Self-potential Survey near the Craters of Stromboli Volcano (Italy). Inference for Internal Structure and Eruption Mechanism R. BALLESTRACCI

Laboratoire de G$ophysique Appliqu~e. Universit~ de Toulon La Garde, 83130 - France

ABSTRACT The presence of self potentials on the upper part of Stromboli Volcano is associated with the existence of convection cells of gases and condensed liquids whose rising parts are situated above the hot zones. A model of a convection cell fed by a hot fluids through its lower middle part was calculated and applied to the various structures of the upper part of the volcano. This model which takes account of a process which is probably fairly c o m m o n in volcanic system, gives a clear explanation of the various potential zones. Although other interpretations could be made, the fact that the positive and negative potentials are of the same importance, the limited size of the convection zone and the presence of gas under pressure beneath permeable grounds lead us to conclude that the phenomenological cause of these potentials is probably the mechanism of electro-filtration. The resets of self-potential measurements showed very close correlation with those we obtained previously by magnetotelluric profiling. These two geophysical methods were very efficientin the precise delimitation of hot zones which give rise to resistivityand potential anomalies. The depth of those zones was estimated from the aspects of the potential anomalies and from the geological environment.

The two methods were furthermore powerful tools in the investigation of volcanological processes and in particular the <~Strombolian ~ rythrnic eruption. The geometrical disposition of the feeding channels involves the obstruction of the path of the gases by the lava. Generally, the force of pressure exerted by the gas increases quickly and in consequence the plug of lava is ejected and shattered. W h e n the gas flow lessens, the forces of viscosity increase Bull. Volcanol., Vol. 45-4, 1982

quicker than the forces of pressure, the discharging outlets are then obstructed and thus there is great hazard that the accumulated energy will be liberated in a very violent fashion. In this case there will be a paroxysmal-type eruption. This mechanism is the consequence of a recent accident in the volcano evolution: the landslip of a segment of the cone, probably associated with the subsidence of the central caldera; the lava spread over the collapsed sector and the rhythmic eruptions described above constructed the present craters.

INTRODUCTION The Stromboli is a strato-volcano situated 75 km North of Sicily (Italy). It rises 3000 meters from the Tyrrhenian seabed and 924 meters above sea level. It is one of two active volcanoes (StromboliVulcano) of the Aeolian arc, whose existence is linked with the presence of a subduction zone situated between the African and European plates (BARBERI et al., 1974). The craters are situated on an elongated terrace 250 meters long at an altitude of about 760 meters (Fig. 1). The two snmrnits (918 m and 924 m) are the remains of the edge of the old crater. The ejected products and lava flows spread over an unstable sloping area stretching from the active craters to the sea (Sciara del Fuoco). Stromboli is considered one of the most active volcanoes in the world. Since the ftrst observations by Aristotle, and

350

R. BALLESTRACCI

FIG. 1 -- T h e central p a r t of Stromboli Volcano. T h e s u m m i t s are r e m n a n t s of t h e upper p a r t of the paleo-volcano. P r e s e n t activity occurs on the terrace where craters 1, 2, 3-4 are situated. T h e various magnetotelluric traverses have b e e n m a r k e d in. T h e self potentials were m e a s u r e d in the pit b e t w e e n t h e craters a n d the overhanging cliff.

probably long before, the craters have been in a state of permanent activity characterized by violent projections of gas carrying incandescent scoria and ash during a period of a few second only (Fig. 2), but repeated several times an hour. The term <~ has been given to this phenomenon, which has been observed in the course of eruptions at m a n y volcanoes. Apart from this fairly moderate rhythmic process, paroxysmal accidents occur during which very violent explosions, lava flows, avalanches of ash and even tsunamis have been observed.

The number, shape and lay-out of the craters change slowly and continuously when activity is normal, or suddenly during paroxysmal episodes. The volcanic manifestations as a whole have been the subject of numerous studies, the most recent concerning the study of gases (TAZmF~, 1970; and LE GUERN, 1972); the statistics of eruption (SETTLE et al., 1974); the mechanism and dynamics of Strombolian activity (BLACKBURN et al., 1976); the paroxysmal eruption of 1975 (CAPALDI et al., 1978); the energy balance of the volcano (Mc

SELF-POTENTIAL SURVEY NEAR THE CRATERS OF STROMBOLI VOLCANO (ITALY)

GETCHIN et al., 1979); a set of magnetotelluric profiles of the upper part of the volcano (BAH.ESTRACCI, 1982). This last study showed the presence of feeding channels situated just below the surface, connecting the active central part of the volcano to the various craters, and formed a working basis for other geophysical investigations. HALWACHS (1979) was the first to point out the existence of large potential differences on the surface around the crater terrace. A map of self-potentials and ground temperatures was therefore set up in September 1980 with the eim of cenfn~ning results obtained by magnetotelluric profiling and to define the physical phenomena which cause Strombolian activity and its paroxysmal manifestations.

FIO. 2 -- (( Strombolian>> eruption: violent projections of incandescent gas scoria and ashes from a discharging outlet situated in crater 1. This type of eruption occur several time an hour and is characteristic of Stromboli volcano. The projections can be several hundred metres high.

351

The self-potentials method has been applied, successfully in general, to the detection of anomalies in geothermal systems (ZOHD¥ et al., 1973; ANDERSON, 1976 and CORWIN, 1979) but the only measurements made on an active volcano were those of ZABLOCKI (1976) on the Kilauea. In all these experiments the results obtained show clearly compatibility between the potential anomalies and thermic sources whose existence has been shown by other methods. T h e main difficulty which remains is to explain the ionic or electronic mechanism which cause these natural potentials, t h a t is, to interpret these anomalies in physical or physico-chemical terms.

EVOLUTION AND RECENT ACTIVITY OF STROMBOLI VOLCANO The present craters of Stromboli volcano are distant from the axis of the initial paleo-volcano (Fig. 3). The paleovolcano is characterized by steep slope which can be attributed to a large quantity of intrusive sills (RITTMANN, 1963) whose number and thickness increase near the paleo-vent and which give rise to the swelling of the upper part of the volcano. The phenomenon which has dominated the recent evolution of the volcano is the landslip along a fault of a segment of cone which has profoundly altered the form and function of this volcano (Fig. 3 b). The landslip, aided by the steep slope, was probably associated with the intrusion of a sill along a plane of stratification of the strato-volcano, thus acting as a lubricant. It is likely that this phenomenon was accompagnied by the collapse of the summit caldera, deprived of the support formed by the lava evacuated during the landslip of part of the paleo-volcano. Thus the subsidence of a segment of the cone forming the paleo-volcano has the effect of forming craters on the subsided ground outside the summit caldera (Fig. 3 c). The collapse of rock masses between the caldera and the outer part of the volcano have modified the top part of the volcano to its present landscape. On Fig. 14 we

352

a, BALLESTRACCI

compare the particular evolution of the Stromboli volcano with the more symmetric evolution of a strato-volcano having a gentler slope (Fig. 3, d. e. f.), which constructs its new crater within the summit caldera (as in the case of the Vesuvius cone, constructed within the Monte Somma caldera following the plinian eruption of 79). The present rhythmic projections of scoria are building a group of 3 cones of pyroclastics around the craters. T h e morphology of this group of cones evolves gradually during normal strombolian activity. On the other hand, the number and behaviour of the discharging outlets inside each crater changes rapidly. Thus, in April 1979 and September 1980, the overall volcanic activity was as follows (using the same numbering of craters as in the previous work: (Fig. 1 and 4). APRm 1979 Crater 1

Crater 2

1 discarging outlet: (2): brief, powerful explosion incandescent gases.

emitl2ng

Crater 3-4

2 independent discharging outlets: (3): eruption of gas and a little fragmented lava. (4): constant emission of steam. SEPTEMBER

1980

Crater 1

2 independent discharging outlets: eruption of scoria, ash and gases. Crater 2

3 discharging outlets: 1 outlet: brief, powerful explosion emitting gas and black smoke, 2 outlets: eruption of scoria, ash and gases. Crater 3-4

2 discarging outlet: (1): eruption of scoria, ash and gases.

d

e

FIG. 3 -- Diagrammatical representation ot the two phenomena (3 a, b, c) which have marked the recent evolution of StromboH volcano: landship of a segment of the cone and collapse of the sumrn]tal caldera. Comparison with the transformation of a volcano whose slope is less inclined (3 d, e, f).

2 synchronized discharging outlets and 1 independant outlet: eruption of scoria, lava and gases. All the eruptions happen in a more or less rhythmic fashion, with the interval between eruptions from any one discharging outlet changing rapidly also. PREVIOUS AUDIOMAGNETOTELLURIC PROFILES Audiomagnetotelluric measurements in the 8 Hz-170 Hz frequency range were taken on the summit of the volcano in April and July 1979 (BAt.t.~STRACCI, 1982). Because of the large electrical contrast between very conductive hot m a g m a and the cooler wallrocks, mainly tufts and basaltic flows with low conductivities, detection of zones of low resistivity was easy. Principal results are clearly apparent on the Fig. 4 where are plotted equiresistivity

SELF-POTENTIAL S U R V E Y N E A R T H E C R A T E R S O F S T R O M B O L I V O L C A N O (ITALY)

section of these conductive bodies, it seems that the elongated zones between the two eastern low resistive bodies, Z 1 and Z~. and the craters are deepening from the southeast to the northwest. This geometry supports a mechanism of volcanic products feeding in which the gas

curves (10, 20 ~ m) measured at 8, 17, 37, 80 Hz along the 7 profiles A 1Bl, A 2 B2, As B3, CD, EF, GH, JK. These apparent resistivity anomaly maps show some low resistivity zones in the old volcanic products of the paleo-volcano. Although it is quite impossible to ascertain the cross 80HZ

J

~ /

37 H z

I !

_

/

~--~

]"--~f

~

tl c.,

(

~t "

~..

, , - - , ....... , ......

.+,

- -

~

..H

"

:d ~" / ......... ' i ~

,~/'

\

_..~'~~

I 8HZ

"

~

/

/ £.+j

i

71~.+,.'-h-

="

K...

I, +"

.,'x /

~

I

I

I,+. •

+t+:?I~N4) : %,~..~4."2 vV,"

.,~'~.~h,y..~:.,+, ~

~_1-~ ~ , y~ /"'--.-,,...-K- - " / "

-.... X

f

% ' <..., \ i ; Y

t/ ,/

z,

~."

~..j .-t~":" ,D

(+~Y~!"

~ ' i : . )~c.x.~-".,,,"/~7/-"

........... +......~ "~"~"~

353

....

~ +.......

\\#~

+. . . . . ~.

Q........ : ~ , X _

. . . . . . . +_: . . . . .

i - - ~ , \ z,

FIG. 4 -- Anomaly maps at 80, 37, 17 8 H z plotted from audiomagnetotelluric profiling, i0 • m and 20 fJm equiresistivitylines make possible a quick identificationof low resistivityzones (on the dotted area the measured resistivity is under 10 fJm).

354

a. BALLESTRACC]

and lava issued from dikes situated in the old volcanics are led to the crater zone by deepening channels. Strombolian activity (well collimated jets of incandescent scoria, gases and lapilli lasting no more than few seconds and occurring at more or less regular interval several times per hour) is explained by the geometrical layout of the feeding channel. The flow of lava, probably arriving in batches, temporarily blocks part of the path of the gases. The plug of lava cools and the increase in pressure of the gases, once the force becomes greater than the force of viscosity, ejects the mass of lava which shatters on leaving the vent, giving rise to the well-known effect of hot spatters and scoria which characterises strombolian activity.

MEASUREMENT OF SELF-POTENTIALS AND TEMPERATURES of about 40 cm below the surface. The measurements of self-potentials on the ground is made quite simply with a fixed reference electrode P and a second electrode which is moved about over the area under examination. The map of self

potentials is defined relative to the potential of P which is arbitrary. The only equipment required is two impolarisable electrodes, a portable high impedence voltmetre and an insulated conducting cable. Figure 5 shows, in an approximate manner the whole of the geographical structure and the level curves. The mobile electrode was displaced from P along 8 traverses approximatively parallel to each other, all of them being situated in the pit between the craters and the steep cliff which overhangs them. Measurements along the line of craters 1 and 3-4 were avoided because of danger there was too great. For each traverse, measurements were taken every 10 metres and after each traverse a closed circuit measurement was made to ensure that the sources of error (drift with time electrode polarisation) remained low. The closure tension was in all cases less than 10 inV. Electrical contact with the ground was very good, since the ground was damp a few centimeters below the surface. For each potential measurement, an associated temperature reading was made with a digital thermometer coupled to a thermometric probe, with a precision of one-tenth of a degree at an average depth of about 40 cm below the surface.

FIG. 5 -- Approximate level curves in the pit and situation of the various ptoffies along which self potentials were measured from point P.

SELF-POTENTIAL SURVEY NEAR THE CRATERS OF STROMBOLI VOLCANO (ITALY)

RESULTS Figure 6 shows equi-potential curves drawn from values measured along the various traverses shown on Fig. 5. The potential being measured relative to the arbitrary potential of P, all the equi-potential ~ r v e s were off-set in order to give a zero value to the potential of the vulcano as a whole, that is to all points distant from potential-generative processes. Thus the reference electrode had a potential of 280 inV. The main features which appear on this map are: 1. Remarkably high potential gradients: they can reach 500 mV over 100 m, ie, of the same order as those obtained at Kilauea (700 mV) by ZABLOCKY (1976), and greater than those measured on most geothermal areas. 2. Positive potential zones which are very clearly defined (Fig. 6) and which correlate very well with results obtained by magnetotelluric measurements. One zone A + is associated with the channel connecting Z1 to crater 1. A second zone B + is situated above the channel connecting zone Z2 to crater 3-4. A third zone R + covers the group of the three craters,

355

although measurements were taken only on its south-east border. A fourth zone S + is connected with the presence of central dykes. 3. Negative potential zones ( A - B - ) which correlate with the coldest zones of this part of the volcano as shown in Fig. 6. The interpretation of these zones may be made fairly easily when one considers the geological structure of the volcano. The whole active area near the summit of Stromboli is geologically formed of 2 parts

(Fig. 13):

-- a stratified area of tufts, scoria and basalt flows emitted by the paleo-volcano, - - t h e craters and the pit constructed on a collapsed zone of the old volcano and formed of recent pyroclastics whose permeability to gases is much higher than that of the previous medium. The results of the self-potential measurements suggest that the positive zones are associated with gases rising by convection. The quantities of gas emitted by Stromboll are indeed very large: the ratio of the mass of gases to the mass of lava emitted was measured indirectly from two eruptions by CHOUET et al., (1974): they found 2.4 and 15.8. Gases, whose pressure changes at the rhythm of the eruptions,

FIG. 6 -- Contour map of the self-potential distribution drawn from potential values measured along previously defined traverses. Zero potential is the potential of the points distant from the potential generating process.

356

R. BALLESTRACCI

cannot escape from dikes or internal channels except for a small fraction which escapes by convection over the whole of the paleo-volcano cliff (zone S + ) with a maximum at the summit, and for another part, which also gives rise to convection at the interface between the two media (zone A-~ and B~). The bulk of the gases escapes during eruptions, but also by convection around the three craters (zone R +). These phenomena are helped by the rhythmical increase in pressure inside the feeding channels. The channels feeding crater 2 (Fig. 4) is very apparent on the magnetorelluric profiles but cannot be observed on the self-potential map. From this we conclude that this channel is embedded in the low-permeability material of the old volcano. The map of measurements of 40 cm deep temperature (Fig. 7) shows clearly a temperature rise (the reading is over 80 °) directly above the upward-moving gases (zone A + and B +) and in the crater zone (R-F) whereas S~- zone remains cold. This map gives a better insight into the nature of the substratrum in the pit, and in particular those parts composed of cold collapsed ground which are less permeable to gases and therefore cooler.

.

1 0 0

CONVECTION OF VOLCANIC GASES IN POROUS MEDIA The presence of self-potentials is linked with the convection of gases in porous media composed of very permeable pyroclastics or in more compact media composed of basalt, scoria and tufts from the paleo-volcano. In this chapter we apply the DONALDSON (1962) model referring to the dynamics of a fluid in a permeable medium, which we adapt to the precise case, very common in volcanic systems, where the convection cell is feed from below over a limited surface by very hot gas. This gas, of density p, of specific heat e, of kinematic viscosity v, saturates an horizontal layer of homogenous porous material of thickness d and permeability K. Using the simplifyins approximations of BOUSS~ESQ, the equations governing the fluid behaviour are as follows: continuity equation • (p~)

ffi 0 ,

Darcy's law:

P--p ~ + ( v / K ) p ~ = O,

m

FIG. 7 -- Map of temperatures measured on the ground. The collapsed zones directly above which temperature is lower are clearly apparent.

SELF-POTENTIAL SURVEY NEAR THE CRATERS OF STROMBOLI VOLCANO (ITALY)

energy conservation equation:

cp~.

V T~Km

357

0=0

0=0

V2T,

where Km is the thermal conductivity of the saturated medium, ~ the volume flow rate vector, P the pressure, T the temperature, ~ the gravity vector and p ~ p [(1 -- a (T -- To)] were a is the coefficient of volumetric expansion of the fluid. T h e s e equations m a y be written in nondimensional form (WOODING, 1957) d

(Km/Kv) K Po = ( l / g ) ~, 0

'~ I t =

T-To

0=1

T1-- To '

(Km/Kv)w

Po -- I

where Kv is thermal conductivity of water, K thermal diffusivity of gas, Po the density of gas at temperature To at the surface, T1 the arrival temperature of gas. T h e first 3 fundamental equations reduce as follows • ~=0,

VH+nO~'+~=

FiG. 8 -- Temperature and flow patterns for n = 10z. The convection cell is fed via its lower part by very hot gas over a limited surface (0 m 1). We assumed that the gas was discharged on the surface and in the lower part (0 = 0,1 and 0 = 0).

Considering the flux in a vertical plane

(xz) we can write:

O,

where ~, ~', and H are volume flow vector, unit gravity vector and pressure expressed in a nonMimensional m a n n e r and:

n

0=0

ka(T1-- ~ ) g d (Km/Kv) VK

is Rayleigh's number. These equations can also be written:

which satisfies the continuity equation and where ¥ it the scalar stream function and j the unit vector perpendicular to xz: 020 020 0 x 2 ~- 0 z 2

0¥ 0x

a2¥

d8

02¥ _

Ox 2 + - ' f f T z2 - n

O0 0¥ 0z + 0z

O0 0x

Ox "

A squared network with m e s h of length h is applied to the convection cell, the n('~O)

xg'+X~

x ~=0

temperature and stream functions at each grid point are then related to these same values at neighbouring grid point by the following (Fox, 1962):

358

R. BALLESTRACCI

4 0 (xm, z,~) =

+ 1--~h

1+-~-

~.

8 ~ ~,,, 8(x~_1, zn)+ 1 - - ~

+

1 + --~-

m,~

~,n 8(Xm, Xn+3

,

4 lp (xm, z~) = lp (x m + 1 , zn) + lp (xm-1, z~) + ~ (x m , Z n +3

+ ¥ (Xm , Zn_l) -- h 2 n (\ 8Sx0 1/m,~+ O(h~). If we e s t i m a t e a starting value of s t r e a m function a n d t e m p e r a t u r e , t h u s 8¥ 8¥ enabling us to calculate ~ and ~ at each grid point, it is then poss~le to calculate a new value for 8, to estimate 88 -at each grid point and to obtain a 8x new value for the stream function. If we repeat this process several times, it converges r a p i d l y for sufficienty large values of n. This mathematical formulation is a p p l i e d to a convection cell fed over a small surface a r e a (1/10 th of t h e cell) of its lower part. S t e a m is p a r t i a l l y discharged on the surface in t h e neighbourhood of t h e cell axis, which is also t h e h o t t e s t part. T h e cell is a s s u m e d to be s a t u r a t e d with steam, b u t in t h e neighbourhood of the i s o t h e r m 8 = 0,1, s t e a m condenses a n d the w a t e r is forced b y its own weight towards t h e b o t t o m of the cell. To avoid applying the p r e c e d i n g calculation to b o t h fluids, we a s s u m e d t h a t t h e r e m a i n d e r of s t e a m was discharged at t h e lower limit of the convection cell (pressure is a s s u m e d to b e equal to a t m o s p h e r i c p r e s s u r e a t all points). Conditions a t the limit are thus as follows: 8 = 1 on t h e lower axial feeding zone,

8 = 0.1 on t h e u p p e r axial zone of gas discharge ( t e m p e r a t u r e s m e a s u r e d on the surface of t h e ground are of t h e order of T1-To/100). 8 ~ 0 a t all o t h e r points. T h e t e m p e r a t u r e a n d flow p a t t e r n s were e s t a b l i s h e d for n ~ 102 a n d n ~ 104 O=OJ

(3= 0

1

e=1

'

e=o

FIG. 9 -- Temperature and flow patterns for n ~ 104. Contrary to the previous example, the flux is entirely controlled by the chargedischarge process.

SELF-POTENTIAL SURVEY NEAR THE CRATERS OF STROMBOLI VOLCANO (ITALY)

(Fig. 8 a n d 9). I n fact, values for n are probably still higher. T a k i n g a m e a n value of a = 10 -~, T1T o = 1 0 s d o, g = 1 0 s cm sec -2, d ~ 1 0 0 0 cm and t h e r m o d y n a m i c values for s t e a m at 500 ° (COLT.mR, 1972), as v = 5.10 -9 cm~] sec a n d K = 10 -~ m2/sec, we obtain values for Rayleigh's n u m b e r of n = 4.106 a n d 4.104 (permeability value of k = 10-14 cm 2 a n d k = 10 - ~ cm 2 are attributed respectively to r e c e n t pyroclastics a n d to older structures of the paleo-volcano). T h e t e m p e r a t u r e a n d flow p a t t e r n (non dimensional) established for n = 10 ~-show t h a t p a r t of the convection cell fluid is in a closed circuit, whereas for n = 104 the flux is entirely controlled b y the process of charge a n d discharge (there is n o t closed convection cell).

359

INTERPRETATION OF MEASUREMENTS T h e m a i n result suggested by Fig. 8 a n d 9 is t h a t the zones where the self potential m e a s u r e m e n t is positive are zones where there is gas discharge at the surface (flux lines cut across the ground surface). Zones where the potential is negative correspond with parts of the convection cell where the s t e a m condenses or where the condensed water is forced b y its own weight downwards. I n Fig. 10, we show schematically the various positive a n d negative p o t e n t i a l zones. T h e overall disposition of the zones is caused b y the convection of gases in several n e a r b y cells separated b y collapsed sectors of the paleo-volcano which iso-

100

Ill

I

FIG. 10 -- The various zones of potential. They can be explained by the interferences of 4 convection cells. Two of them (A and B) are fed by gases escaping by convection from the channels feeding craters 1 and 3-4. The descending parts of the convection cells being directly below the higher parts of the collapsed structures. The other two are associated with the convection of gases above the dikes (S+) or in the craters (R+). The thick lines show the boundaries of the collapsed zone. Double circles indicate the most intense convection in the ascendant zone.

360

R. BALLESTRACCI

late them from each other in a more or less impervious fashion. These structures are at present hidden beneath a uniform covering of pyroclastic, but are easily identifiable on the temperature pattern. The descending zones of the convection cells, thus are situated directly above those structures which are less permeable and colder than the more recently ejected products. The zone A-t-, an ascendant axial zone of the convection cell relative to channel feeding the crater 1, is associated with three zones A--. Similarly zone Bqis coupled with two descending zone B--. Zones R~- and S-{- which related to the craters a n d to the central dikes respectively are positive potential areas. The zones of negative potential, which should be associated with them, cannot appear separately as is shown by Fig. 11 and 12 which were set up diagrammatically from the results of Fig. 8 and 9. The positive potentials are higher in the craters zone (R-t-) than in the central dikes zone (S+), which is a consequence of the greater permeability of the recent pyroclastics. The low value (50 mV) of the potential (S+) zone allows to show up the zones A - and B-- which overlap it.

FIG. 11 -- Diagrammatical representation of temperature and flow patterns in the craters zone. The extremely permeable material situated in the neighbourhood of the discharging outlets makes this zone a very active convective cell.

FIQ. 12 -- DiagrAmmatical representation of temperature and flow patterns in the top part of the volcano situated above the dikes where the lava rises. The lower permeability of the old structures makes this part of the volcano a less active convective cell (except for directly above the dikes).

On Fig 13 we show the traversing of the speculative lava feeding channel through the collapsed structure of the paleovolcano. The channel detected on the magnetotelluric profiles, as feeding crater 2 does not give rise to gas convection, it is likely that this channel emerges into the recent pyroclastics at a point situated in zone R A-, too close to the craters to be detected by our potential measurements. The resistivity and self-potential measuremens therefore concur in showing the presence of feeding channels which bring the lava from the higher part of the central dykes to the discharging outlet which are the lowest points of the collapsed structures. Although their depth cannot be determined precisely, it would seem that they are not deep because some of these structures are still vis~le, and the temperature on the ground above the channels is relatively high considering the low thermal conductivity of the pyroclastics. In addition the shape of the negative zones A - and B-- superimposed on zone S-t- confirms the observations made by means of magnetotelluric profiling that

SELF-POTENTIAL SURVEY NEAR THE CRATERS OF STROMBOLI VOLCANO (ITALY)

the channels are at a sharp angle with respect to the ground surface and that plane sections of the convection cells perpendicular to the feeding channels have dimensions which increase with distance from the interface pyroclastics-old structure. Fig. 13 represents the possible paths of the various channels and the discharging outlets inside the three craters. In fact the lava carried by the various channels spread over the roughly horizontal surface which form the ceiling of the collapsed structure. Thus it is possible that a discharging outlet, situated inside one crater, may be fed by a channel which seems to be associated with another crater. This possibility is suggested by the drawing of crater 2 which gives rise to eruptions which could apparently be classifted into two distinct categories (D1, D~ D ~ refer to the discharging outlets keyed in Fig. 1).

361

where o is the charge density, 5 is the thickness of a cylindrical condenser composed of the group of opposite charges, L the conductance and v the viscosity of the fluid. Streaming potential measurements (Mc INNES, 1961; AHMAD, 1964) have been carried out on liquids, and it is difficult to evaluate the importance of this phenomenon for gases; however one may consider that the low viscosity of gas should promote the existence of high potentials, which is the case for steam passing through a tubular path (KI2NKERBERG et al., 1958). A study of the porosity of pyroclastics by the mercury intrusion method enabled us

THE ORIGINS OF SELF POTENTIAL The above study has shown that the potential anomalies observed are associated with the presence of convection cells of gas and condensed liquid, but we have not identified the mechanism which causes the observed electromotive forces. Various physical, chemical or electrochemical processes can create such potentials. In the specific case which we have studied, they could be generated by an electrofiltration phenomenon, the result being an effect which is called streaming potential. Contrary to electro-osmosis, streaming potentials are caused by the preferential adsorption of an ion on the grain surface when a fluid is forced by a pressure difference A p to penetrate a porous environment (Mc INNES, 1961), thus the fluid carries the opposite charge from that of the adsorbed ions. The potential which is created m a y be expressed in the case of a capillary by the Helmoltz equation. o5 E=Ap

Lv

'

FIG. 13 -- Paths of the various channels and discharging outlets within the three craters. The paleo-vol~nic material is single hatched if still in place or cross-hatched if collapsed. Recent pyroclastics are stippled. The arrows indicate the point of interface between the two types of material. At this points the convection phenomenon is very marked.

362

R. BALLESTRACCI

to evaluate its capacity of ionic adsorption; the total surface of the pores is 1.34 m2/g. The total volume of the pores is 0.17 cma! g and the average diameter is 0.5 micrometer. These values are comparable to those measured for classical mineral adsorbants. For the purposes of comparison, pore surfaces are around 0.1 m2/g for simple geological material (quartz) and equal to or greater than 100 m2/g for chemical catalysts (GREEG, 1967). T h e importance of self potential anomalies generated by the streaming potential phenomenon are also dependent of the density of charge present in the fluid. The presence of saline deposits on the surface directly above the gas outlets, together with the analysis of gases emitted by the volcano (TAZmFF, 1970 and LE GUERN, 1972) lead us to suppose that the density of charge is high. Thus conditions seem favorable to the existence of streaming potentials. T h e positive potential zones are associated with rise by convection of gas (and condensed liquid near the surface). The negative zones are due to electroflfltration of condensed liquid above the colder zones of the volcano. Other processes could generate self potentials. Potentials of thermo-electric origin, if they had an effect on the measured anomalies, would be present over all the higher part of the volcano; this is not the case. The diffusion potentials between solutions of differing salinity cannot be of very great importance, since the rainfall is low and the temperature too high. Potentials due to salts impregnating the zones situated above the fluid rise path cannot explain the presence of negative anomalies.

MECHANISM OF STROMEOLIAN ERUPTION The conclusion proposed in the previous study (BALT,ESTRACCI, 1982) relating to the mechanism of Strombolian eruptions remain valid. The geometric disposition of the group dike-feeding tubedischarging outlet (Fig. 13) means that the lava flow, which, like the emission of gas,

is continuous, temporarily blocks part of path of the gases, the plug of lava cools and the increasing pressure of gases, once the force it creates is greater than the force of viscosity ejects forcibly the lava mass which shatters on leaving the vent, giving rise to sprays and scoria at high temperature, accompanied by gas and lapilli. This process is typical of what has been called <>; this ambiguous t e r m represents for us eruptions which happen for each discharging outlet at more or less regular intervals and which release energy accumulated by the increase in gas pressure in the feeding channels. This relatively simple process has to be reconciled with the fact that the number of discharging outlets is in general greater than the number (3) of feeding channels. After eruption of a first discharging outlet, ejected material cools, falls down and obstructs the orifice. A second discharging outlet fed by the same channel, and whose plug has been heated by gases since its previous eruption, will erupt in turn. The existence of several layers of lava situated at various levels, tends to give a complex aspect to the chronology of eruptions in the various craters. Paroxysmal eruptions of Stromboli volcano are fairly common (the most recent occurred in 1956, 1967, 1975). A precise description of this type of eruption was given by RITTMANN (1931, 1933) concerning the eruption of l l t h September 1930. The eruption began with two very violent explosions followed by an eruptive cloud 2.5 km high above the craters, and blocks weighing several tons were projected as for as 3 k m away. This first phase was followed by a hail of scoria and ash which claimed several lives. In the following twelve hours, lava flows originating just below the crater terrace spread over the Sciara del fuoco down to the sea. Before the explosion, the whole island had been lifted 1 metre, and it suddenly regained its initial position as soon as the explosions had freed the accumulated gases; at the same time, a wave two metres high broke over the island. This eruption totally transformed the crater terrace (Fig. 14).

SELF-POTENTIAL

SURVEY NEAR THE

CRATERS

The elements which characterise the paroxysmal eruption described by Rittmann can be explained quite simply by the modes which we have worked out. Lava flow and gas emission are continuous phenomena, the gas being the driving

~.~

"'~2,

• .....

"

~

.

OF STROMBOLI

VOLCANO (ITALY)

363

force of the eruptions. It is also the gas which heats up the surveyed parts of the volcano which are at some distance from the feeding dikes. If the gas flow should slow down, all the discharing outlets cool down and the force of viscosity increases faster than the force exerted by the pressure of gases. The outlets then become blocked. The pressure of gases inside the volcano then increases dramatically. When the pressure (which is exerted over a relatively small area) is too great, part of the crater terrace is suddenly fractured, blocks are projected into the air, gases loaded with scoria and ash expand suddenly to a high altitude. Large quantities of lava are emitted on the crater terrace. It cannot be ejected by the normal process and thus breaks through the weak wall of the pyroclastics cone and spread over the Sciara (Fig. 14). Meanwhile lava fourtains create new craters. The simple conclusion one can give on the prediction of a paroxysmal eruption of Stromboli volcano is to observe the rhythm of eruptions. As soon as eruptions cease and there is no more gas escaping, there is danger of paroxysmal eruption. CONCLUSION

.. < , ~,' :

•

.,

-:~

| FIG. 14 -- Transformation of the crater terrace after the paroxysmal eruption of llth September 1930 (a-before; b-ajar). The lava flows which, having crossed the pyroclastics crater spread over the Sciara, are clearly visible (after RITTMANN, 1931).

1. - The upper part of Stromboli volcano is the site of important phenomena of convection of gases and condensed liquids. 2. - These phenomena give rise to electrical potentials on the ground surface whose distribution suggest that they are caused by a dominant mechanism which is probably electroffltration. 3. - The map of potentials in the neighbourhood of the craters may be interpreted ad being the superimposition of potentials created by several convection cells: two are situated immediately above two feeding tubes, several adjacent cells are associated with the group of craters, and another cell originates in the presence of dikes within the paleo-volcano. 4. - Resistivity and self-potential surveys of the upper part of the Stromboli volcano proved to be very efficient in

364

R. BALLESTRACCI

delimiting the position of the channels which feed the various discharging outlets with lava and gas. 5. - <>activity is explained by the geometric disposition of the dike - feeding channel - discharging outlet grouping. T h e outlet, situated at the lowest point, is regularly obstructed by a plug of lava which is ejected by the gas when the pressure becomes sufficiently great. 6. - T h e disposition of the craters originates in a recent collapse of a section of the volcanic cone, probably linked with the subsidence of the central caldera. 7. - T h e mechanism of strombolian eruption is possible only because the quantity of gas emitted is high. It is difficult to explain this peculiarity by degasing of the m a g m a during its rise. It remains to understand, therefore, the origin and genesis of the gases of Stromboli in order to clarify the overall working of this volcano.

AKNOWLEDGEMENTS This work was supported by the P.I.R.P.S.E.V. program (Programme Interdisciplinaire de Recherche, de Pr~vision et de Surveillance des Eruptions Volcaniques), from the Centre National de la Recherche Scientifique and the Institut National d'Astronomie et de G~ophysique.

REFERENCES AHMAD, M.U., 1964, A Laboratory Study of Streaming Potentials. Geophys. Prospect., 12, p. 49-64. ANDERSON, A.A. and JO~a-ISON, G.R., 1976, Application o / t h e Self-potential Method to Geothermal Exploration in Long Valley, California. J. Geophys. Res., 81, p. 15271532. BAU.~STRACCl, R., 1982, Audiornagnetotelluric Profiling on the Volcano Strornboli, Internal Structure and Mechanism of the Strombolian Activity. J. Volc. Geother. ires., 12, p. 317-337.

BARBERI, F., GASPARINt,P., INNOCENTI,F., and VILLARI,L., 1973, Volcanism of the Southern Tyrrhenian Sea and Its Geodynamic Implications. J. Geophys. res., 78, p. 5221-5232. BLACKBURN, E.A., WILSON, L., and SPARKS, R. S. J., 1976, Mechanism and Dynamics of Strornbolian Activity. J. Geol. Soc. Lond., 132, p. 429-440. CAPALDI, G., GUERRA, I., Lo BRASCIO, A., LUONOO, G., PECE, R., RAPPOLLA, A., SCARPA, R., DEL PEZZO, E., MARTINI, M., GHIARA, M. R., IARER, L., MUNNO, R., and LA VOLPE, L., 1978, Stromboli and Its 1975 Eruption. Bull. Volcanol., 41, p. 260-285. CHOUET, B., ~ E V l C Z , N., and GETCI-nN, T.R., 1979, Photoballistics of Volcanic Jet Activity at Stromboli, Italy. J. Geophys. Res., 79, p. 4961-4976. COLLmR, J.G., 1972, Convective Boiling and Condensation. Mc Graw Hill, London. CORW~, R. F., and HOOVER, D. B., 1979, The Self-potential Method in Geothermal Exploration. Geophysics, 44, p. 226-245. DONALDSON, I.G., 1962, Temperature Gradients in the Upper Layers of the Earth's Crust due to Convective Water Flows. J. Geophys. Res., 67, p. 3449-3459. Fox, L., 1962, Solution of Ordinary and Partial Differential Equations. Pergamon Press, London. GRESS, S. J., and SING, K. S. W., 1967, Adsorption, Surface Area and Porosity. Academic Press, London and New York, 1967. HALWACHS, M., Personnal communication, 1979. KLmKENBERa, A., and VAN DER MINNE, 1958, Electrostatics in the Petroleum Industry. Elsevier, New York. LE GUERN, F., 1972, Etudes dynarniques sur Ia phase gazeuse ~ruptive. Rapp. CEA. R-4383, sere. de doc., Centre Etudes Nucl. Saclay, France. Mc GETCHIN, T. R., and CHOUET, B. A., 1979, Energy Budget of the Volcano Strornboli, Italy. Geophys. Res. Letters, 6, p. 317-320. MCINNES, D. A., 1961, The Principles of Electrochemistry. Dover Publications, New York. R1TTMANN, A., 1931, Der ausbruch des Strornboli am 11 sept. 1930.. Zeitschr. fur Vulk., 14, p. 47-77. - - , 1933, Beitrag zur kenntnis des Stromboli craters. Zeitschr. fur Vulk., 16, p. 184190. - - , 1963, Les volcans et leur activitd. p. 211-212, Masson, Paris, 1963. SETTLE,M., 1973, A Statistical Analysis of the Activity of Stromboli, Italy, during Early September 1971. Eos Trans. AGU, 54, p. 509510.

SELF-POTENTIAL SURVEYNEAR THE CRATERS OF STROMBOLI VOLCANO(ITALY) TAZIEFF, H., 1970, New Investigations on Eruptive Gases. Bull. Volcanol., 34, p. 421438. WOODINGS, R.A., 1957, Steady State Free Thermal Convection of Liquid in a Satured Permeable Medium. J. Fluid Mech., 2, p. 273285. ZABLOCKI, C.J., 1973, Mapping Thermal Anomalies on an Active Volcano by the Selfpotential Method, Kilauea, Hawaii. Proc. 2nd U.N. Syrup. on the Development and

365

Use of Geothermal Resources, San Francisco, U.S. Government Printing Office, Washington, D.C., 2, p. 1299-1309. ZOHDY, A. A. R., ANDERSON, L.A., and MUFLER, L. J. P., 1973, Resistivity, Self-potential, and Induced-polarization Surveys of a Vapor-dominated Geothermal System. Geophysics, 38, p. 1130-1144.

Ms. received Nov. 1981; sent to review Nov. 1981. Revised ms. received Dec. 1982.