped part of a deeply rooted volcanic chim-

Ž .ney Principe et al., 1987 . We deem it very usefulpointing out that the above outlined physical descrip-tion fully conforms to the observed SP positivesource centre in Fig. 5, profile SN-1, to the lateralmost superficial accompanying negative nuclei againvisible in Fig. 5, profile SN-1, closely correspondingto the westernmost fractured portion of the Somma

caldera rim, and to the resistivity low disclosed bythe DG tomography of Fig. 2 in the same centralshallow zone.

Finally, it seems worth mentioning to this respecta very recent experiment of 2D seismic tomographyŽ .Zollo et al., 1996a,b , by which a relatively highvelocity plug, with a P-wave velocity transition from3.0 to 3.8 kmrs, was found underneath the Vesuviuscone in the depth range from about 600 m down to atleast 3000 m b.g.l., with a poorly resolved lateralextension in the range 500–1500 m towards theSomma caldera northwestern rim. It includes in theupper portion the previously discussed SP anomalousbody. The whole high-velocity plug has been inter-preted by the above quoted authors as a body ofmagmatic origin, being either a high concentration ofslowly cooled magmatic dikes or a part of the vol-cano that has suffered an intensive alteration byhigh-temperature hydrothermal fluids.

5. Magnetotellurics

We carried out 12 MT soundings located as inFig. 1. Almost all MT stations were placed on

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238230

volcanic rock outcrops. A wide frequency rangeŽ .0.001–100 Hz was adopted in order to have infor-mation, as complete as possible, about both theshallow and the deep structures. As it concerns thefield data acquisition technique, reference is made to

Ž .Giammetti et al. 1996 .

5.1. MT data processing

After the usual MT data pre-processing, i.e. spikeremoval, window-tapered FFT of electric and mag-netic field records and data convolution by the in-strument transfer function, least-squares impedanceestimate procedures were initially applied. In particu-lar, in order to have a control over the influence ofthe various noise sources, the two estimates includ-ing autopowers of the magnetic and electric field

Žcomponents, respectively, were adopted Sims et al.,.1971; Vozoff, 1972 . If the em components are

affected by random noise, downward biasedimpedance amplitude estimates are obtained by thefirst procedure, while upward biased estimates arethe result of the second approach.

Then, making reference to the phase estimates,which are not affected by random noise and hencehave a unique least-squares determination indepen-dently of the procedure which is used, the secondapproach, based on the autopowers of the electriccomponents, was definitively accepted for the reasongiven here below.

Ž .Boehl et al. 1977 demonstrated theoretically thatfor a 1D earth MT impedance amplitude and phaseare related to each other by the Hilbert transform

Ž .pair. Fischer and Schnegg 1980 presented a methodfor calculating the impedance from the phase for a

Ž .2D rotated response. Weidelt and Kaikkonen 1994proved that impedance amplitude and phase are aHilbert transform pair for the 2D response in the TM

Ž .mode. Yee and Paulson 1988 suggested that thisproperty is valid for a general 3D earth system.Therefore, by Hilbert transforming likely undisturbedexperimental phases, equally undisturbed impedance

Žamplitude estimates are expected Sutarno and Vo-.zoff, 1991 . Thus, by comparing numerical phase-

transformed amplitude estimates with the two aboverecalled standard estimates, one can, in principle,ascertain how noise affected the data, whether on

both or in one or the other estimate. In our case, wededuced that the data in the Vesuvian area werestrongly affected by random noise essentially in themagnetic channels. This result is confirmed by thecomparison with the DG data shown in Fig. 2.

We finally applied the coherence weighting androbust regression estimate method recently proposed

Ž .by Egbert and Livelybrooks 1996 , slightly modi-fied to meet the assumption of robustness of theelectric channels.

5.2. MT data representation

Fig. 7 shows the whole set of TM and TE modeapparent resistivity diagrams. As they did not showany regularity independent of frequency, we con-cluded that no 2D strike direction exists in thesurveyed area.

The interpretation of MT data collected in vol-canic areas is a difficult task and the question ofwhich method is more appropriate to reach a reliablesolution is still a matter of debate. A volcanic envi-ronment is thought, in principle, as a 3D complexstructure, but a 3D modelling would require such alarge set of MT data and a-priori constraints that arigorous solution can hardly be achieved in practice.This is particularly true in those situations as theVesuvian area, where the huge dimensions of thevolcanic apparatus and the large extension of inac-cessible sites do not actually permit to program adense and widely extended grid of measuring sta-tions. Neither the conditions exist for a 2D modellingof a 3D structure as outlined by Wannamaker et al.Ž . Ž .1984 and Ingham 1992 , since our data, as previ-ously said, do not indicate the presence of any

Ž .prevailing strike direction elongated 3D body .Due to such inherent difficulties, we at last de-

cided to compute the rotationally invariant apparentresistivity deduced from the determinant of theimpedance tensor elements, as indicated by many

Žauthors e.g. Ranganayaki, 1984; Ingham, 1988; Park.and Livelybrooks, 1989 . We followed this approach

for all MT soundings, except for sounding MT3,whose TM and TE mode diagrams manifested aclear static shift. For this sounding we selected theTE mode because most conforming to the local DGdata. It is worth mentioning that for a 2D assumption

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 231

Ž . Ž .Fig. 7. Rotated MT apparent resistivity curves for the TM losanges and the TE mode squares .

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238232

the TE mode is, in principle, the least affectedŽpolarisation by local inhomogeneities Wannamaker

.et al., 1989a .

Fig. 8 shows two smooth pseudosections of thedeterminant apparent resistivity, relative to the twoprofiles indicated in Fig. 1, namely the nearly N–S

Ž . Ž .Fig. 8. MT apparent resistivity pseudosections for a the N–S profile and b the W–E profile.

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 233

Ž . Ž .Fig. 8a and W–E Fig. 8b profiles. In drawingthese cross sections we selected the soundings show-ing the highest degree of mutual consistency, mainlyin view of the following approximate 1D Bostickinversion.

The most relevant information contained in bothpseudosections is the presence of a rather uniformconductive central zone extending along the wholeprofiles.

5.3. MT data interpretation

Accounting for the previous explained strong lim-itations in using 2D and 3D MT modelling tech-niques, we made recourse to the 1D Bostick inver-

Žsion Bostick, 1977; Torres-Verdin and Bostick,. Ž .1992 . As is known e.g., Wannamaker et al., 1984 ,

heavy interpretative problems can arise by such asimplified 1D approach, when applied to MT sound-ings essentially located over a laterally bounded very

Ž .conductive shallow layer a few ohm meter , con-fined within the first 1–2 km of depth. In fact, theMT data show a downdipping asymptotic distortionof the apparent resistivity curve, having no relationwith the underlying structures, which consequentlycannot be identified. In our case, the MT pseudosec-tions showed that the central conductive layer is,however, very likely located within a much greaterskin-depth interval, greater than 1–2 km as alsodeducible from the previously shown DG transect inthe same central zone. Hence, the distortion effect isaccordingly notably scaled down, and becomes irrel-evant for the purposes of our study.

Fig. 9 shows the 1D Bostick inversion of bothprofiles of Fig. 8. The conductive zone along theW–E profile has a resistivity as low as 30–50 Vmat a depth around 7 km b.s.l. Of course, owing to theresidual spreading of the experimental data manydifferent models, all compatible with the soundingcurves, can be obtained by the classical direct inter-pretation with horizontal layers. Hence, we cannotexclude that lower resistivities within thin layerscould equally well fit the MT data. Noteworthy isalso the presence of the conductive layer beneathsounding MT9, for which we estimated a lowerresistivity around 20 Vm at a shallower depth ofabout 5 km b.s.l. Conversely, in the N–S profile the

resistivity of the conductive layer tends to increasefrom sounding MT5 both northwards and south-wards. Beneath sounding MT7 it reaches about 400Vm, while below sounding MT2 it reaches about150 Vm. Also the depth increases in both directionsand reaches about 15 km b.s.l. below sounding MT7and about 11 km below sounding MT2.

Most silicic magmas are believed to originate bydifferentiation or partial crustal melting within in-tracrustal magmatic reservoirs, which can be as shal-low as 5–10 km from the earth’s surface. Detectionof this type of reservoir is very important from boththe volcanological and the geothermal point of view,as they are often the source of catastrophic eruptions

Žas well as a long lasting heat source Capuano et al.,.1988 .

By analysis of fluid inclusions in ejected nodulesand other geochemical evidences, Belkin and De

Ž .Vivo 1993 postulated for Vesuvius a crystallisationtrapping zone of a silicate melt, with pressure of1–2.5 kbar and temperature of 1000–12008C, in thedepth range 4–10 km. Moreover, the weight percentŽ .wt% of water within lava phenocrysts related toVesuvian activity preceding A.D. 1631 is in the

Ž .mean around 1.1–1.2 Belkin et al., 1998 .At present we not yet have laboratory estimates of

electrical conductivity of Vesuvian samples, simulat-ing the above physical conditions and with samewater content. Nevertheless, it is well known fromlaboratory measurements carried out on many other

Ž .samples of basaltic rocks Angenheister, 1982 thatin the temperature range 1000–12008C resistivitymay vary even within three orders of magnitude,

Ž .from 0.5 to 300 Vm see Fig. 10a . Moreover, it isŽ .also known e.g., Lebedev and Khitarov, 1964 that

the resistivity of silicic melts is strongly dependentŽ .on the water content see Fig. 10b . A change from 1

to 4 wt% of water content in the melt at 10008Ccauses resistivity to change from 100 to 1 Vm, andan even greater increase of resistivity occurs if tem-perature decreases below 10008C.

Moreover, by a study of the Vesuvian mi-Ž .croearthquake activity Bianco et al. 1998 derived

that hypocentres are clustered around the verticalaxis through the Vesuvius cone with focal depths notexceeding 6 km b.s.l.

Ž .Finally, Zollo et al. 1996b identify a P-to-Sphase conversion in their active seismic profile at a

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238234

Ž . Ž .Fig. 9. MT 1D Bostick inversion for a the N–S profile and b the W–E profile.

depth of about 10 km, estimated by a 2D ray-tracingmodel at the top of a low-velocity zone, which isassumed to represent a melting zone.

Hence, it seems reasonable to admit that a moltenrock system might likely exist with resistivities ofthe order of some tens of ohm meters, that is what

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 235

Ž .Fig. 10. a Electrical conductivity versus temperature clusterŽdiagram for natural and synthetic volcanic melts redrawn and

. Ž .implemented after Angenheister, 1982 . b Electrical conductiv-Žity-versus-water content for a granitic melt redrawn after Burn-

.ham, 1975 .

we obtained from the interpretation of the MTsoundings performed along the W–E profile in the

Ž .Vesuvian area see again Fig. 9b .

6. Concluding remarks

We have presented and discussed the results of anintegrated electric and electromagnetic survey in the

active volcanic area of Mount Somma–Vesuvius.Ž . Ž .Dipolar geoelectrics DG , self-potential SP and

Ž .magnetotellurics MT were used to investigate theshallow and deep regions of the volcanic area.

As it concerns the DG method, an apparent resis-tivity pseudosection along a N–S profile across theVesuvius cone was realised. The main result was thedetection of a largely extended relatively low resis-tivity zone in the central part of the profile, closelyin correspondence to the Somma caldera, includingin the middle the top terminal part of the Vesuvius

Ž .main plumbing system Fig. 2 . The maximum depthof this zone was estimated around 2500 m b.g.l.

The SP data, collected over the whole volcanicarea, clearly showed a W–E-directed wide band ofpositive anomalies, bounded, mainly to the north, bymuch stronger negative anomalies. The band in-cludes the whole Somma caldera and extends wellbeyond the western Vesuvian foothills, reaching the

Ž .Tyrrhenian sea coast-line Fig. 3 . The negative-to-positive amplitude asymmetry was interpreted as theeffect of a more conductive zone beneath the centralband. This interpretation not only conforms to theresult brought to light by the DG tomography, butalso suggests that the volcanic activity most probablydeveloped and migrated along a large W–E-orientedfracture system, part of which would extend underthe sea water. Noteworthy was also the evidence of aroughly N–S-trending narrow fracture system, run-ning almost over the whole survey area and cuttingthe lowest Mount Somma eastern slopes.

The second step in the SP data analysis was theapplication of a new tomographic imaging proce-dure, properly formulated to single out, in a proba-bilistic sense, both position and extension of electriccharge accumulations underground. The main resultof such a new approach was the detection of a largepositive nucleus in the central zone in the depth

Ž .range 600–2200 m b.g.l. Fig. 5 . With respect to thecentral Vesuvius cone it appears sensibly displacedtowards the Tyrrhenian sea, roughly about the west-ern border of the Somma caldera, more preciselybeneath a reconstructed circular portion of a former

Ž .caldera see Fig. 3 , which is believed to be relatedto the Avellino plinian eruption occurred 3800 years

Ž .before present Principe et al., 1987 . The geophysi-cal interpretation of this large positive charge accu-mulation anomaly was made using Onsager’s theory

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238236

of coupled flows, which includes electrokinetic andthermoelectric polarisation phenomena. The majorconclusion, very well accounting also for some re-cent seismic results, is that the conductive centralzone is most probably made up of an intensivelyaltered and mineralised block of cemented volcanicbreccia, characterised also by a strongly reduced

Ž .vuggy porosity Fig. 6 .The MT data, distributed along two perpendicular

Ž .profiles a N–S and a shorter W–E profile , enabledus to obtain the first significant picture of the deepelectrical structure of the volcano. The Bostick inver-sion of the soundings placed along the N–S profilerevealed the existence of a well defined, largelyextended conductive intracrustal layer with a resistiv-ity of some hundreds of ohm meters at most. Themost remarkable evidence is the presence of aperched more conductive zone roughly beneath thecentral Vesuvius apparatus, to which the lowest in-

Ž . Žterpreted resistivity 30–50 m was associated Fig..9a .

Ž .The shorter perpendicular profile Fig. 9b , cross-ing the previous one in correspondence of soundingMT5, showed that the conductive layer is clearlyelongated in the E–W direction and that a scarpfeature seems to delineate its eastern boundary.Moreover, a thickening of the intracrustal conductivelayer appears to characterise the western edge of theprofile. Both such features seem to be closely linkedwith the similar, but much shallower features weoutlined concerning the SP W–E-oriented positiveanomaly band and the disclosed anomalous chargeaccumulation underground, apparently displacedwestwards, with respect to the Vesuvius main con-duit. Altogether these two consistent evidences wouldindicate that the huge shallow and deep structure ofMount Somma–Vesuvius is best featured by a W–E-elongated system, stretching westward towards theTyrrhenian sea.

Finally, like the SP data, which showed a sharpN–S-oriented distortion line, probably associable toa narrow, permeable fracture system, also the MTW–E profile line disclosed a similar effect. In fact,just where the two lines intersect, the MT data put inevidence a strong dislocation of the intracrustal con-ductive layer. To the easternmost edge of the MTline, the conductive body rapidly rises up to a meandepth of about 5 km b.g.l.

Acknowledgements

The authors wish to thank the anonymous refereefor his useful suggestions which helped improve thequality of the text. Study performed with financialaid from the Italian Group of Volcanology of theNational Research Council.

References

Anderson, L.A., Johnson, C.R., 1976. Application of the self-potential method to geothermal exploration in Long Valley,California. J. Geophys. Res. 81, 1527–1532.

Ž .Angenheister, C. Ed. , 1982. Physical Properties of Rocks, Sub-Ž .volume b. In: Hellwege K.H. Editor-in-chief , Landolt-Born-

stein, Numerical Data and Functional Relationships in Scienceand Technology. New Series. Group V: Geophysics and SpaceResearch, Volume 1. Springer, Berlin.

Arno, V., Principe, C., Rosi, M., Santacroce, R., Sbrana, A.,´Sheridan, M.F., 1987. Eruptive history. In: Santacroce, R.Ž . Ž .Ed. , Somma Vesuvius. Quad. Ric. Sci. 114 8 , 53–103.

Bablo, M., Bjornsson, A., 1978. Magnetotelluric investigation of¨ ¨the lower crust and upper mantle beneath Iceland. J. Geophys.45, 1–16.

Balducci, S., Vaselli, M., Verdiani, G., 1985. Exploration well inthe ‘Ottaviano’ permit, Italy; ‘Trecase 1’. In: Straub, A.S.,

Ž .Ungemach, P. Eds. , European Geothermal Update. Proc. 3rdInt. Seminar on the results of EC Geothermal Energy Re-search. Reidel, Dordrecht.

Ballestracci, R., 1982. Self-potential survey near the craters ofŽ .Stromboli volcano Italy . Inference for internal structure and

eruption mechanism. Bull. Volcanol. 45, 349–365.Belkin, H.E., De Vivo, B, 1993. Fluid inclusion studies of ejected

nodules from plinian eruptions of Mt. Somma–Vesuvius. J.Volcanol. Geotherm. Res. 58, 89–100.

Belkin, H.E., De Vivo, B., Torok, K., Webster, J.D., 1998.Pre-eruptive volatile content, melt-inclusion chemistry, and

Žmicrothermometry of interplinian Vesuvius lavas pre- 1631. Ž .A.D. . J. Volcanol. Geotherm. Res. 82, 79–95, this volume .

Bianco, F., Castellano, M., Milano, G., Ventura, G., Vilardo, G.,1998. The Somma–Vesuvius stress field: seismological andmesostructural data. J. Volcanol. Geotherm. Res. 82, 199–218,Ž .this volume .

Boehl, J.E., Bostick, F.X., Jr., Smith, H.W., 1977. An applicationof the Hilbert transform to the magnetotelluric method. Tech.Rep. Electr. Geophys. Res. Lab., Univ. Texas, Austin, TX.

Bostick, F.X., Jr., 1977. A simple and almost exact method of MTŽ .analysis abstract . Workshop on Electrical Methods in

Geothermal Exploration, Snowbird, UT, 1976.Burnham, C.W., 1975. Water and magmas; a mixing model.

Geochim. Cosmochim. Acta 39, 1077–1084.

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238 237

Capuano, P., Gasparini, P., Patella, D., 1988. Contouring magmareservoirs by magnetotellurics. Pap. Int. Workshop of Vol-canic Seismology, Capri, Italy, 1988.

Cassano, E., La Torre, P., 1987. Geophysics. In: Santacroce, R.Ž . Ž .Ed. , Somma Vesuvius. Quad. Ric. Sci. 114 8 , 175–196.

Cundari, A., 1979. Petrogenesis of leucite-bearing lavas in theRoman Volcanic Region, Italy. The Sabatini lavas. Contrib.Mineral. Petrol. 70, 9–21.

Dahlin, T., 1996. 2D resistivity surveying for environmental andengineering applications. First Break 14, 275–283.

Di Girolamo, P., 1978. Geotectonic setting of Miocene–Quaternaryvolcanism in and around the eastern Tyrrhenian sea borderŽ .Italy as deduced from major elements geochemistry. Bull.Volcanol. 41, 3.

Di Maio, R., Patella, D., 1991. Basic theory of electrokineticeffects associated with earthquakes. Boll. Geofis. Teor. Appl.33, 145–154.

Di Maio, R., Patella, D., 1994. Self-potential anomaly generationin volcanic areas. The Mt. Etna case-history. Acta Vulcanol. 4,119–124.

Di Maio, R., Di Sevo, V., Giammetti, S., Patella, D., Piscitelli, S.,Silenziario, C., 1996. Self-potential anomalies in some Italianvolcanic areas. Ann. Geofis. 39, 179–188.

Di Maio, R., Mauriello, P., Patella, D., Petrillo, Z., Piscitelli, S.,Siniscalchi, A., Veneruso, M., 1997. Self-potential, geoelectricand magnetotelluric studies in Italian active volcanic areas.Ann. Geofis. 40, 519–537.

Egbert, G.D., Livelybrooks, D.W., 1996. Single station magne-totelluric impedance estimate: Coherence weighting and re-gression M-estimate. Geophysics 61, 964–970.

Ercan, A., Drahor, M., Atasoy, E., 1986. Natural polarizationstudies at Balcova geothermal field. Geophys. Prospect. 34,475–491.

Finetti, I., Morelli, C., 1974. Esplorazione sismica a riflessione neigolfi di Napoli e Pozzuoli. Boll. Geofis. Teor. Appl. 16,175–222.

Fischer, G., Schnegg, P.-A., 1980. The dispersion relations of themagnetotelluric response and their incidence on the inversionproblem. Geophys. J. R. Astron. Soc. 62, 661–673.

Fitterman, D.V., 1983. Modeling of self-potential anomalies nearvertical dikes. Geophysics 48, 171–180.

Giammetti, S., Patella, D., Siniscalchi, A., Tramacere, A., 1996.The Siena Graben: combined interpretation of DES and MTsoundings. Ann. Geofis. 39, 189–200.

Gough, D.I., McKirdy, D.M., Woods, D.V., Geiger, H., 1989.Conductivity structures and tectonics beneath the EMSLAB

Ž .land array. J. Geophys. Res. 94 B10 , 14099–14110.Hersir, G.P., Bjornsson, A., Pedersen, L.B., 1984. Magnetotelluric¨

survey across the active spreading zone in southwest Iceland.J. Volcanol. Geotherm. Res. 20, 253–266.

Ingham, M.R., 1988. The use of invariant impedances in magne-totelluric interpretation. Geophys. J. 92, 165–169.

Ingham, M.R., 1992. Audiomagnetotelluric soundings on WhiteIsland volcano. J. Volcanol. Geothern. Res. 50, 301–306.

Landau, L.D., Lifsits, E.M., 1978. Fiziceskaja kinetika. Nauka,ˇMoscow.

Landau, L.D., Lifsits, E.M., 1986. Elektrodinamika splosnychˇ ˇsred. Nauka, Moscow.

Lebedev, E.B., Khitarov, N.I., 1964. Dependence of the beginningof melting of granite and the electrical conductivity of its melton high water vapour pressure. Geokhimiya 3, 195–201.

Loke, M.H., Barker, R.D., 1995. Least-squares deconvolution ofapparent resistivity pseudosections. Geophysics 60, 1682–1690.

Loke, M.H., Barker, R.D., 1996. Rapid least-squares inversion ofapparent resistivity pseudosections by a quasi-Newton method.Geophys. Prospect. 44, 131–152.

Marzocchi, W., Scandone, R., Mulargia, F., 1993. The tectonicsetting of Mount Vesuvius and the correlation between itseruptions and the earthquakes of the Southern Apennines. J.Volcanol. Geotherm. Res. 58, 27–41.

Massenet, F., Pham, V.N., 1985. Mapping and surveillance ofactive fissure zones on a volcano by the self-potential method,Etna, Sicily. J. Volcanol. Geotherm. Res. 24, 315–338.

Mauriello, P., Patella, D., Siniscalchi, A., 1996. The magnetotel-luric response over two-dimensional media with resistivityfrequency dispersion. Geophys. Prospect. 44, 789–818.

Mizutani, H., Ishido, T., Yokokura, T., Ohnishi, S., 1976. Elec-trokinetic phenomena associated with earthquakes. Geophys.Res. Lett. 3, 365–368.

Newman, G.A., Wannamaker, P.E., Hohmann, G.W., 1985. Onthe detectability of crustal magma chambers using the magne-totelluric method. Geophysics 50, 1136–1143.

Nishida, Y., Tomiya, H., 1987. Self-potential studies in volcanicŽ .areas 1 : Usu volcano. J. Fac. Sci. Hokkaido Univ. Ser. VII 8,

173–190.Nourbehecht, B., 1963. Irreversible thermodynamic effects in

inhomogeneous media and their application in certain geoelec-tric problems. Ph.D. Thesis, M.I.T., Cambridge.

Onsager, L., 1931. Reciprocal relations in irreversible processes I.Phys. Rev. 37, 405–426.

Papoulis, A., 1962. The Fourier Integral and its Applications.McGraw-Hill, New York, NY.

Park, S.K., Livelybrooks, D.W., 1989. Quantitative interpretationof rotationally invariant parameters in magnetotellurics. Geo-physics 54, 1483–1490.

Park, S.K., Torres-Verdin, C., 1988. A systematic approach to theinterpretation of magnetotelluric data in volcanic environmentswith applications to the quest for magma in Long Valley,

Ž .California. J. Geophys. Res. 93 B11 , 13265–13283.Patella, D., 1987. Tutorial: Interpretation of magnetotelluric mea-

surements over an electrically dispersive one-dimensionalearth. Geophys. Prospect. 35, 1–11.

Patella, D., 1997a. Introduction to ground surface self-potentialtomography. Geophys. Prospect. 45, 653–681.

Patella, D., 1997b. Self-potential global tomography includingtopographic effects. Geophys. Prospect. 45, 843–863.

Patella, D., Tramacere, A., Di Maio, R., 1990. Self-potentialŽ .anomalies. In: Barberi, F., Bertagnini, A., Landi, P. Eds. , Mt.

Etna: the 1989 Eruption. Giardini, Pisa, pp. 58–61.Pescatore, T.S., Sgrosso, I., 1973. I rapporti tra la piattaforma

campano-lucana e la piattaforma abruzzese-campana nel caser-tano. Boll. Soc. Geol. Ital. 92, 925–938.

Principe, C., Rosi, M., Santacroce, R., Sbrana, A., 1987. Explana-Ž .tory notes to the geological map. In: Santacroce, R. Ed. ,

Ž .Somma Vesuvius. Quad. Ric. Sci. 114 8 , 11–51.

( )R. Di Maio et al.rJournal of Volcanology and Geothermal Research 82 1998 219–238238

Ranganayaki, R.P., 1984. An interpretive analysis of magnetotel-luric data. Geophysics 49, 1730–1748.

Scandone, P., 1978. Origin of the Tyrrhenian sea and Calabrianarc. Boll. Soc. Geol. Ital. 98, 27–34.

Scandone, R., Giacomelli, L., Gasparini, P., 1993. Mount Vesu-vius: 2000 years of volcanological observations. J. Volcanol.Geotherm. Res. 58, 5–25.

Sill, W.R., 1983. Self-potential modeling from primary flows.Geophysics 48, 76–86.

Sims, W.E., Bostick, F.X. Jr., Smith, H.W., 1971. The estimationof magnetotelluric impedance tensor elements from measureddata. Geophysics 36, 938–942.

Skokan, C.K., 1993. Overview of electromagnetic methods ap-plied in active volcanic areas of western United States. J.Volcanol. Geotherm. Res. 56, 309–318.

Sutarno, D., Vozoff, K., 1991. Phase-smoothed robust M-estima-tion of magnetotelluric impedance function. Geophysics 56,1999–2007.

Torres-Verdin, C., Bostick, F.X. Jr., 1992. Principles of spatialsurface electric field filtering in magnetotellurics: Electromag-

Ž .netic array profiling EMAP . Geophysics 57, 603–622.Turberg, P., Barker, R., 1996. Joint application of radio-magneto-

telluric and electrical imaging surveys in complex subsurfaceenvironments. First Break 14, 105–112.

Vozoff, K., 1972. The magnetotelluric method in the explorationof sedimentary basins. Geophysics 37, 98–141.

Wannamaker, P.E., Hohmann, G.W., Ward, S.H., 1984. Magne-totelluric responses of three-dimensional bodies in layeredearth. Geophysics 49, 1517–1533.

Wannamaker, P.E., Booker, J.R., Filloux, J.H., Jones, A.G., Chave,A.D., Tarits, P., Waff, H.S., Egberk, G.D., Young, C.T.,Stodt, J.A., Martinez, M., Law, K.L., Yukutake, T., Segawa,J.S., Write, A., Green, A.W. Jr., 1989a. Magnetotelluric obser-vations across the Juan de Fuca subduction system in the

Ž .EMSLAB project. J. Geophys. Res. 94 B10 , 14111–14125.Wannamaker, P.E., Booker, J.R., Jones, A.G., Chave, A.D., Fil-

loux, J.H., Waff, H.S., Law, L.K., 1989b. Resistivity crosssection through the Juan de Fuca subduction system and its

Ž .tectonic implications. J. Geophys. Res. 94 B10 , 14127–14144.

Wannamaker, P.E., Wright, P.M., Zi-xing, Z., Xing-bin, L., Jing-xiang, Z., 1991. Magnetotelluric transect of Long Valleycaldera: Resistivity cross-section, structural implications, andthe limits of a 2-D analysis. Geophysics 56, 926–940.

Ward, S.H., Fraser, D.C., 1967. Electrical methods, Part B: Con-duction of electricity in rocks. In: Hansen, D.A., MacDougall,

Ž .R.E., Rogers, G.R., Sumner, J.S., Ward S.H. Eds. , MiningGeophysics, Volume II: Theory. SEG, Tulsa, OK.

Washington, H.S., 1906. The Roman Comagmatic Region.Carnegie Inst. Washington 57, 1–199.

Weidelt, P., Kaikkonen, P., 1994. Local 1-D interpretation ofmagnetotelluric B-polarization impedances. Geophys. J. Int.117, 733–748.

Worthington, M.H., 1984. An introduction to geophysical tomog-raphy. First Break 2, 20–26.

Yee, E., Paulson, K.V., 1988. Concerning dispersion relations forthe magnetotelluric impedance tensor. Geophys. J. 95, 549–559.

Zablocki, C.J., 1976. Mapping thermal anomalies on an activevolcano by the self-potential method, Kilauea, Hawaii. Proc.2nd UN Symposium on the Development and Use of Geother-mal Resources, San Francisco, 1975, 2, 1299–1309.

Zohdy, A.A.R., Anderson, L.A., Mufler, L.J.P., 1973. Resistivity,self-potential and induced-polarization surveys of a vapour-dominated geothermal system. Geophysics 38, 1130–1144.

Zollo, A., Gasparini, P., Biella, G., De Franco, R., Buonocore, B.,Mirabile, L., De Natale, G., Milano, G., Pingue, F., Vilardo,G., Bruno, P.P., De Matteis, R., Le Meur, H., Iannaccone, G.,Deschamps, A., Virieux, J., Nardi, A., Frepoli, A., Hunstad, I.,Guerra, I., 1996a. 2D seismic tomography of Somma-Vesuvius.Description of the experiment and preliminary results. Ann.Geofis. 39, 471–486.

Zollo, A., Gasparini, P., Virieux, J., Le Meur, H., De Natale, G.,Biella, G., Boschi, E., Capuano, P., De Franco, R.,Dell’Aversana, P., De Matteis, R., Guerra, I., Iannaccone, G.,Mirabile, L., Vilardo, G., 1996b. Seismic evidence for alow-velocity zone in the upper crust beneath Mount Vesuvius.Science 274, 592–594.