Post-Minoan Volcanic Activity of the Santorini Volcano. Volcanic Hazard and Risk, Forecasting Possibilities

This makes it possible to plot a volcanic risk zonation map.

The forecasting of a probable future reactivation that would display the same behaviour as that of the post-Minoan activity is estimated to be possible by an integrated and efficient monitoring network.

From the comparison between post-Minoan activity and pre-Minoan recurring volcanic cycles, it emerges that it is not possible to use the pre-Minoan evolutionary trends in predicting the most probable evolution of volcanic activity.

Considering all the possible scenarios of the volcanic evolution it is concluded that a cessation of volcanic activity is improbable, while the continuation of the post-Minoan activity is considered most probable.

The occurrence of a catastrophic eruption in the immediate future is excluded, while the occurrence of volcanic activity fed by basic magmas is considered probable.

INTRODUCTION

The Santorini volcano is the best studied volcanic complex of the South Aegean Active Volcanic Arc (Fig. 1). There are a lot of geological, volcanological and petrological studies and published papers about the pre-Minoan volcanic activity and the Minoan eruption. There are also a lot of historical descriptions and studies and a few recent, mainly petrological, studies on the post-Minoan eruptions.

However, none of the existing studies has systematically elaborated the various hypotheses on the probabilities and the type of the evolution of volcanic activity in the future. There are a few recent studies on the probability, the hazard and the risk of the most probable post-Minoan eruptions (Fritzalas and Papadopoulos 1988; Papadopoulos 1990; Papazachos 1990). The probability and possibility of, and also a method of successfully forecasting the start and the type of volcanic activity have not been studied either.

In this study our aim is to contribute to answering those questions, concentrating mainly on the post-Minoan volcanic activity. To that purpose, we traced premonitory phenomena and studied the manifestation and evolution of the historical volcanic activity in order to estimate the possibility and the method of forecasting, and also the hazard and risk of a repetition of similar activity.

Connecting and comparing the post-Minoan and the pre-Minoan volcanic activity with its recurring volcanic cycles and their specific characteristics, we tried to localize the position of the post-Minoan activity in relation to the pre-Minoan volcanic cycles. We also tried to evaluate the probabilities of a change in the behaviour of a future volcanic eruption.

POST MINOAN VOLCANIC ACTIVITY

After the last paroxysmal eruption of the Santorini volcano, in the 17th century BC, which deposited the Minoan pumice and produced the last calderic collapse (Heiken and McCoy 1984), volcanic activity continued, localizing mainly in the intra-calderic area.

Extrusive, effusive and slightly explosive activity produced the dacitic lavas and pyroclastics that built up Palaea and Nea Kameni (Fig. 2).

Outside the calderic depression, volcanic activity was manifested only once, in the region that is now occupied by Kolombos reef, 6.5 km north-east of the homonymous Cape of Thera.

     Intra-calderic post-Minoan activity:

     197 BC: The explosive and extrusive volcanic activity which was mentioned by Strabo built up in a short time a cone consisting mainly of pyroclastics. It was called Iera. According to Fouqué (1879), Iera was most probably sited where Bankos reef is found today. Bankos reef could be considered as a remnant of Iera (Fig. 3.1).

     AD 46-47: Mainly extrusive activity mentioned by Aurelius Victor in the Historia Romana formed Thia island which coincides with Palaea Kameni (Fig. 3.2).

     726: Highly explosive activity, from a centre located to the north-east, near Palaea Kameni, produced large quantities of pumice and ash. The explosive activity was followed by lava extrusion which joined with Palaea Kameni, forming Ayios Nikolaos lavas (Fig. 3.3) (Fouqué 1879; Akylas 1925).

     1570-1573: Volcanic activity shifted towards the north-east. Extrusive and explosive activity formed Mikri Kameni island (Fig. 3.4) (Fouqué 1879).

     1707-1711 (23/5/07 - ?/9/11): The eruption began westside of Mikri Kameni with a very slow and calm extrusion (Aspronisi) and continued shifting northward (Mavronisi) with the same eruptive mechanism. The explosive activity began two months later. Alternating extrusive, effusive and explosive activity formed Nea Kameni island over the next four years (Fig. 3.5) (Fouqué 1879; Akylas 1925).

     1866-1870 (4/2/66 - 15/10/70): After an increase of the sea-water temperature and a subsidence of the shores in Volcanos bay, the first lavas emerged by a very slow and calm extrusion. Two days later, the first explosions began. The maximum height of the pyroclastic columns in the paroxysmal eruptions was 2,200 m. The volcanic activity, mainly extrusive and effusive, continued until 1870. During that time, apart from Georgios centre, which was continuously active, two other volcanic centres appeared: Afroessa, with an initial very slow and calm extrusion, followed by a vigorous effusive activity, and May's Islands centre, which emerged from the sea with a slow and calm lava extrusion (Fig. 3.6) (Fouqué 1879; Akylas 1925).

1925-1928 (11/8/25 - 17/3/28): The eruption was preceded by an increase of the sea-water temperature in Kokkina Nera bay and the subsidence of Nea Kameni's east shore. The volcanic activity began with jets of steam and water in Kokkina Nera bay, and very soon the first lava outflows and explosions appeared. Soon the volcanic activity centre shifted north-west, to the site occupied by Dafni's pyroclastic cones. In contrast with the other eruptions, effusive-explosive activity dominated here, with subordinate extrusive activity. The maximum height of the pyroclastic columns in the paroxysmal eruptions was 3,200 m. The effusive and explosive activity continued until January 1926. After a four month interval without any event, effusive and explosive activity reappeared with a change in the pyroclastic depositional mechanisms. A limited water-magma interaction produced an overloading of the pyroclastic columns creating some limited lateral pyroclastic flows. A second quiescent interval, from May 1926 to January 1928, was interrupted by four phreatic explosions. These were followed by an extrusive and slightly explosive activity which built up Naftilos dome (Fig. 3.7) (Georgalas and Liatsikas 1926a, b; 1932; Ktenas 1926a, b; 1927; Ktenas and Kokkoros 1928a, b; Washington 1926; Reck 1936).

     1939-1941 (20/8/39 - ?/7/41): After an increase of the sea-water temperature and a subsidence of the shores in Ayios Georgios bay a submarine explosion opened the vent. An extrusive and slightly explosive activity coming from it formed Tritona's dome. The volcanic activity shifted north-east, then south, focusing finally at the centre of Nea Kameni island. An effusive, extrusive and slightly explosive activity built up the lava flows and domes of Ktenas, Fouqué, Smith-Reck and Niki. Phreatic explosions pre-announced the location of the lava outflow in all four cases. A fifth phreatic explosion ('twin explosive funnel') was not followed by magma outflow (Fig. 3.8) (Georgalas and Papastamatiou 1953; Georgalas 1959; 1962; Georgalas and Kokkoros 1940; Liatsikas 1942).

     1950 (10/1/50 - 2/2/50): After some phreatic explosions, extrusion and effusion built up Liatsikas lavas (Fig. 3.9) (Georgalas 1953).

     Kolombos eruption:

     1650 (27/9 - 6/12): Volcanic activity occurred at the end of an intense seismic excitation of tectonic origin that lasted for two years. An initial slow and calm extrusion was followed by very strong explosions which produced very large quantities of pumice and ash. On 29/9/1650 a tsunami caused serious damage to Thera's coasts and to other islands, within a radius of 150 km. The S0₂, which was produced in large quantities from Kolombos centre, proved to be a serious danger when the volcanic plume was directed by the wind towards Thera. More than 50 people lost their lives and more than 1000 animals were killed. The pyroclastic cone which emerged some metres above sea level was fast eroded. Now the cone's base is found at the depth of 280 m and its top 20 m below sea level (Fouqué 1879; Akylas 1925).

PHYSICO-CHEMICAL CHARACTERISTICS OF THE POST-MINOAN PRODUCTS

The lavas of Palaea and Nea Kameni are typical calc-alkaline dacities with a restricted range of composition (64,1 ~68, 4wt% SiO₂). The magma rises from shallow magma chambers, located at a depth between 2-4 km, with an initial H₂O content 3-4 wt% and temperatures between 950-1000°C (Barton and Huijsmans 1986).

The total volume of the intra-calderic products was estimated to be about 2.5 km³ (Heiken and McCoy 1984). The most reliable data about the volumes of the erupted product, at each period of activity relate to that of the 1925-1928 period which was calculated to be 0.1 km³ (Washington 1926; Reck 1936). We cannot estimate the erupted volumes in the other periods of activity. We have no accurate data either on the bathymetry of the caldera, or on the topography of the emerged lavas. Moreover, we believe that submarine extrusions were not registered, especially during the first centuries of the volcanic activity.

The main volumetric rate of the produced lavas is 1,164,000 m³/year, while the rate for 1866-1925 is 1,694,000 m³/year and the rate for 1711-1866 is 903,000 m³/year. As to the latter, we considered reliable the 0.14 km³ products volume calculated by Barton and Huijsmans (1986). These data prove that there is no constancy in the production of magmatic products in time, and no correlation between produced volumes and repose time between successive periods of activity.

TECTONIC FEATURES

Tectonic data concerning the area of research come from a series of field observations, from the study of the aerial photographs, and also from the evaluation of all data found in the bibliography. Additional information was collected from historical descriptions concerning very recent faults and fractures and the setting and shifting of the volcanic centres.

Evaluating all the data (Fig. 4) we point out the following: All the volcanic centres of the post-Minoan intra-calderic activity are distributed along a zone 600 m wide and 4,500 m long with a direction of N 65° E. The south-west extrusion centres of Palaea Kameni are not included in this zone. We have no reliable data to evaluate a possible different behaviour.

The characteristic shifting of the volcanic centres, observed on Nea Kameni during most of the periods of its volcanic activity, occurs in the same direction as that of the distribution zone of the volcanic centres, e.g. Georgios-Afroessa-May's Islands (1866), Kokkina Nera-Dafni's cones (1925), Tritonas-Ktenas-Fouqué (1939). There also occurs a shifting, on a much smaller scale, in a north-south direction, e.g. Aspronisi-Mavronisi (1707), Fouqué-Reck-Smith (1939). That could be considered negligible compared to the N 65° E direction.

This behaviour could be interpreted only by the existence of an extensional strain in a direction perpendicular to the distribution zone of the volcanic centres. This stretching enables the feeding dykes to develop in a direction perpendicular to the σ min (Nakamura 1977).

The presence of an extension, active until now in the central Aegean region, with a NW-SE σ min direction, is confirmed both by the neotectonic data of this region and the study of the focal mechanism of the recent earthquakes (McKenzie 1978; Jarrige 1978; Mercier 1979; Seismotectonic map of Greece, IGME, 1989).

Let us add, finally, two more points which reinforce the above observations. The first is the direction of the Bouguer gravity anomaly (Budetta et al. 1984) in the central part of the caldera. This anomaly shows a trend in the N 60° E direction. The second point concerns the observed alignment in the N 50° E direction of the two cinder cones of Megalo Vouno (Mavro and Kokkino Vouno) with Kolombos maar and Kolombos reef.

VOLCANOLOGICAL COMMENTS AND REMARKS

Keeping in mind the description of the post-Minoan volcanic activity, and knowing the physico-chemical characteristics of the feeding magma as well as the tectonic setting of the area, we can make some comments concerning the mechanism of triggering, ascent and outlet of the magma, as well as the explosive mechanisms.

The frequency of the eruptions, their duration in time, and the characteristics and the volume of the lavas produced lead us to the conclusion that there exists a relatively unstable dynamic equilibrium between the pressure exerted by the magma on the chamber walls and the overlying lithostatic pressure. The disturbance of this equilibrium, which gives rise to magma outlet, is 'temporary' and does not produce a radical change in the stress ratio. The equilibrium is restored easily, after a relatively short time of 'relassative' volcanic activity.

A slow and calm extrusion of viscous magma with no explosive activity is the characteristic of the first stages of any eruption. This means that the upper layers of the ascending magma are very poor in volatiles which leak out before and during the slow ascent. The leakage of the magmatic volatiles is slow enough to prevent fragmentation of the magma and explosive phenomena. The high viscosity of the magma could be attributed both to the leakage of the volatile percentage and to the temperature decrease during the slow ascent. The latter is due to the heating of the walls of the feeding dykes.

A short time before and during volcanic activity, a subsidence of Kameni's shores is systematically registered. The area affected by this process is generally near to or on the site at which the volcanic activity later occurs.

A self-triggered mechanism due to a concentration of volatiles at the roof of the magmatic chamber, which would decrease the density of the upper layers of the magma, should be excluded. If the above happened, strong explosive activity would be characteristic of the first stages of the eruption and an uplift of the existing formations would occur due to the magma pressure.

Completing these observations by those formulated for the tectonic setting of the area and the distribution of the volcanic centres, it appears that the most probable triggering mechanism of volcanic activity is the occurrence in the region of a tensile stress in a direction NNW-SSE. This stretching reduces the lithostatic pressure and allows to the magma to intrude along feeding dykes directed perpendicularly to the σ min, i.e. N 65° E. The concentration of the volcanic centres in such a narrow zone could be justified considering that this hot and fractured unstable zone makes the release of the accumulated tensile stress easier and therefore more likely in this area.

The above conclusions are mainly based on data and observations of the intra-calderic activity. They could also apply in the case of Kolombos due to the following facts:

-   The dominating tectonic lineament in this area is the N 50° E direction.

-   The volcanic eruption of 1650 was triggered by tectonic earthquakes, which are due to the extensional strain prevailing in the central Aegean region.

Explosive activity begins to appear after the first phase of ascent, when all the upper layer of magma poor in volatiles has come to the surface. No water-magma interaction has been observed in any of the post-Minoan eruptions. The contribution of sea and meteoric water to the eruptive activity is restricted to two cases:

   a)   The creation of 'white nubes' which are formed by the volatilization of the water which rushes into the crater, giving rise to weak explosions. In this connection the explosions of the second period of the 1925-1928 eruptive activity should be mentioned, when a limited water-magma interaction was registered.

   b)   The creation of phreatic explosions. These occur when the volcanic activity centres are located in the area of sub-aerial lavas, e.g. 1928, 1939-1941, 1950. Magma and magmatic volatiles overheat the water contained in the lavas. When the water pressure exceeds lithostatic pressure, phreatic explosions occur. The new magma appears later, in the created phreatic craters.

At this point, a question arises: what is the cause of the difference in the intensity of the explosive phenomena in the different post-Minoan eruptive periods? What is the cause of the high explosivity of 726, 1650, 1925-1928? Considering the volume of the magma produced by these eruptions it must be concluded that in all three cases there existed a high discharging rate. For the 1925 eruption, where reliable data are available, a flow rate of 450,000 m³/day was estimated (Reck 1936). It was much higher than all the previous and following eruptions of Nea Kameni. Since there is no difference in either the chemical composition or the initial content of volatiles in comparison with the rest of the lavas of Nea Kameni (Barton and Huijsmans 1986), it is assumed that the high explosivity is due to the faster ascent of the magma: during periods of high tensile stress a faster ascent of the magma is facilitated. The magma arrives at the surface richer in volatiles and an intensive explosive activity is manifested.

The eruptions of 726 and 1650 displayed some peculiarities: the eruption of 726 occurred after a repose time of volcanic activity which lasted 680 years. The erupted magma had a more acid composition (68.42 wt% SiO₂, 0.86wt% MgO) and a lower phenocrystal content (6-8 vol %) than the other intra-calderic products. Probably all these differences as well as the existence of an intensive tensional strain led to the occurrence of strongly explosive phenomena.

Unfortunately, we have no data for the 1650 eruption as regards the physico-chemical conditions of the magma. From the pumice description (bright white colour), we assume that the magma was more acid than the intra-calderic ones. Also, the explosion occurred in a region where there was no previous volcanic activity. The only registered eruption in this area was that of maar Kolombos, 30,000 years BP. Most probably, it was an eruption fed by a new batch of magma which intruded into the area and remained there a sufficient time to differentiate, accumulating a large amount of volatiles.

Based mainly on geochemical and petrological data, observations and studies, Barton and Huijsmans (1986) suggested a model of the intra-calderic magmatic chamber of limited dimensions which periodically, and up to now, continues to be replenished from deeper magma chambers with basaltic magmas.

The coexistence of small magmatic chambers in a restricted area is confirmed by the pre-calderic volcanic activity, where small chambers supplied synchronous eruptions with different magmatic products from different centres, e.g. M. Vouno-Thera, Profitis Ilias-Thera, M. Vouno-Skaros. This was also confirmed in the post-Minoan activity by the coexistence of the intra-calderic activity with that of Kolombos, 16 km from the intra-calderic centre.

Another fact that could be connected with the existence of another hypothetical magmatic chamber was manifested in the area between Oia and Therasia. On 28/11/1978, at a distance of 500 m from Cape Ayios Ioannis Thalassinos, an abrupt delivery of a large amount of gases caused intense agitation of the sea, so that a small boat found in the epicentral area was in danger of sinking.

Concluding, as to the number and size of the magmatic chambers, the presence of a series of small magmatic batches intruding into the upper layers of the crust alongside the tectonic lineaments of the area is considered most probable. The presence of a large magmatic chamber which extends over the whole area where the above phenomena occurred is considered less probable.

VOLCANIC HAZARD

The data available on the pre- and post-Minoan volcanic activity do not permit the elaboration of a quantitative statistical model which could be used for the estimation of the probability of appearance and of the characteristics of a future reactivation, i.e. a quantitative definition of the volcanic hazard in the area.

For the post-Minoan activity, which could be simulated with a very simple model of two states (eruptive state and repose state), a very low number of eruptions is available (10). It is also probable that some eruptions have not been observed (submarine) or registered, or have been registered in sources lost or unknown.

The modelling of the pre-Minoan activity which requires a very complex 'Markov chain' of states is impossible. This is due first to the activity of different volcanic centres which are supplied by different magmatic chambers, and second to the large number of eruptions, for only some of which are data available.

For the above reasons a qualitative schematic model of volcanic activity will be presented in an attempt to trace the main trend of evolution during the many cycles of volcanic activity.

     Pre-Minoan activity:      Table 1 contains all the available data about the activity of the volcanic complex of Santorini. These data come from many published papers concerning Santorini, mainly from the synthetic works of Pichler and Kussmaul (1980), Huijsmans (1985), Druitt et al. (1989), and our field observations. In the above Table, the centres of Thera, Therasia and Skaros are considered as one (Thera group centre). The reason for this is that all the existing data lead to the conclusion that there are no differences between them, so there is no reason for them to be considered as individual centres.

All the existing volcanic products are divided into three groups:

   -     The first group contains the dacitic lava domes and flows and the associated acid subordinate pyrolastic layers which usually underlie dacitic lavas.

   -     The second group contains the main acid pyroclastic formations which are deposited by high explosive magmatic or phreatomagmatic activity, including fall, flow and surge deposits, i.e. highly destructive acid products.

   -     The third group contains the basic products. Here two sub-groups are distinguished: the first (3a) concerns andesitic lava flows, cinder cones and subordinate basic pyroclastic fall products; the second (3b) concerns pyroclastic products of basic composition which are deposited by a mechanism of pyroclastic flow or hydromagmatic-maar type products. These highly destructive products display the same characteristics with regard to volcanic hazard as those of the second group.

Data from Table 1 allow the following deductions:

1. During the period of volcanic activity it seems that a passage from group 1 to group 3a without the interposition of group 2 or group 3b should be excluded. The only exception is the Akrotiri centre, but in this case a very long repose time and very active tectonics have occurred between the last two phases.
2. A synchronous activity of two different volcanic centres was observed, with the eruption of basic volcanic products belonging to group 3a (M. Vouno, Profitis Ilias) and products belonging to groups 1 and 3 (Thera group centre).
3. Paroxysmal eruptions of the second group are preceded by an important cessation of the volcanic activity of the interested centre. After a second cessation of the volcanic activity, a repetition of the paroxysmal eruptions is registered, e.g. Lower pumice 1 - Lower pumice 2, Cape Riva ignimbrite - Minoan Pumice.

The above observations enable us to formulate a hypothesis on the present stage of activity and the probable evolution of the system. According to the scheme of Table 1, the volcanic system is in phase 1, producing viscous dacitic lavas. After the completion of this phase, a cessation of the volcanic activity would be expected followed by a paroxysmal eruption which would deposit acid pyroclastics.

There are no reliable data concerning the duration of the repose time between phase 1 and phase 2. The following indications enable us to estimate the duration of the repose time to be at least 10³ years:

   -     The repose time between the two adjacent paroxysmal eruptions of Cape Riva ignimbrite and Minoan, which lasted 10⁴ years.

   -     The erosion observed in the upper part of the products belonging to phase 1 as well as the thickness of the existing palaeosols between the products of phase 1 and phase 2.

   -     The time necessary for a dacitic magma to be differentiated into a rhyodacitic-rhyolitic one, as this is the chemical composition of magmas which fed paroxysmal eruptions; this is estimated to be longer than 10³ years.

Unfortunately, in the post-Minoan activity two characteristics are present which do not permit its identification with phase 1 of the pre-Minoan cycles:

   a)   In both cases of the pre-existing 'phase 1' of the Thera group centre, an outflow of basic products (Lower and Upper Megalo Vouno andesitic lavas) was observed, synchronous with the dacitic activity. This was not observed in the post-Minoan activity.

   b)   In the pre-Minoan activity, the large paroxysmal eruptions were followed by an effusion of a large quantity of andesitic and basic andesitic lavas which filled the depressions caused by large calderic collapses, e.g. Lower Therasia and Profitis Ilias andesitic lava flows, Upper Therasia and Skaros andesitic lava flows. The above action indicates that in the pre-Minoan activity there is a lack of 'normal' evolution of volcanic activity, with extrusion of rhyodacitic-dacitic lava domes in the intra-calderic area following the catastrophic eruption which causes the calderic collapse.

These deviations from pre-Minoan behaviour deprive us of the possibility of using the model of the pre-Minoan activity even for a qualitative forecasting of the most probable evolution of the volcanic activity.

      Post-Minoan activity:      As the pre-Minoan behaviour cannot be used in forecasting the post-Minoan activity, all the possible scenarios of the volcanic evolution need to be considered:

   -     The continuation of volcanic activity with the same characteristics as the post-Minoan eruptions.

   -     The change of volcanic activity to a type of 'phase 2' or 'phase 3' of pre-Minoan activity.

   -     The definite cessation of volcanic activity.

Consider the following facts:

   a)   The existence of a shallow magma chamber in the intra-calderic area, which, 40 years ago, during the last eruption, fed a volcanic activity with dacitic products having the same chemical composition as the lavas of the last 500 years.

   b)   The existence, in the vast area, of an extensive tectonic field maintaining the same direction as that of the post-Minoan period, proved by the study of the source mechanism of all the recent earthquakes.

These lead us to conclude that a cessation of volcanic activity is improbable, while it is most probable that the occurrence of eruptions having the same characteristics as the last 500 years' activity will continue. Observing the repose time between the registered intra-calderic eruptions (Table 2), no estimate of periodicity in the occurrence of the various eruptions can be made. A tendency for the repose time to decrease during the last three eruptions is contradicted by the most recent repose time which is still continuing (40 years). If the period of 923 years which appears between the vigorous eruptions of 197 BC, AD 726 and AD 1650 is not accidental, then it probably indicates a periodicity of this magnitude in the release of the accumulated tensile stress in this area.

The probability of a catastrophic eruption (phase 2) is very small for the immediate future. It presupposes either a long cessation of volcanic activity, which could produce a differentiation in magma (> 10³ years), or the creation by stretching of large open leaking faults which would make possible the rapid ascent of the existing dacitic magma. The latter hypothesis does not seem as probable, since a gradual release of the regional tensional stress, in the already 'weak' hot and fractured zone of Nea Kameni or Kolombos, is possible.

The production of large quantities of primitive basaltic fluid in the area of Thera, reaching to the upper lithospheric layers and feeding a basic volcanic activity is one of the probable scenarios for the evolution of volcanic activity in the above area. All generative causes of the primitive basaltic fluid are present - causes suggested by the available models (subduction, mantle plume system, etc.). Such an action does not necessarily presuppose the cessation of activity in the area of Kameni. A synchronous production of dacitic and andesitic magmas is possible, as is the case in the pre-Minoan activity. The eruptive mechanism of this type of activity will depend on the location of the points of the magma outlet. In the case where the centre of activity is located on land, there occur lava flows, cinder cones and subordinate pyroclastics (phase 3a products). In the case where the centre of activity is located in a submarine area, submarine effusion with pillow lavas and hyaloclastites will result at first. In the case where the volcanic products reach shallow depths or appear on the surface creating conditions which will permit the magma-seawater interaction (Kokelaar 1983; Shepherd and Sigurdsson 1982), or when they intrude into a hydrothermal system (Druitt et al. 1989), then there will occur a deposition of highly risky products, like those of group 3b.

VOLCANIC RISK

The volcanic risk of the Santorini eruptions (volcanic hazard x value x vulnerability) depends directly on the type of the expected volcanic activity, as determined in the previous section.

In the case of a catastrophic eruption, the most probable location of the centre of activity is in the calderic area (Druitt 1990). So the caldera walls are of great importance in the distribution of the pyroclastic products as flows and surges, which are controlled by the topography. Anyway, the pyroclastic fall and surge deposits would cover all the area of the Santorini island, as in the case of the Lower pumice and of the Minoan eruption. The risk of such an eruption is very low because of the very small volcanic hazard, i.e. the probability of the occurrence of the eruption in a short time. In this case, there is no reason for plotting a zonation map of volcanic risk.

In the case of volcanic activity supplied by basic magmas, the risk from the lava outflow is non-existent in either the submarine centres or in centres located in the area of Kameni. The risk from the activity of centres located on Thera or Therasia is also non-existent; it is well-nigh impossible for centres of volcanic activity to appear in the above area since there exists the weak zone of the caldera and the area of Kolombos. In the case of occurrence of volcano-morphological conditions necessary for the occurrence of hydro magmatic explosions, the walls of the caldera act as barriers for the regions beyond their limit, at least as regards the most dangerous wet surge and flow deposits. Taking into account the probabilities for the manifestation of volcanic activity, as well as its characteristics in general, we would consider such activity of basic volcanic products as one of low risk.

The volcanic risk, if the post-Minoan volcanic activity were to be continued, can be calculated with relative precision from the already known effects of this type of activity. Table 3 gives the radius of activity of all dangerous volcanic effects of the eruptions recorded from 1650 to 1950. Data from pre-existing eruptions were rejected either because their radius of activity was not precisely determined or because their veracity is in doubt (e.g. tsunami in AD 46). The Table does not include earthquakes of volcanic origin; volcanic earthquakes are not considered volcanic risks for two reasons:

   a)    The volcanic earthquakes in all post-Minoan activity never exceed the magnitude of 5 degrees on the Mercalli scale.

   b)    The tectonic earthquakes appearing in the region are of much higher frequency and magnitude (8-9 on the Mercalli scale) than volcanic earthquakes, so that in the earthquake planning for the area, based on the tectonic earthquakes, the much lower risk from volcanic earthquakes is largely included.

From Table 3 a volcanic risk zonation map was plotted, which is given in Table 5:

     Zone A is the zone of the maximum risk. The reason is that the phreatic explosions, like those registered in Nea Kameni, when the centres of the magma outlet are located under the emerged lavas, constitute a grave danger to visitors to this island.

     Zone B includes the area where there is a danger from ballistic projectiles. In the historical eruptions (especially in 1866-70) some cases of boats sinking and injuries to crews from the ejected blocks were recorded.

     Zone C includes the area which would be affected by a tsunami with the same magnitude as that of 1650. Apart from a narrow shoreline zone, the highest risk is expected in the area between Vlyhada-Perissa and between Kamari-Monolithos, where the smooth morphology permits the penetration of the sea water to about 200 m from the shore. It should be noted that those nearshore areas are the most crowded in the island. The risk is of course low, because of the low probabilities of the occurence of a tsunami phenomenon. It is, however, a fact that should be taken into consideration in public defence planning against volcanic risk.

Let us note, in this connection, that the mechanism of generation of the gravitative wave of 1650 has not yet been elucidated. The mechanism of the generation of the tsunami has been attributed to a seismic event by Latter (1981). This is opposed by the historical records (Akylas 1925) where a pause of all volcanic activity for one hour was reported, followed by the occurrence of the gravitative wave, without it being announced or followed by any other phenomenon. We consider as more probable a mechanism of generation either by a collapse caused by the evacuation of the magmatic chamber, or by a gravitative sliding of a large amount of pyroclastics, which accumulated in steep slopes of the volcanic cone of Kolombos.

The toxic gases risk zone, and a zone of pyroclastic ash-fall, mud rains and acid rains (zone D) covers the whole area of Thera. The configuration of both zones is controlled according to the direction of the wind prevailing in the area during the period of the eruption. Taking this into consideration, zone D concerning Kolombos centre was drawn pre-supposing the existence of north-east winds. The hazard of gases is of course much higher than that of pyroclastic ash, mud rains and acid rains. The latter can affect agriculture and apiculture, while toxic gases threaten human life itself. Gases from the eruption of 1650 caused the death of 50 people and more than 1000 animals on Thera. The subsequent intra-calderic eruptions were less dangerous, causing only respiratory and ophthalmological problems to the people of Santorini.

FORECASTING POSSIBILITES

The possibility of the advance forecasting of the beginning of volcanic activity depends on the existence of an integral, efficiently operating monitoring network. The possibility of forecasting is proportional to the risk of the expected eruptions.

Our possibilities of forecasting the least probable event, the sudden creation of an open tectonic fault which could produce a catastrophic eruption, are almost nil. An abrupt tectonic event does not give sufficient time to permit the evaluation of the changes in the parameters under inspection (temperature and chemical composition of fumaroles and thermal waters, microseismic events, gravity field, geomagnetic field and geodetic data).

The probabilities for the forecasting of volcanic activity fed by basic magmas are higher. The difference in the physico-chemical features of this magma would cause grave changes in the registered parameters. Moreover, the ascent from the source region to the upper lithospheric layers of this magma is likely to be registered by the observation of a corresponding migration of the hypocentres of the micro-earthquakes.

The precursory phenomena, which have been registered since the post-Minoan activity (Table 4) permit us to estimate that an integral and operating monitoring network would make possible the forecasting, with relative accuracy and certainty, of the repetition of this type of eruption. Table 4 omits eruptions which occurred before 1707 because no reliable data are available. The lack of any record of a shore subsidence as well as of a change in the temperature of the fumaroles and sea waters is probably due to deficient registration rather than the absence of such phenomena. The activity of 1925-1928 has been estimated as two different eruptions, because the long repose time (June 1926 - January 1928) can be considered as separating two different eruptions. In 1939-1941 no earthquakes were reported. However, we have taken their existence for granted, since a fracturing of the area was reported ten days before the eruption started.

CONCLUSION

The most reliable model for the storage of the magmatic fluids in the area of Santorini seems to be the one based on the presence of a series of small magmatic batches which intrude into the upper crustal layers alongside the ENE-WSW tectonic lineaments caused by a NNW-SSE regional tensile stress. The same tensile stress is the one which triggers volcanic eruptions, allowing magma ascent along feeding dykes of a direction ENE-WSW.

The calm and slow ascent of the dacitic magma during the first stages of the eruption is due to the leakage of magma volatiles and its temperature decrease, caused by the heating of the walls of the feeding dykes.

Explosive activity begins to appear after the first phase of ascent, when the whole upper layer of the magma poor in volatiles has come to the surface. Differences in explosivity are due to the faster or slower ascent of the magma caused by more or less high tensile stress.

The role of sea and meteoric water in the eruption activity is restricted to a subordinate feeding of the explosive activity and the creation of the phreatic explosions.

The schematic model of the pre-Minoan activity of the volcanic complex of Santorini, including three different recurring cycles of activity, does not permit the forecasting of the most probable evolution of the contemporary volcanic activity because of the differences between the behaviour of the pre-Minoan and post-Minoan activity.

Considering all the possible scenarios of volcanic evolution, it is concluded that:

   -     A cessation of volcanic activity is improbable. It is more propable to expect the manifestation of eruptions having the same characteristics as those of the post-Minoan activity.

   -     The probability of the occurrence of a catastrophic eruption in the immediate future is very low.

   -     The probability of volcanic activity fed by basic magmas is high.

   -     The volcanic risk from a repetition of a post-Minoan type volcanic activity has been quantitatively estimated and expressed in a volcanic risk zonation map.

   -     The lack of any periodicity in the repose time and of any constancy in the rate of the produced volumes in time does not permit any long term forecasting for this type of activity.

Considering all the precursory phenomena which occurred during the post-Minoan activity, it is estimated that short-term forecasting in Santorini is possible using an integral and efficient monitoring network.

----------------------------------------------------

----------------------------------------------

Source: "Thera and the Aegean World III, Vol 2" (pp. 183 - 197)

Authors: M. Fytikas, N. Kolios, and G. Vougioukalakis    

Institute of Geology and Mineral exploration, Athens.cccc