Volcanic Risk Zoning in the Island of Ischia (Italy)

Volcanic risk zoning in the island of Ischia (Italy)

M. Matteral, J,F. Martfn-Duquel, J. Pedrazal, M.A. Sanzl, R.M. Carrasco2 & J.M. Bodoquel ‘ Univemidad Complutense, Spain 2 Universidad de Castilla-La Mancha, Spain

Abstract

Ischia constitutes the largest and more populated island of the Neapolitan archipelago. As other active volcanic areas along the middle-west Italian coast (Somma-Vesuvius and Phlegrean Fields), the volcanic activity of Ischia stems from deep fractures associated with the Tyrrhenian sea-floor spreading over the last 10 million years. The record of historic eruptions along with signs of thermal activity show that there is a potential for hazardous volcanic events. The paper constitutes a first approach to the zoning of the volcanic risk of the island, by considering the three main factors involved — the elements at risk, or value (population), the hazard posed by the volcanic phenomena, and the degree of damage resulting from the hazard (vulnerability). The analysis of the hazard was carried out by using Geographic Information System (GIS) techniques and eruptive models. The analysis was based not only on the evaluation of the probability of occurrence of a fiture new eruption within the island, but also on the evaluation of the probability of occurrence of different intensities and topologies. As those topologies have different energy and potential for destruction, they also condition the vulnerability of the value, which is different for each volcanic phenomenon. The results show lower volcanic risk levels than for similar volcanic areas, However, if two characteristics of the analysed territory are taken into account —the high tourist affluence and the insularity-, then the risk shouldn’t be underestimated.

1 Geological and vulcanological setting of Ischia

The island of Ischia is located northwest of the gulf of Naples (figure 1). The volcanic activity of the Neapolitan archipelago is related with the recent geological evolution of the Tyrrhenian Sea. About 10 million years ago, the © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

16 Risk Analysis III

Italian peninsula and the islands of Corsica, Sardinia and Sicily were bound together. Since then, a process of sea-floor spreading formed the Tyrrhenian Sea, determining an anticlockwise rotation of the Italian peninsula. This movement, still in progress, has generated the stretching of the crust and the formation of deep faults, favouring the outflow of magmas to the surface. Ischia has been formed by different pyroclastic eruptions and effimive activity, Absolute dating and volcanological studies have allowed to summarize this activity in five main phases, accurately studied and described by Vezzoli [1]. Different from sole volcanoes, Ischia doesn’t show an only vent — previous eruptions are aligned to fault-associated centres. This circumstance was important for the hazard evaluation, which focused not only on the absolute probability of occurrence of a new eruption, but also on the spatial probability of formation of new eruptive centres —and their possible eruptive topologies-.

Figure 1: Digital Elevation Model of Ischia obtained from a grid of 10 m. The figure shows a superimposed one-square-kilometre grid (UTM coordinates, map n“ 183 of the Italian Military Geographic Institute, scale 1:25 ,000), to which several calculations will be referred to.

2 Methodology

Despite there are numerous procedures developed for the evaluation of natural risks, perhaps the most widely accepted by the scientific community is that proposed by UNESCO [2]. Specifically considered for volcanic risk, Scandone and others [3] state:

Risk = (Value) x (Vulnerability) x (Hazard) (1) where: Value, total amount of lives or properties at risk for a volcanic eruption; Vulnerabili&, percentage of lives or goods likely to be lost because of a given © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

Risk Analysis III 17

volcanic event; Hazard, probability that a given area may be affected by a certain volcanic phenomenon, Among thew the most difficult factor to evaluate is the hazard, because it includes the evaluation of the probability of occurrence of a new eruption —with different intensity-, and the evaluation of the probability of the different topologies that might occur for each eruption.

2,1 Hazard evaluation

The hazard evaluation was based upon the hypothesis that a future volcanic eruption will take place behaving similarly to previous episodes, Thus, the first step was the characterization of the volcanic record of Ischia [1], This was done for the last 55,000 years, which is the age of the main stratigraphic unit of the island —the Tufo Verde- and the beginning of the present volcanic dynamics, The second step was to assign the Volcanic Explosivity Index (VEI) of Newhall and Self [4] to each eruptive episode. For this, the volumes of each eruption were evaluated, and the type of eruption —from the characteristics of the deposits [1]— were deduced, as they are the main factors which allow the assignment of a VEI. The volume evaluation was carried out by the measurement of a three- dimensional representation of the different volcanic units (by using both the Surfer 7.0 and CartaLinx 2.0 software), Results are shown in table 1,

Table 1. Volcanic eruptions in Ischia for the last 55,000 years and associated VEI [4]. (+) cubic meters; (*) thousands of years.

EPISODE VOLUME (+) VEI AGE (*) EPISODE VOLUME (+) VEI AGE (*) Tufo Verde 6040421764 6 55,00 S,Anna 65889405 3 22.60 Pietre Rosse 1537937600 5 46.00 Costa Sparaina 44729669 3 4,00 TtiI del Giglio 1510000000 5 33,00 C.s Costalrzo 32638920 3 38.40 Piano Liguori 1103815279 5 5.50 Schiappa and Pomicione 21897696 3 24,00 Selva de] Napolitano 695314783 4 10.00 Pilaro 17391059 3 25,00 Unkrrown centers 386421760 4 3,10 Ciglio and Cava Pelara 13701178 3 23.00 Puuta Irqeratore 272666079 4 19.00 Monte Cotto 11942697 3 28.00 Cantariello 232410831 4 5,00 Monte Vezzi 11048663 3 27,00 Camotese 225207245 4 17.00 Monte Trippodi 9299688 2 1.80 S. Montano 99000000 3 34.00 Vatoliero and others 7972679 2 1.70 Zaro 96249528 3 6.00 Rotaro III 7559731 2 1,70 Bosco dei Conti 94806104 3 2.50 Punts dells Carrrumia 7226958 2 4,50 Upper Cava Pelara 91452237 3 20,00 Lower Cava Pelara 5875222 2 24.00 %arrupo di Parrza 91183976 3 24.00 Catieri 3467587 2 15.00 Rotaro I 91093313 3 2.10 Rotaro IV 2898951 2 1.61 Montagnone I 86082744 3 2.70 Rione Bocca 2736900 2 2.40 Grotta di Terra 85499103 3 19,50 Monte Tabor 984127 1 3.50 Rotaro II 83253603 3 1.80 Porto dkhia 908138 1 2.30 Montagnone II 73218441 3 1.90 Castiglione 795491 1 2.80 Arso 68307999 3 0.70 Grotta del Mavone 557234 1 29,00

Once the VEI was assigned to each episode of the volcanic record, the probability of occurrence for each VEI was evaluated. For the statistics of volcanic eruptions, it is classical the distinction between volcanoes “without © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

18 Risk Analysis III memory” and “with memory” (Wickman [5]), For the former, the number of eruptions in a given time interval has a pattern that follows the Poisson distribution — the probability of occurrence of an eruption within a time interval is independent of time. Ischia behaves like volcanic areas with memory, where the probability of occurrence of an eruption is not independent of time, because the larger the volume of an eruption, the longer the time between an eruption and the next, For these situations, the Poisson distribution can not be applied directly, but subdividing the volcanic activity in different VEI classes, Then, the probability of eruption for each VEI can be generalized according to a Poisson distribution, Specifically, for explosive volcanoes, the relationship between the frequency and each VEI has an exponential pattern (Scandone and Giacomelli [6]):

where: Q is the eruption frequency per year; a is an eruptive fi-equency index of the volcano; and b represents an index of the style of activity of the volcano. This relation can be expressed similarly to the Gutenberg-Richter equation [6]:

logo = log(a) -b* J“H (3) where: b is now a coefficient that represents the slope of the regression line established for VEI 2 2. Eruptions with VEI < 2 are not considered in the analysis, because it is assumed that they are underestimated in historical records, Actually, table 1 shows clearly how the number of events of VEI <2 decreases noticeably with time. For this reason, to calculate the frequency for each VEI, a different time span was considered. By implementing the frequency results for each VEI in the relation (3), a simple regression was established between frequencies and VEI classes. Results are shown in figure 2.

0.01 n = -1.96016983 = y = -1.96016983-0,46014296x ~ = .0,46014296

0.001 t’= ‘w

0.0001 I @= l,096~l@ ~ Jo-(o 4614”~U 1 \ ‘-squued=0998074

Figure 2: Mathematical development and fashion of the simple regression between eruptive frequencies and VEI classes. © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

Risk Analysis III 19 With this relation, the Age-Specific Eruption Rate (ASER) for each VEI was obtained. Then, the eruption probabilities (10 and 100 years) for each VEI (probability of observing n eruptions for a given t time) were calculated (table 2).

Table 2. ASER and VEI eruption probability. The probability of observing n eruptions for a given ttime is defined by: P(O,QVEI = exp (- cD*O.

Time interval Number of Age-Specific I m,, , n, I r,, , .nfi. VEI r(l.lu)v~~ r(l.l UU)VEI (years) eruptions Eruptior n Rate 1 4,500 3 0.0037!‘9915 0.03657522 0.25983129 2 5,000 6 0.00131687 0.01299642 0.11543884 3 20,000 10 0.0004:5646 0.00454381 0.04360928

4 W7,”.. I .. nnn 5 0.00015822 0.00157970 0.00155736 , r!r,rmr15440 5 55,000 I J 4--I V.uuvu 0.00054810 0.00545400 6 I .55.000. .. . . I n..nnon....1901 0.00019006 0.00189739

2.2 Spatial favorability for new eruptive centres by means of GIS analysis

As it has been already pointed out, Ischia doesn’t have a preferential vent, For this reason, the whole territory of the island is subject to the formation of a new crater. The objective of this epigraph is the evaluation of the spatial probability for the occurrence of a future eruptive centre. The evaluation was referred to a grid of 10xl O m (raster map), and the GIS utilised to perform the analysis was Idrisi 32, The basis for the evaluation were different signs of potential volcanic activity. This information was compiled from various bibliographic sources, In all cases, deterministic approaches were followed — a zoning of the territory, for each sign, as a fimction of the tendency of each pixel to favour a new volcanic event. The methodology varies with the sign type, considering their specific conditions and dynamics.

2,2.1 Distance analysis — fault conditioning and hydrothermal activity Due to the close relationship between volcanic eruptions and both factors, as reflected by the eruptive history of the island, a distance analysis was carried out from both faults and hydrothermal sites (Vezzoli [1]). By plotting the fault and hydrothermal distance maps with the topological location of all the eruptive centres occurred within the island in the last 55,000 years, new maps —showing the distance of both factors with respect to the eruptive centres— were obtained. Through the analysis of the histogram obtained from both resulting images, the mathematical fimctions that fitted with the pattern, and the establishing of the mathematical-statistic parameters that define those fimctions, were recognized. Once these algorithms were obtained, they were implemented in the digital base map —through the image calculator and reclass modules—, obtaining the respective propensity maps for both factors (figures 3A and 3B), Calculation limitation of Idrisi 32 required to break the initial fimction down into several compounds, and each one was expressed by means of the algorithms considered to be more accurate, © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

2(I Risk Analysis III 2.2.2 Anomalies analysis — gravity, radon and magnetic anomalies The starting point for the gravity, radon and magnetic anomalies analysis was the digitisation of the contour lines corresponding to the different values for each factor. This was carried out by using the CartaLinx 2,0 software, and from existing data sources (Nunziata and Rapolla [7], Maino and Tribalto [8]). From the contour maps, the corresponding digital terrain models (through the conversion tools of Surfer 7.0 and Idrisi 32 software) were obtained. From the digital terrain models, the corresponding slope maps (expressed in radians) were derived, These maps show the spatial variation of the Z values for each anomaly. The figures of these maps were later standardized — the total sum of the pixels

was made equivalent to 1 (or 100 ‘XO), and the obtained figures were recalculated. By these procedures, propensity -or favourability— indexes were obtained (figure 3C, 3D and 3F). They show the relative probability to the formation of a new crater, based on the information provided by each factor.

2.2,3 Mogi analysis During its recent geological history, Ischia has undergone vertical movements. Maino and Tribalto [8] (see [9]) studied the evolution of this phenomenon for the period 1913-1967. This information was analysed by adapting to this particular situation the model proposed by Mogi [10]. Mogi’s model links the depth~ of a magmatic chamber of a radius r with the vertical deformations Ah recorded in the land surface d. From [8], a digital model of vertical movements was produced. Once the point of maximum aptitudinal variation was located, the average values of vertical deformation Ah for different ranges of horizontal distances to that point were established. By plotting these data in a Cartesian diagraw a curve was drawn, Then, the parameters for the polynomial curve of second order that fitted with the fashion of the curve were established, This allowed us to establish the depth to which the magmatic chamber would be situated ~= 2,750 ~ 250 m), The variation in the vertical deformation Ah related with the horizontal distance was expressed by a function of normal distribution, which higher value corresponds with the distance d, By a process of standardization (the total sum of the values was made equivalent to 1 --or 100 %--, and then the values were recalculated) a zoning of the island, showing the propensity for the formation of a new crater -considering the signs supplied by elevation variations- was obtained. The results (relative probability) are shown in figure 3F.

2.2.4 Data integration Figure 3G shows the integration of all the favorability indexes. The final data, referred to the grid subject to analysis (985 x 776), were simplified and converted to the grid of figure 1. To make this calculation, it was considered that the sum of all the probabilities values had to be one — it is to say, that in case that an eruption will occur, there is a 100% of probability that it will occur in one of the cell of the island. Giving the same weight to each factor, the sum of each map had to be 1/6. Figure 3H shows the simplification of the results. © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

Risk Analysis III 21

—

Fismre–e 3: A to F. favourabilitv indexes for the formation of a new eruptive centre by considering the ~nformation supplied by different volcanic signs, A) Faults. B) Hydrothermal activity. C) Gravity anomalies. D) Magnetic anomalies. E) Radon anomalies; F) Mogi analysis. G) Hazard zoning by integration all the favorability indexes. H) Simplification and conversion to the one-square-kilometre grid (see figure 1). © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

22 Risk Analysis III 2.3 Hazard zoning of eruptive topologies by means of simplified modelling

A simplified hazard zoning of the different eruptive topologies likely to occur in Ischia -pyroclastic flows and falls, and effusive activity— was made. It was carried out by a modelling of the different phenomena by: (a) considering the physical mechanism of each eruption type; (b) taking into account the specific parameters for each VEI class; (c) referring them to a one-square-kilometre grid of the topographic map (see figure 1). The probability, for each eruptive typology t,was evaluated by the following expression:

(4)

where: k is the number of events with a probability of formation of a new crater Pc, which, on the bases of the modelling, reach the cell i.

2.3,1 Pyroclastic flows hazard Pyroclastic flows are one of the most damaging volcanic processes [6]. As considered here they include other phenomena, like surges and blasts. Simulations were carried out for eruptions of VEI >3, because eruptions with a lower VEI don’t produce pyroclastic flows. Also, eruptions with VEI >6 are considered to affect to the whole island, regardless of where they occur, For simulations of eruptions with VEI between 3 and 5, the concept of “energy cone”, as proposed by Sheridan [11], was essential. In this model (figure 4) it was considered that the distribution of the area of influence of pyroclastic flows would take place within the extent of the circumference of the cone. The parameters that define the shape and dimensions of the cone depend on both the VEI and the elevation of the crater, For this reason, the minimum height of collapse of the eruptive column H,Ol of a pyroclastic flow (for each VEI) is the sum of the height of the energy cone OV, plus the average altitude of the cell. The collapse height of an eruptive column, as a fimction of the volumetric flow O, is given by [12]:

HCO,= 2.9043+ 0°5’87 (5)

According to Siebert and others [13], if the angle 5 (figure 4) —friction angle- has a value of 6°, it is possible to calculate for each VEI —through a simple trigonometric relationship- the circumference radius of the basis of the cone, The energy cone was superimposed to the topography, considering that it is situated in the middle point of each cell of the grid, This process revealed those areas non vulnerable, for being out of the reach of the pyroclastic flows, either because of their altitude or because they are protected by natural barriers, By adding the probability values for each class of VEI, the total probability of pyroclastic flow hazard was obtained. Results are shown in figure 4.

2,3,2 Pyroclastic fallout hazard With this expression, we refer the hazard posed by those volcanic materials that, propelled by eruptive Plinian columns, fall out from the air by the gravity effect, © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

Risk Analysis III 23

Energy cone (Sheridan [11]) Hazard map

@ Hx w I 1 4m.40.. .’0. ,01,. <6,0. ,0.. ‘!.. ,!,0. ,),,. “~~ (vohunetic (heightof the eruptive (cone Probablllty flow) collapsecone) radius) 3 105.7 32,6 310.2 4 1681.8 136,9 1302,5 5 65695.6 916.0 2751,2 6 262376.3 18411 17516,9

Figure 4: Pyroclastic flow hazard. Physical model, dimensions of the energy cones used for simulations, and hazard zoning.

Spatial distributions of fallout deposits have an elliptic shape, with the volcano occupying one of the ellipse focus, and the major axis facing the direction of the prevailing high winds. In the proximities of the volcano, pyroclasts fall is not conditioned by winds, so that the deposits distribution is not elliptic but concentric to the crater (see figure 5). Considering this physical fallout pattern, the hazard evaluation was based on simulations for eruptions of 3,4, 5 and 6 VEL Figure 5 shows the physical model for the simulations, and the geometrical parameters of the corresponding ellipsoid for each VEI class. Those parameters were calculated by analysing the volcanic record (eruptions of Baia, Pomici Principal and Tufo Giallo, to which VEI values of 4, 5 and 6 were assigned respectively). The ellipse circle and the evaluations are referred to deposits with a thickness higher than 1 metre. In each simulation, the eruptive centre was located in the ellipse focus, and superimposed to the topography. The maximum length of the ellipse was established considering the influence of the prevailing high winds (> 5,500 m for VEI = 3; >12,000 for VEI > 4), By applying (4), the modelling was carried out. Results appear in figure 5.

2.3,3 Effusive activity hazard The maximum distance that lava flows can reach depends on the emission rate and on the topography through which they flow, The parameters needed for those simulations were obtained from trigonometric relations and from the analysis of the geometry and dynamics of two historical eruptions, Rotaro III and Arso [1] [14], with VEI 2 and 3 respectively, Higher VEI classes don’t produce significant lava flows. For these simulations, the model proposed by Walker [15] was followed, © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

24 Risk Analysis III

Geometry of the distribution of pyroclastic falls Hazard map I I ‘“~

Figure 5: Pyroclastic fallout hazard, Physical model, dimensions of the ellipsoids used for the simulations, and hazard zoning.

In Walker’s model, lava flows geometry depends on their chemistry. Geometry is defined by H —horizontal spreading in terms of the circle diameter which surface is equivalent to the lava flow— and by V —average thickness of the lava flow-. The rate between both parameters defines the “aspect ratio”. Calculations of the aspect ratios for the considered VEI classes were carried out according to existing studies [1] [12] [14]. These calculations allowed us to establish specific values of the geometrical parameters, By adding the probability values for each VEI class for each cell of the grid, it was possible to obtain the total probability of lava flows reach for the whole island (figure 6).

2.4 Value and Vulnerability

On the basis of total number of inhabitants living in the island from 1951 to 1991, human lives have been the value (1) considered for the risk zoning. Therefore, vulnerability (1) was estimated in terms of percentage of lives likely to be lost by the different topologies of volcanic phenomena. The most common estimations for this index are the mortality rates due to different volcanic phenomena in recent times, as reflected by the literature [3]. Thus: - New eruptive centres, As assumed, the formation of a new eruptive centre, at any location, will cause the destruction of an area of approximately one square kilometre. Thus, the vulnerability associated within this extent is 1 (100 %). To refer it to the different municipalities, the value of l/N (where N is the number of cells for each municipality) was considered. - Pyroclastic flows. The high virulence and destructiveness of this phenomenon implies that the survival rate is virtually zero. Therefore, vulnerability can be considered 1 within the extent of pyroclastic flows reach. - Pyroclastic fallout. This is one of the less dangerous volcanic phenomena. Deposit thickness of 10 cm can only cause damage to cultivations, whereas thickness of about 1 m can trigger collapses of some roofs. A figure of 0,1 of vulnerability was considered for potential deposits thicker than 1 m. © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

Risk Analysis III 25

Figure 6: Effusive activity hazard. Physical model, values of maximum distances for the simulations, and hazard zoning.

- Lava jlows (e#usive activiq). Lava flows in Ischia are characterised by a high viscosity, so that the mortality due to this phenomenon is always very low. Based on similar assessments, a figure of 0.01 was assigned to this process,

3 Results, discussion and conclusion

By considering the three parameters involved in the risk evaluation (1) -by the sum of the partial risks for each volcanic phenomeno~, the total volcanic risk for each municipality was calculated (figure 7). Average values are lower than for similar and nearby volcanic regions [3] [16]. However, if two characteristics of the analysed territory are taken into account —the high tourist affluence and the insularity—, then the risk shouldn’t be underestimated. The high tourist affluence has a severe influence on the elements at risk, or value. Thus, considering than more than 350,000 tourists visit the island in summer, the total risk would be actually much higher than that shown in figure 7. The insularity factor would create problems at the time of implementing evacuation plans regarding a possible volcanic eruption, as the present-day means of transport by sea from Ischia would be insufficient.

Total Risk

Municipalities 1951 1961 1971 193.1 1991

B?.mo @ Ischia 1342 1298 1303 1418 1717

I Casamicciola I 536 I 5,98 I 648 I 7,13 I 7.89 I

Forkd%chia 1604 17,52 19,78 23,44 2800

lschia Pono 1801 19,79 24,85 2781 28.52

!-am Amno L56 1,90 2.32 267 298

Serara Fontana 298 3,02 307 337 3.71

Figure 7: Volcanic risk evaluation and zoning of Ischia. Data by municipalities. © 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved. Web: www.witpress.com Email [email protected] Paper from: Risk Analysis III, CA Brebbia (Editor). ISBN 1-85312-915-1

26 Risk Analysis III Risk reduction and mitigation would imply a simultaneous improvement and revalorisation of some unique volcanic features, which rather than being exclusively signs of potential volcanic activity, constitute an invaluable natural and cultural heritage. While this study is preliminary, it does suggest avenues for future improvement: (1) computer-aided quantitative models for the eruptive mechanisms; (2) extend the value analysis to the social-economical situation.

References

[1] Vezzoli, L., (cd). Island of Ischia, Quademi dells Ricerca Scientific n“ 114, Vol. 10, CNR: Roma, 1988, [2] UNESCO, Report of consultive meeting of experts on the statistical study of natural hazards and their consequences, SC/WS/500, UNESCO: Paris, 1972. [3] Scandone, R., Arganese, G. & Galdi, F,, The evaluation of volcanic risk in the Vesuvian area, Journal of Volcanology and Geothermal Research, 58, pp. 263-271, 1993, [4] Newhall, C.G. & Self, S., The Volcanic Explosivity Index (VEI): an estimate of explosive magnitude for historical volcanisrq Journal of Volcanolo~ and Geothermal Research, 87, pp. 1231-1238, 1982. [5] Wickman, F.E., Repose period patterns of volcanoes, V: General discussion on a tentative stochastic model, Ark, Mineral. Geol., Paper, 4-5, pp. 351-367, 1966. [6] Scandone, R. & Giacomelli, L., Vulcanologia, principi jlsici e metodi d ‘indagine, Liguori Editore: Napoli, 1998. [7] Nunziata, C, & Rapolla, A., A gravity and magnetic study of the volcanic island of Ischia, Naples (Italy), Journal of Volcanology and Geothermal Research, 31, pp. 333-344, 1987. [8] Maine, A. & Tribalto, G., Rilevamento gravimetrico di dettaglio dell’isola d’Ischia (Napoli), Bollettino Servizio Geologico d’Italia, 92, pp. 109-123, 1971. [9] Luongo, G., Cubellis, E. & Obrizzo, F,, Ischia, Storia di un ‘isola vulcanica, Liguori Editore: Napoli, 1987, [10] Mogi K., Relations between the eruptions of various volcanoes and the deformations of the ground surface around them Bulletin of the Earthquake Research Institute, 36, pp. 99-134, 1958, [11] Sheridan, M. F., Emplacement of pyroclastic flows: a review. Ash-flow tufls, ed, C.E. Chapin & W.E. Elston, Geological Society of America Special Paper 180, Geological Society of America: Boulder, pp. 125-136, 1979. [12] Mattera, M., Valutazione del Rischio Vulcanico nell ‘isola d’Ischia, Tesi di Laurea, Universit~ degli Studi di Napoli: Napoli, 1995. [13] Siebert, L., Glicken, H. & Ui, T., Volcanic hazards from Bezymianny and Banday-tipe eruptions, Bulletin of Vzdcanology, 49, pp. 435-459, 1987, [14] Chiesa, S,, Poli, S. & Vezzoli, L., Studio dell ‘ultima eruzione storica dell ‘isola d ‘Ischia: la colata dell ‘Arso 1302, Dipartimento di Scienze dells Terra, University di Milano: Milano, 1986, [15] Walker, G.P. L., Explosive volcanic eruptions, a new classification scheme, Geologische Rundschau, 62, pp. 431-446, 1973. [16] D’Andrea, M., Valutazione del Rischio Vulcanico nei Campi Flegrei. Tesi di Laurea, University degli Studi di Napoli: Napoli, 1993.