Volcanism and Seismicity in Mediterranean Island Arcs

Since 1962, the author has proposed the hypothesis of a time-link between internediate-depth eathquakes occuring beneath a volcano and the climactic phase of its activities. That delay time is related to the focal depth, the type of the volcano and the pattern of its eruption.

The apparent velocity (depth versus time value) of these correlations varies form 0.2 to 2 km/day.

Research subsequently carried out on seismicity and volcanism int he New Hebrides archipelago and other arcs int he Pacific Ocean condirmed these basic correlations between feep earthquakes and volcanic eruptions. Since 1963, tests for predictions of eruptions of volcanoes have been carried out successfully in the South West Pacific.

Spatial-temporal correlations between intermediate-depth earthquakes and volcanic eruptions are also observed in the Mediterranean Sea.

In the Eolian islands, earthquakes at a depth of about 280 km precede (by about 17 months) changes of activity at Stromboli, the mean value of the apparent velocity being 0.53 km/day.

In the Eastern Mediterranean, because of the complex tectonics and the dormancy of the volcanoes, the temporal correlation between intermediate-depth earthquakes and volcanic activity is not clearly observed. However, the eruptions of Thera (Santorini) were preceded (by about 20 months) by the strongest subcrustal shocks which have taken place in the Aegean volcanic arc. The slow apparent velocity (0.25 km/day) of these correlations is in agreement with the type of eruption: viscous lava-forming domes.

INTRODUCTION

The association of andesitic volcanoes with island arcs is well known, and the close spatial coincidence of andesitic volcanoes and inclined deep seismic zones is embodied in the work of Gutenberg and Richter (1954):

"The occurence of earthquakes at depths of 100 - 250 kilometers under active volcanoes suggests a common cause, but hardly a direct relation of cause and effect. Probably a single system of stresses is responsible for both shocks and eruptions".

The composition of the lavas in volcanic arcs seems to be related in some way to the depths of the underlying Benioff zone. For Kuno (1966) close correspondence between the variations of depth of earthquake foci in the mantle and of basalt magma type in the Japanese island arcs indicates that differnt magmas are produced at different depths where the earthquakes are generated by stress release.

Dickinson and Hatherton (1967) have drawn attention to the correlation between potash content in erupted lavas of oceanic volcanoes and he vertical depth of the inclined seismic zone. They have inferred that andesitic magmas are generated at Benioff zones and rise to the ground surface without undergoing sufficient contamination to mar the pattern of chemical variations established at the sites of partial melting in the mantle.

Taylor (1960) studying the possible correlation between regional seismic activity and the incidence of eruptions at the colcanic centres of Papua and New Guinea, drew attention to the broad correlation between large volcanic eruptions and conditions of previous abnormal crustal stress.

Eaton (1962) made similar observations in Hawaï. Periodic swarms of earthquakes from a zone in the mantle about 60 km beneath the summit of Kilanea outlined a possible source of magma for the volcano. The Kilanea eruption in November 1957 followed a swarm of earthquakes about 55 km deep, between the 14th and the 19th of August 1957.

The search for direct cause and effect between tectonic earthquakes and volcanic eruptions was taken up by the author and others workers (Blot & Priam 1963) after the catastrophic "nuée ardente" eruption of Lopevi volcano in the New Hebrides, which occured without warning in July 1960.

From the observations made in the New Hebrides island arc on t e seismicity and the volcanic activity it has been possible to derive the following concept:

1. Most intermediate-depth focus shocks in the vicinity of the active volcanoes have been followed by eruptions, several months afterwards.
2. The eruption intensity is often dependent on the earthquake magnitude: intermediate-depth shocks of magnitude about 7 have preceded very large eruptions, those of magnitude about 6 have been followed by moderate to strong eruptions, those of magnitude about 5 have been forerunners of small to medium activity.
3. The time interval (t) between the intermediate shocks and eruptions for the same volcano depends on the depth (h) of focus. It varies between 4 and 12 months.
4. The apparent velocity of the respective effect v = h/t depends both on the type of volcano, and on the character of the eruption, which is often connected with the magnitude of the initial earthquake; it varies between 0.5 and 2 km/day (Blot 1972, 1976).

Another interesting result of the study of the temporal and spatial distribution of earthquakes in the New Hebrides has been the discovery of regular earthquake migrations within the inclinded seismic zone (along the sinking lithosphere, following the theory of plate tectonics) (Blot 1964, 1976).

In the same tectonic structure deep shocks signal the beginning of a migration of shocks up to the intermediate levels, producing a series of rising events and leading to a seismic crisis at the surface. The phenomena from shallow earthquakes sink back to the level of the deep shocks, and then it begins all over again. When this tectonic process is moving through the roots of a volcano (at the intermediate depth level) an eruption may be expected.

Grover extended this hypothesis to account for large shallow earthquakes by assuming convergence effects of several deep and intermediate focus earthquakes in the Solomon Islands (Blot & Grover 1966; Grover 1967).

Latter (1969, 1970) has made a computer search for significant relations between volcanic eruptions and specific earlier deep and intermediate earthquakes, and positive results emerged from this study.

Since 1963 tests for the prediction of eruptions of volcanoes have been carried out in the New Hebrides and other island arc regions, such as the Solomon islands and New Zealand. Form 1970 to 1975, 26 eruptions were forecast, of which 20 took place with an accuracy of ± 15 days (Blot 1976).

Carr and Stoiber (1973) have reviewed the recent volcanic activity in Central America and compared this to the distribution of intermediate-depth earthquakes. They noted that the appearance of concentrations of intermediate-depth earthquakes may signal that quiescent volcanoes are about to enter a period of renewed activity.

Tokarev (1967) pointed out that a positive correlation is observed in the Kurile - Kamchatka zone between eruptions and intermediate-depth earthquakes (of magnitude 6 and over). For instance an increase of seismic activity at depths of 70 - 250 km in the Northern Kamchatka proves that a relation exists between the giant eruption of the Sheveluch volcano and the displacement process in the upper portion of the Earth's mantle.

The tectonic and temporal correlations between volcanism and seismicity in island arcs are summarized in Fig. 1.

THE MEDITERRANEAN ISLAND ARCS

In the Mediterranean Sea two typical islands arc structures have been recognized:

In both areas, the disposition of foci, axes of negative gravity anomaly, and deep trenches along the outside edge of the seismic belt resemble the features of island arcs in the Pacific. Similarly, there too the active volcanoes lie above intermediate-depth earthquakes.

• In the Tyrrhenian Sea, the intermediate earthquakes are distributed along an inclined plane which deepens towards the NW, a distribution originally interpreted by E. Peterschmitt (1956) as evidence for the existence of an inclined seismic (Benioff) zone dipping under the Calabrian arc at a mean angle of about 55^o.
• In the Aegean Sea, the intermediate earthquakes lie close to an "amphitheatrical" surface which dips NNE under the volcanic arc at a mean angle of about 30^o (Papazachos & Comninakis 1971).

The two zones of intermediate-depth earthquakes deepining beneath the Calabran and the Hellenic arcs may be explained by underthrusting of the lithosphere, originating in the motion of the African plate relative to the Euro - Asiatic plate (Le Pichon 1968).

Accorind to D.P. McKenzie (1970) the spatial distribution of seismicity and volcanism in the eastern Mediterranean Sea is closely related to the exisence of two small, rapidely moving plates: the Aegean and Turkish plates, with the Aegean plate moving towards the south-west. The underthrusting og the Tyrrhenian Sea may be explained in a similar manner by a south-easterly motion of a small Tyrrhenian plate (Ritsema 1969).

The K₂O versus SiO₂ value in volcanic rocks of the volcanoes of the Calabrian and Hellenic arcs increases with the increasing depth of underlying earthquakes. This is similar to the pattern found in Pacific volcanic arcs (Ninkovich & Hays 1971).

The analogy between the Mediterranean and Pacific island arcs is also demonstrated by the time correlation between intermediate-depth earthquakes and volcanic activity.

THE ERUPTIONS OF STROMBOLI

The Eolian islands lie in the central part of the Calabrian volcanic arc. They comprise seven islands but active volcanoes are limited to Vulcano and Stromboli.

Vulcano has been quiescent since 1890 and present activity consists only of intercrateric and scattered solfataric phenomena.

Stromboli volcano is in nearly continuous activity with alternations of eruptive and quiescent stages. Jets of incandescent gases loaded with lapilli characterize these eruptions, for which the term Strombolian activity has been coined.

Most intermediate-depth focus earthquakes in the region are concentrated near this volcanic island. Form time to time paroxysms take place with large explosions and lava flows, and these paroxysmal eruptions are preceded by deep shocks under the volcano.

Table I gives the chronology of intermediate-depth earthquakes in the vicinity of Stromboli and large eruptions since 1943. Correlations inferred from these data are illustrated in Fig. 2.

In the Calabrian volcanic arc earthquakes at depths of about 280 km precede by a 17 months or so, changes of activity at Stromboli.

From Table I, the mean values of the parameters of these correlations are given in Table II.

THE ERUPTIONS OF THERA

The volcano of Thera (Santorini) lies in the centre of the chain of volcanic islands streching from Kromnynia (near Corinth) to the Bodrum peninsula of southwestern Turkey, and is the only actual active volcano in the Aegean volcanic arc.

The last eruptions of Thera were characterised by the effusion of viscous dacitoid lava, forming domes, with accompanying violent explosions.

Table III gives the chronology of intermediate depth earthquakes in the Aegean volcanic area (h ≥ 100 km, M ≥ 6 ½) during the period 1900 - 1950 and of eruptions of Thera.

From 1908 to 1918 the intermediate-depth earthquakes were not followed by any apparent activity of Thera. According to Gutenberg and Richter (1954) the epicentre determination of the earthquakes during that period have an error of ± 3^o in latitude and longiude and about 80 km in depth. After 1920 the errors were estimated to be about 2^o and 50 km respectively.

The strong intermediate depth shocks in 1923, 1926, 1937, and 1948 were followed by eruptions of Thera in 1925, 1928, 1930, and 1950.

Some features of the seismicity in the Aegean Sea, which have been noticed by various authors, are as follows:

• "In Greece the periods of greater activity are initiated by intermediate depth shocks, i.e. the seismic activity in the area of Greece is induced by preocesses occuring under the Earth's crust". (Galanopoulos 1965)
• "In the Aegean area over the observation period 1901 - 1970 the largest amount of seismic energy were released by intermediate-depth earthquakes prior to 1944... If we consider the central and southern parts of the Aegean region we observe a migration of activity during the twentieth century. At the beginning of the century the activity was concentrated at depths of 150 - 200 km; in 1935 - 1950, the foci moved to shallower depths of 60 - 100 km; and after 1950 there was an increase in crustal activity" (Kárnik 1972).

This upward migration of foci may explain the present quiescence of Thera volcano.

Considering all the intermediate-depth earthquakes detected beneath the Aegean volcanic belt during the years 1923, 1926, 1937 - 38 and 1948, a good correlation may be established between the deepest shocks (h≥150 km) and the subsequent eruptions of Thera (Table IV, Fig. 3).

The three latest eruptions have been proceded by 580 ± 13 days by shocks at depths of 160 ± 10 km.

The eruption on the 11th August 1925 followed, by a greater delay time, the shocks on 1st August 1923, possibly because of the long period of quiescence (1870 to 1925) of the volcano. According to this concept the great eruption in January 1866 would have been connected with the strong intermediate-depth earthquake:

- 1863, April 22: 36 ^o ½ N 28 ^o E  h = intermediate M = 8.5 (the preceeding eruption of Thera was in 1707 - 1711).

For the correlations between intermediate depth earthquakes and eruptions of Thera, the apparent velocity is h/t = 155/620 = 0.25 km/day

This low value may be compared with the pattern of the dome extrusion of Santiaguito at Santa Maria Volcano (Guatemala) in  1922. That noticeable eruption (June 1922) was preceded (by 1 year and 4 months or 485 days) by a remarkable intermediate-depth earthquake (the strongest deep shock observed near this volcano) on:

- 1921 February 4 : 15 ^o N 91 ^o W h = 120 km M = 7.5

- 1922 June 5 : 14,75 ^o  N 91,55 ^o W Santiaguito eruption.

In this case the apparent velocity of correlation was:

h / t = 120 / 485 = 0.25 km/day.

THE MINOAN ERUPTION OF THERA

According to archaeological research of S. Marinatos, the city of Akrotiri had been partly destroyed and reconstructed before the dramatic Minoan eruption of Thera took place.

Using the time correlations between deep shock and volcanic eruptions in the Pacific, P. Hédervari (1971) has postulated a delay time of about 10 months between the great destructive earthquake (probably at intermediate depth) and the apocalyptic eruption of Thera.

A delay time of 2 years ± ½, derived from the actual observations of seismic and volcanic activity in the Aegean volcanic arc, would be more appropriate.

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

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