Factors Controlling the Evolution of Large Explosive Eruptions

Roberto Scandone

Dipartimento di Fisica
Università di Roma Tre
Via Vasca Navale 84
00146 Roma
This version of the paper is longer than the one published on JVGR. It contains a few more examples that were cut because of editorial reasons


A simple model, based on observations of recent explosive eruptions, is proposed for the patterns of variable discharge rate during explosive eruptions. Such eruptions are characterized by a slowly increasing waxing phase and a phase of more rapid waning activity. This behavior is associated with the response of the magma chamber walls. Elastic readjustment is likely during effusive eruptions; a rigid behavior, until brittle fractures occur, is instead typical of explosive ones. The waxing phase is related to a progressive increase of the pressure gradient driving magma out of the reservoir. This gradient is controlled by the vesiculation of saturated melt that re-equilibrates the decompression of the magma chamber as soon as it is connected with the surface. The following increase of discharge permits an early Plinian phase which culminates in collapse of the eruption column and emplacement of pyroclastic flows . The emission of abundant lithic fragments, making up the roof of the chamber, signals the beginning of the waning phase. The eruption lasts until the vesiculation, caused by the pressure decrease, can counterbalance the lithostatic load. Collapse of the chamber occurs as soon as the pressure becomes lower than the lithostatic load by an amount similar to the strength of rocks. The collapse tend to re-establish the original pressure conditions preventing further vesiculation and thereby resealing the magma chamber.


An overpressure above the lithostathic value is often invoked as necessary to start an eruption (Blake, 1981, Parfitt et al, 1993). The overpressure, driving the magma to the surface, can be made available, when the reservoir ruptures, either immediately, in the form of elastic energy stored in the magma and surrounding rocks, or, with a certain delay, as stored internal energy of the magma (see for example Burnham, 1972,1985). This last mechanism occurs through expansion of the already exsolved H2O and exsolution of additional H2O upon decompression of the magma body.

I propose that the relative importance of the two processes can be analyzed by the temporal evolution of the magma discharge during an eruption, because in one case, elastic energy is, by definition, made available immediately at the start of the eruption, whereas in the latter case, stored internal energy can be made available after a measurable delay time since the beginning of the eruption.

Some basaltic eruptions show an unsteady behavior with a rapid waxing phase and a slower waning in the rate of magma discharge (fig.1a)

Fig.1 - Examples of trends of variation of magma discharge during effusive and explosive eruptions

(several cases are shown in Scandone, 1979, and an extensive bibliography is provided by Wadge, 1981, although contrasting examples have been exibithed during the Puu Oo eruption, Wolfe et al, 1988). This typical feature has been attributed to a progressive decrease of the overpressure driving the eruption from a closed magma chamber (Machado, 1974). A different pattern with a slow increase of magma discharge and subsequent rapid waning, is observed in the course of explosive events (fig.1b). This pattern has been observed during the eruptions of Mt St Helens, El Chichon and Pinatubo, having a Volcanic Explosivity Index (Newhall and Self, 1982) in the range of 4-6, but also several other smaller eruptions like the ones of Pavlov, Redoubt, and Oshima display a similar behaviour (McNutt,1987, Aramaki, 1988, Power et al, 1994).

In these cases the associated deposits have a typical succession of facies (Sparks et al 1973, Walker, 1985). The beginning of eruption is always characterized by vent opening and a plinian phase with a reversely graded, pumice-fall deposit. This phase is followed by a wavering plinian stage with partial column collapse and intraplinian ignimbrites. The climactic stage is reached with the eruption of an ignimbrite characterized, in proximal areas, by the deposition of a lag-breccia deposit. Overall, this sequence suggests a progressive increase of magma discharge rate.

The most voluminous and intense eruptions cause the formation of calderas. Scandone (1990) suggested that the mechanism of caldera formation is due both to the size and aspect ratio of the magma chamber as well as the nature of the rocks making up the roof of the chamber. Scandone (1991) further hypothesized that the cause of ignimbrite emission could be related to a decompression of the magma during eruption with a consequent increase of the pressure gradient driving the magma to the surface.

Factors which are likely to increase the pressure gradient include:

i) An increase in fragmentation depth, reducing column height and magma column pressure (Sparks, 1978, Scandone and Malone, 1985).

ii) Vesiculation of the column reducing the density and magma column pressure (Druitt and Sparks, 1984).

iii) Vent erosion at the surface allowing the exit pressure to decrease and therefore reducing the overall pressure gradient (Wilson et al, 1980, Papale and Dobran, 1994).

iv) Delayed bubble growth (Proussevitch et al, 1993).

In this paper I will examine in more detail the delayed bubble growth hypothesis and show that this process is the main factor controlling the temporal evolution of large-volume explosive eruptions. A qualitative model that explains the typical facies succession will be derived taking into account both the geological sequences associated with explosive eruptions as well as the typical seismicity related with these events.

The model is based on phenomena observed in the course of the eruption of Mt St Helens in May, 1980. It will be shown that the same features have been observed in the course of several other monitored eruptions (El Chichon,1982, and Pinatubo, 1991).

2-The model

The deposition of the typical sequence of an ignimbrite has been explained as due to an initial Plinian eruption column which later collapses giving rise to pyroclastic flows or surges (Sparks and Wilson, 1976, Sparks et al, 1978). Reverse grading of the pumice deposit, as well as the generation of ignimbrites, suggest an increase in the discharge rate of magma until the emission of the pyroclastic flow, followed by a rapid waning in discharge rate. Fluid-dynamical models, based on the analysis of quasi-steady phases of eruptions, have explained that an increase in magma discharge can actually lead to the collapse of the eruption column (Wilson et al, 1980, Woods, 1988).

2.1 Pressure gradient increase: relevance of changing vent geometry and properties of the magma column in the conduit

Wilson et al (1980, 1981) suggest that conduit geometry and magmatic gas content play an important role in the overall evolution of events during an eruption. The common occurence of lithic material within the pumice deposit has been taken as an evidence of conduit erosion during eruption. This produces a progressive widening of the conduit, an increase in mass discharge rate, and the consequent collapse of the Plinian column.

Varekamp (1993) suggests that an increasing content of lithics in the deposits does not necessarily implies a widening of the conduit. An erosion of the conduit causes an increase in magma discharge rate which scales as R4 (R= radius of the conduit), whereas the lithics increase at a smaller rate; as a consequence, the erosion of the conduit should result in a decrease of the weight percent of lithic fragments in the deposits. The actual increase of lithics observed with increasing mass discharge is presumably related with catastrophic modification of the lower conduit section related to incipient caldera collapse (Varekamp,1983).

Conduit erosion and widening of the vent is a requisite for eruptions with supersonic exit velocities for the gas-pyroclast mixture. Erosion probably occurs in the upper conduit, whereas the deep parts of it, below the fragmentation level and the exsolution level, are likely to remain of the same size until collapse occurs.

No relevant increase of discharge is allowed if a large part of the conduit remains of the same section, unless it is created a substantial change of the pressure gradient driving the flux. A non-equilibrium, two-phase model of magma flow along a volcanic conduit shows that the variation of exit pressure with changing eruption conditions, during the May 18, 1980 eruption of Mount St Helens, was relatively small (10-20 bar) (Papale and Dobran, 1994). These same authors (Papale and Dobran,1994) have also shown that the variation of the bubbly flow region in the course of the eruption has been of the order of 100-200 m, not sufficient to justify relevant change of the pressure gradient. In order to explain the large variation of magma discharge in the course of this eruption we have to invoke a different cause.

2.2 Mechanisms driving the initial phases of Mt St Helens eruption

Several driving mechanisms may initiate an eruption:

1) exsolution of a volatile phase or intrusion of new magma into a reservoir (Blake, 1981,1984, Tait et al 1989) with development of an overpressure above the strength of the rocks surrounding the magma chamber;

2) Instability of fluid-filled cracks driven by the Peach-Koehler pseudo-buoyancy force (Weertman, 1971) due to exsolution of a gas phase or crack propagation by stress corrosion (Anderson and Grew, 1977);

3) Buoyancy of liquids lighter than surrounding rocks (Ryan, 1987) with development of stress above the strength of the rocks.

The basic difference between the first mechanism and the other two may be established by the ground deformations preceeding the eruption. In the first case, the deformation geometry is correlated to a source fixed in space but with increasing intensity (an increase of the width and amplitude of the deformation as pressure increases). In the latter cases, the deformations are due to a source moving toward the surface (the width of the deformation decreases on approaching the surface). Deformation sources of the first type may be those observed before the Rabaul eruption of 1994, or the seismic crisis of Campi Flegrei in 1983-84 (McKee et al1985, Bianchi et al, 1987). An example of deformations related to a source approacing the surface is provided by the vertical deformations before the 1944 eruption of Usu volcano (Minakami et al, 1951). In both cases, the depth of the pressure source may be estimated by the geometry of the deformation.

Geodetic measurements made prior to the May 18, 1980 eruption of Mt St Helens indicated a shallow source of deformation of the northern flank of the mountain (Lipman et al, 1980). Contrary to this trend, the horizontal deformation measured before and after the explosive eruptions following the one of May 18, 1980, were detected on the external network of observation thus indicating a deeper source of deformation (Swanson et al, 1981). These observations are of crucial importance, as they indicate that the magma chamber, whose top was at 7-10 km (Scandone and Malone, 1985, Rutherford et al, 1985), was not responsible of the surface phenomena observed in the weeks before the May 18 eruption. The phenomena immediately preceeding the eruption were likely due to the intrusion of the cryptodome already detached from the main magma body because of its own buoyancy. The role of the deeper reservoir became evident only later, in the course of the paroxismal eruption.

2.3 Chamber wall behavior during the eruption

Scandone and Malone (1985) analyzed the seismicity occurring mostly after 17:49 of May 18 and identified an earthquake-free volume, whose top was at about 7 km depth, interpreted as the magma chamber of the volcano. Barker and Malone (1991) further developed this idea by analyzing the focal mechanism solutions of post-eruption earthquakes. They found that the earthquakes were caused by a stress concentration caused by a perturbation of the regional stress field at a depth of 7-11 km, explained by a decrease in pressure within a cylidrical magma chamber.

A detailed analysis of the readjustment of the chamber wall during the first day of eruption is obtained by the rate of earthquake occurrence on May 18 and 19 following the onset of the eruption. Shemeta and Weaver (1986) subdivided this time interval into five periods as shown in Table 1.

Table 1

Period Time (PDT)
1 08.49- 11.30
2 11.30- 14.15
3 14.15- 16.30
4 16.30- 17.30
5 17.30- 0830

During the first two periods (fig.2),

Fig. 2 - Depth-time distribution of earthquakes in the first day of May 18, 1980 eruption of Mt St Helens. The vertical bar indicates the location of the magma chamber (slightly modified after Shemeta and Weaver, 1986)

the earthquake release rate was low (low cumulative seismic moment) and the events were located in a volume below the volcano at depths ranging between 3 and 9 km, most earthquakes occurring between 3 and 6 km, with only a few events as deep as 12 km. During period 3, there was the maximum release of seismic energy (about 3/4 of the total), a scatter in epicentral distribution as wide as 3 km, with most events occurring at depths between 2 and 8 km, and only a few between 8 and 12 km. During periods 4, and 5 there was a decrease in the release of seismic energy, and a deepening of events (4-14 km) with a lack in the shallow portions of the system. After 19:00 of May 18 nearly all events were located at depth between 5 and 12 km.

The corresponding pattern of volcanic activity is analyzed by Criswell (1987) and Carey et al (1990). There was a progressive increase in the magma discharge rate from the end of the lateral blast until about 15:00-16:30 of May 18. The initial plinian phase occurred between 09:00 and 12:15; the progressive increase of magma discharge led to the generation of pyroclastic flows at 12:15; the maximum rate of magma output occurred between 15:00 and 16:30. At 16:35 there was a decrease in magma discharge rate and a return to a plinian phase. At 11:40 the seismographs around Mt St Helens began to show a a gradual increase in strong ground shaking and by 13:30 it was impossible to discriminate individual earthquakes (Foxworthy and Hill, 1982, Scandone and Malone, 1985).

Fig. 3 - Variation of the main physical parameters of the eruption of 18 May, 1980 of Mt ST Helens. Tremor amplitude (top) (Scandone and Malone, 1985); Seismic strain release (middle) (Shemeta and Weaver, 1986); Magma Discharge rate (bottom) (Carey et al, 1990)

Fig.3 shows the general pattern of the eruption described by these different observations. It is similar to fig. 3 of Criswell (1987), but the magma discharge is modified to take into account the observations of Carey et al (1990).

I propose that the low rate of occurrence of earthquakes during the plinian phase, as well as the low amplitude of ground shaking, as an evidence of the absence of collapse. The increase of magma discharge is indicative that there was no stored elastic energy immediately available at the beginning of the eruption. These two features both indicate a rigid behavior of the wall of the chamber.

2.4 Pressure in the magma chamber

A first decompression of the magma chamber occurs in response to the sudden removal of the top part of the mountain as a consequence of the landslide of the north flank of Mt St Helens. The pressure release is of the order of 10-20 MPa (Kieffer, 1984).

Assume that the pressure exerted by the magma on the reservoir walls, before the eruption, is equal or higher than the lithostatic load. The magma is decompressed as soon as it is connected, through a conduit with the surface (because it 'sees' an infinite volume, or some magma is evacuated from the chamber. The amount of decompression is either equal to the lithostatic load, if no magma is present in the conduit, or it is equal to the difference between the lithostatic load and that of a column of magma filling the conduit (the buoyancy of the magma). I transcure, in a first approximation,the compressibility of the magma which is small compared to that of surrounding rocks (Blake, 1981). As long as the chamber does not collapse (or else, the chamber walls behave as a rigid envelope) there is no surrounding pressure acting on magma and, as long as bubbles do not nucleate, no pressure is exerted by the magma on the overlying rocks. Even in the case of limited collapses, the pressure exerted on the magma is no more lithostatic. The depressurization of the reservoir is like the opening of an underground cavity. The top beds of the chamber fall and the upper beds sag down onto them and gradually form an area of relaxed roof spanned by transverse arches (Thomas, 1973). The pressure acting onto the magma body is different from the pressure that was exerted before the drainage. The pressure is equivalent to that of a block of rock of height H=a(h+l), when the depth of the opening is >1.5 (h+l), where h is the height of the opening, 'b' is the width, and 'a' is a coefficient varying between 0-0.5 for hard rock to 2.-4.5 for incoherent material (Cestelli Guidi, 1980).

The decompression causes exsolution of volatiles, if magma is near to volatile saturation, and the magma responds to depressurization by growing bubbles, increasing its volume and attempting to re-establish the original lithostatic pressure. The flow of magma occurs as soon as a sufficient pressure gradient is re-established between the chamber and the surface and the flow rate is basically controlled by the bubble growth rate. The maximum flow rate is attained when the pressure gradient is again nearly equal or slightly less than the difference between the lithostatic pressure and that of the overlying column of magma. This gradient is sufficient to drive magma flow along a conduit as shown by Wilson and Head (1981), Papale and Dobran (1994) even without invoking overpressures of the magma chamber.

2.5 Bubble growth rate

How fast does the chamber responds to decompression? The vesiculation may not occur instantaneously, but depends on the nature of magma, amount of oversaturation and ambient pressure (Proussevitch et al, 1993). Proussevitch et al (1993) made a parametric study describing diffusion-induced growth of closely spaced bubbles in magmatic systems; I report here their main results as they are relevant to understand the factor influencing the pattern of an eruption. The numerical modelling corresponds to an instantaneous depressurization of a saturated magma and subsequent bubble growth (Proussevitch et al, 1993, pag 22297). Upon decompression, the bubbles start to grow after a variable delay time due to high surface tension within small bubbles. The bubbles do not grow indefinitely because there is a limited amount of volatiles in the surrounding melt and they reach their final sizes when the pressure inside the bubbles equals the saturation pressure. High volatile oversaturation causes a faster growth rate. Other things being equal, liquid with basaltic composition have a faster growth rate than liquids of rhyolitic composition. The ambient pressure is the factor which predominantly controls the growth rate. The time for complete growth of bubbles in a silicic melt varies from a few seconds at atmospheric pressure to days or also a few weeks at pressure of 200-400 MPa.

Fig. 4 - Diffusive growth of bubbles in rhyolitic melt for different values of ambient pressure from 0.1 to 400 MPa (saturation pressure for 8 w % of water). (Redrawn after Proussevitch et al, 1993)

I report in fig. 4 the results of Proussevitch et al, (1993) relative to the influence of the ambient pressure. It is possible to see that the bubbles attain their final sizes in a matter of a day, at pressure corresponding to about 7 km. A time delay of about 7 hours between the beginning of the eruption and the attainment of the highest output rate at about 15:00 is perfectly compatible with a delayed bubble growth if we take into account the lower silicic content of the dacitic magma of Mt St. Helens. In conclusion it is perfectly plausible to think that the magma is driven out of the chamber because of its increase in volume due to bubble growth exactly like opening a bottle of champagne.

2.6 Collapse of the chamber

We can estimate, through mass conservation, the volume DV of drained magma producing a pressure decrease equal to the strength of the material (~20 MPa) in a magma chamber containing a volume V1 of melt:


where r1 and r2 are the density of the original magma and of the magma plus bubbles respectively.

The density of the magma plus bubbles can be estimated (Wilson and Head (1981) according to:

where n is the weight fraction of exsolved gas, and rg , rl are the density of gas and liquid respectively. The gas density can be estimated through the equation of state of gas, assuming a good thermal contact between gas and magma:

R= 8.3143 JK-1mol-1 is the universal gas constant and m is the molecular weight for water (=0.018kg).

The content of water for the mafic dacite of St Helens has been estimated by Rutherford et al (1986) at 4.6 wt %. This estimate is between the value reported by Melson and Hopson (1982) of about 7%, and that of Eichelberger and Hays (1982) of 0.7-1.7. I will hypothesize that the magma was saturated, and use, for modeling purposes, the exsolution law for water provided by Sparks (1978) for rhyolite, as there is no equivalent law for mafic dacite. The weight % of water dissolved in the melt is given by (Sparks, 1978):

n=4.11*10-6P0.5 (P in Pascals)

A rhyolitic melt at a pressure of 185 MPa (depth of ~7 km) has a saturation content of 5.6 wt % of water. A pressure decrease of the order of 20 MPa (corresponding to the strength of rocks) produces an exsolution of about 0.3 wt% of water, causing a density decrease from 2400 to 2350 kg/m3. The amount of magma necessary to be drained to produce such a pressure decrease is approximately 2% of the original volume. This value gives an estimate of 12.5-25 km3 for the volume of the Mt St.Helens magma chamber, by the volume of products erupted during the May 18 eruption (0.25-0.5 km3) (Lipman et al, 1981b, Criswell, 1987). The value agrees well with the estimate of 10-20 km3 made by Scandone and Malone (1985) of the earthquake free zone considered to be the magma reservoir.

This rough estimate supports the idea that delayed vesiculation in the chamber, the amount of saturated melt and the rate of drainage are the most important factors affecting the development of the eruption.

The growth of bubbles tends to re-establish the original magmatic pressure (lithostathic or higher), causes a steady increase in the discharge of magma until, eventually, conditions are attained to produce pyroclastic flows by non-convecting eruption columns. The process can last as long as the chamber behaves as a rigid envelope and as long as vesiculation of the water-rich melt occurs. The magma chamber becomes progressively filled with an higher proportion of gas and the density of the chamber may become sufficiently low that it becomes mechanically unstable (Tait et al, 1989).

A slow rate of drainage is likely to cause an elastic readjustment of stress, whereas rapid drainage favors a brittle behavior until fracture occurs. Such a condition was observed for the May 18, and May 25 eruptions of Mount St Helens in 1980, during which a high number of deep, high-frequency earthquakes, were observed. In contrast, only a few, deep earthquakes were observed during the later eruptions, which were characterized by a low drainage.

2.7 Deposit associated with the collapse

The collapse of the magma chamber, as soon as the pressure difference between magma and lithostatic load reaches the strength of the country rock, tends to re-establish the lithostatic load. Collapse may not occur in a single phase as arching of the roof eventually determines a confining pressure less than the original one, but, as a rule, collapse is followed by a rapid waning in the rate of discharge. The violent earthquakes occurring around and immediately above the magma chamber possibly cause extensive fracturing and production of lithic breccia, emitted during this phase of the eruption (Walker, 1985, Scandone, 1990, Varekamp, 1992).

During the 18 May eruption of Mt St.Helens, the emission of lithic breccia (lithic breccia zone of Criswell, 1987) between 15:00 and 16:30 occurs contemporaneously with the highest seismic energy release and peak magma discharge (Carey et al, 1990). After this, there is a decrease in seismic energy release and magma discharge.

2.8 Harmonic tremor

Strong harmonic tremor was measured throughout the eruption (Malone et al, 1981). After 11:40, the tremor rapidly increased and saturated the entire seismic network of Washington State. Harmonic tremor is the most elusive seismic signal recorded in a volcanic environment. It is a continuous ground shaking with frequency content similar to that of low-frequency earthquakes (1-5 Hz), and several models have been proposed to explain its origin. Aki et al (1977) proposed that tremor may originate by the repeated opening of cracks filled with magma. This model was developed to explain episodes of tremor at the basaltic volcano Kilauea in Hawaii. However, a viscous dacitic magma, like that erupted at St.Helens, is more likely to flow in conduit-like passageways than in fluid-filled cracks.

Crosson and Bame (1985) have proposed a simplified model based on a magma chamber filled with a gas phase. This model is able to produce a resonant radial oscillation that is dependent on the existence of a froth region inside the magma chamber and which explains many of the characteristics of harmonic tremor. The frequency is relatively insensitive to the size of the magma chamber. The synchronous oscillation of the magma chamber and surrounding rocks can exist at low frequency only when the froth region is sufficiently large. As magma is withdrawn, the amount of magma versus gas in the chamber will change, with a consequent change in the seismic wave generation because of the large acoustic impedance between the magma-gas mixture and the country rock. Although, I am not aware of systematic changes in the character of the tremor during the eruption of St Helens, a systematic change of the frequency of tremor was observed, for example, during the 1944 eruption of Vesuvius at the transition between the low rate effusive phase and the higher rate explosive one (Imbò, 1952).

According to the present model, the appearance of strong harmonic tremor during the May 18 eruption is indicative of extensive bubble nucleation within the magma chamber; drastic reduction of tremor occurs after 16:15 when the lithostatic pressure is restored by the collapse of the walls.

3- Other examples of monitored large explosive eruptions

3.1 El Chichon, March-April, 1982

Precursory seismicity was observed starting on March 1, 1982 (Havskov et al, 1982, and Gutierrez et al, 1983) with events similar to the B type of Minakami (low frequency) but with a wide range of magnitudes and readable S-phase. All the events until 10:15 March 27 (time in GMT) were of this type. The activity slightly subsided with the occurrence of the largest of these earthquakes on March 6 (M=4.0). On March 27 there was the first occurrence of clear long period events (type 2); they declined rapidly and, on March 28 at 06:00, there was the beginning of type 3 events (long period with lower frequency than type 2). The seismic activity declined suddenly at 03:27 on March 29 and was followed by an eruption at 05:15 on March 29. Seismic activity resumed at 13:00 on March 30 with type-3 events and remained high for 20 hours. Two small eruptions occurred at 08:40 and 10:03 on April 3. There followed a quiescence of 1.5 hours after which seismic activity again increased, with type-2 events that climaxed and then subsided. A large eruption occurred at 01:39 on April 4 followed by a few type-4 (high frequency, tectonic) events. Another large eruption occurred at 11:10 the same day. During the final stage of this eruption, and immediately afterwards, there was an intense swarm of type-4 events (approximately 300 events/hour). The two largest (M=4) occurred at 11:47 and 12:01 on April 4. The number of events decayed with time; by April 5, the number was 4/hour and continued to decrease until 16 April. The probable depths for different events are: type 1 = ~5km; type 2 = ~2km; type 3 32 km ; and type 4 =~15-20.

It is proposed that the entire sequence of eruptions should be considered as a unique event with a progressive increase of pressure in the chamber until 01:39 on April 4. This eruption is characterized by the maximum discharge (1.9x108 kg/s during the plinian phase, Carey and Sigurdsson, 1986). The episode began with the generation of a pyroclastic surge and emplacement of a debris flow, then a plinian phase, and new surge deposition. Minor collapse in the chamber may have occurred during this phase as testified by debris flow emplacement, the high lithic content of the plinian-fall deposit and the few, high-frequency, deep earthquakes. This possibly resulted in minor vesiculation, and the following eruption at 11:22 on April 4 was characterized by a lower discharge rate (1.3x108 kg/s during the plinian phase, Carey and Sigurdsson, 1986). The chamber was finally closed by an extensive collapse, testified by the swarm of deep, high-frequency earthquakes.

3.2 Mt Pinatubo, May-June 1991

At the beginning of May 1991, several small explosions occurred from a series of vents on the north flank of the summit dome of Mount Pinatubo, Philippines (Pinatubo Observatory Team,1991, and Wolfe,1992). Small high-frequency earthquakes continued throughout the month in a zone about 5 km north, north-west of the summit and at depths between 2 and 6 km. A new cluster of earthquakes was recorded on June 1, at 1 km north-west of the summit and at depths of less than 5 km. Small explosions occurred on June 3 with emission of ash and episodes of harmonic tremor.

The seismicity increased until the afternoon of June 7 when an explosion generated a column of steam and ash, 6-7 km high, and was followed, on June 8, by the extrusion of a small lava-dome close to the fuming vents. Weak ash emission and episodes of harmonic tremor continued until June 12 when a burst of intense harmonic tremor (40 minutes long) occurred at 03:01. An ash cloud, 3 km high, was seen on the following morning.

The first major steam eruption occurred at 08:51 on June 12, generating an ash cloud, 19-km high (accompanied by a high amplitude seismic signal lasting 35 min). Small pyroclastic flows were emplaced north and northwest of the vent. This was the first of a series of brief explosive eruptions which occurred until June 15.

On the morning of June 14 a swarm of long-period earthquakes were registered without detectable explosive activity. Violent eruptions occurred at 13:09 on June 14, with the seismic activity characterized by high frequency events as well as long-period events, and episodes of harmonic tremor. Large brief explosive eruptions occurred throughout the night and the morning of June 15 (probably issuing from two or more vents).

The climax of the eruption began at 05:55 on June 15, with the emission of a relatively low (12 km) ash cloud spreading laterally. There followed at least seven pulses as activity evolved into a continous strong eruption by the early afternoon; seismic signals saturated the instruments at 13:42. By 14:30, only one seismograph was still functioning.

The first of a series of strong earthquakes was felt at 15:39. These recurred throughout the night and at least 17 earthquakes with magnitude between 4.4 and 5.6 accompanied and followed the eruptions while hundreds had magnitude greater than 3. The next day, earthquakes (high frequency) larger than magnitude 1.5 occurred at a rate of 150 per hour. This rate rapidly declined, and, by the end of June, was 10-20/hour.

As in the previous examples, the Pinatubo eruption was characterized by a slow build up of activity with an increasing tempo. The occurrence of strong tremor and long period earthquakes marks, in my opinion, the approach toward the climactic phase indicating extensive vesiculation in the chamber.

Importantly, the eruptions at 22:52 June 12 , and at 08:41 June 13 were preceded by 2 to 4-hour swarms of long-period earthquakes. Seismicity was resumed with long period earthquakes on the evening of June 13. The first long period events occurred on May 25 and June 9-10; the first swarm occurred after the eruption at 0851 of June 12 . After the explosion at 08:41 on June 13, long period events began to increase dramatically in frequency and size.

Although the duration of the climactic phase is different, the pattern of the eruption is similar to that of Mount St.Helens. Long period earthquakes and tremors occurred more frequently as the phase of maximum discharge was approached. The eruption evolved from a plinian phase to the emission of pyroclastic flows in accordance with the increase in the discharge rate. In addition, as also occurred at Mt St Helens, lithic-rich debris were emitted with the last pyroclastic flow at the end of the climactic phase (Scott et al, 1993). High frequency earthquakes, possibly around the magma chamber, characterized the last part of the eruption in the afternoon of 15 June and continued for several days (fig.5).

Fig. 5 - Comparison of post eruption seismicity of Mt St Helens (top) (Scandone and Malone, 1985) and Pinatubo (bottom) (Wolfe, 1992). Note that in both cases, the collapse of rocks defined by the seismic zone around the seismic-free magma chamber occurred toward the end of the eruption.

According to Scott et al (1993) the waning phase is associated with the foundering of the caldera which produced abundant lithic clasts, and effectively sealed the conduit.

4-Ancient analogues of Mount Vesuvius

Mount Vesuvius had violent eruptions during its history like the one of 79 A.D. which destroyed Pompei and Herculaneum, and like the one which occurred in 1631 (Rosi et al, 1993, Rolandi et al, 1993). Although these events were not monitored with modern techniques, they were all observed from a close vantage point (like the city of Naples, 12 km away from the volcano) and the main features of the eruptions were recorded: more than two hundred contemporary reports exist for the 1631 eruption.

The 79 A.D. eruption began with a sustained eruption column (vividly described by Pliny the Younger) after a precursory seismic phase. Pliny testifies, in his second letter, that continuous earthquakes were felt at Misenum (22 km from Vesuvius) during the "Plinian" phase of the eruption, and that these were so strong that carriages could not stand still even on flat ground. He also observed the transition to pyroclastic flow emplacement: "From the other side, black and horrible clouds, broken by sinuous shapes of flaming winds, were opening with long tongues of fire (..) After a little while descended onto the land, opened the sea, covered Capri and prevented the sight of Misenum ". At the end of the climactic phase, Pliny returned to Misenum but still feared for the occurrence of many earthquakes. The succession of the products observed in many outcrops around the volcano confirms the words of Pliny. Reverse grading of the pumice fall deposit of the 79 A.D. eruption indicates an increasing discharge rate which eventually led to column collapse and emplacement of pyroclastic flows and surges (Lirer et al, 1972, Sigurdsson et al, 1985). After deposition of the last surge (S-5 of Sigurdsson et al, 1985) there was a sudden increase to over 70 volume % in the proportion of lithic ejecta mostly of deep seated provenance (marble, skarn, cumulate Barberi et al, 1989). The eruption ended with the emplacement of a lithic-rich debris flow (Sigurdsson et al, 1985) and a final phreato-magmatic phase producing deposits of alternating pisolitic ash and cross laminated sands (Barberi et al, 1989).

A similar sequence of events was observed during the 1631 eruption. Activity began with the formation of a plinian column after several days of earthquakes felt all around the volcano. A peculiar seismic signal accompanied the sustained eruption column and several witnesses reported that continuous earthquakes could be felt in Naples (Rosi et al, 1993). Giuliani (1632) states that " a small but continuous tremor of the houses was felt in Naples (...) it grew in such a way that everyone thought that they were to die". A more violent earthquake was distinctly felt at the climax of the eruption and it accompanied the emplacement of pyroclastic flows and lahars (Giacomelli and Scandone, 1992, Rosi et al, 1993, Rolandi et al, 1993). Giuliani (1632) also states that, after these events, more than a hundred earthquakes were counted and were so long and strong that the Neapolitans again feared the collapse of their houses. The deposits of the eruption (Barberi et al, 1989, Rosi et al,1993, Rolandi et al, 1993) confirm the contemporary reports and have been subdivided into different units by Rosi et al (1993). A reversely graded fallout deposit was produced from the early Plinian phase. The amount of lithics increases toward the top of the sequence and the lithic-rich fall-out (Unit IV of Rosi et al, 1993) contains approximately 47 wt % of lithics of deep provenance. This deposit marks the transition to ash flows and surges. The eruption ends with a phreato-magmatic phase and a debris flow characterized by abundant lava lithics and sedimentary blocks (Rolandi et al, 1993).

Additional strong evidence of magma supersaturation in volatiles has been provided by Belkin and De Vivo (1993). They found primary H2O-CO2 fluid inclusions in nodules ejected during the Plinian eruptions of Vesuvius, thus indicating the presence of a supercritical H2O-CO2 fluid in the nodule-forming environment.

5-A remarkable exception: Hekla volcano, Iceland

A remarkable different trend is observed during the eruptions of Hekla in Iceland. The large eruption of 1947-48, as well as the relatively minor events of 1970, 1980-81, and 1991 begin with a violent explosive phase with peak discharge rate right at the beginning of the eruption (Einarson, 1949, Thorarison and Sigvaldason, 1972, Gronvold et al, 1983, Gudmundsonn et al,1992).

The eruption column of the 1947 eruption reached 26 km (Porarinsson, 1949) in a matter of minutes after the beginning of the eruption at about 06:50 on 29 March, 1947, with a mean output of 13500 m3/s of magma (DRE) in the first two hours of eruption. The tephra phase was followed by lava emission steadily decreasing during the following 20 hours with an average effusion rate of 1350 m3/s of magma (DRE) (Einarson, 1949). On the second day the effusion rate was reduced to about 100 m3/s and to a few m3/s in the following months with occasional surges again to 50-100 m3/sec in June. The total volume of lava erupted during the eruption of 1947-49 was estimated in 0.8 km3 of lava and 0.215 of tephra (Gronvold et al, 1983).

The same pattern, even if at a smaller scale, was observed during the eruptions of 1970, 1980-81 and 1991.

It is important to note that all these eruptions were not preceeded by long-term precursory seismicity or deformation, with only a few earthquakes recorded minutes before the eruptions (Porarinson, 1949, Thorarinson and Sigvaldason, 1972, Gronvold et al, 1983, Gudmundson et al,1993). Linde et al (1993) measured a strain pulse at strainmeters located 45-15 km from the volcano only 30 minutes before the beginning of the 1991 eruption. This pulse was interpreted as due to the upward migration of a dyke and the concomitant deflation of a magma chamber whose top was at 4 km from the surface. The purely elastic model predicted a pressure decrease of 14 MPa as a consequence of the drainage of approximately 0.1-0.15 km3 of magma effectively erupted onto the surface. Also the seismic activity related to the volcano and following the eruption was limited (Stefansson et al, 1993) thus supporting the idea of a purely elastic readjustment of the feeding system. The model of the feeding system proposed by Gudmundsson et al (1993) envisages a density stratified reservoir, whose top is located at a depth of 7-8 km, and connected to the surface by a zone of crustal weakness that occasionaly opens as a dyke.

The large initial eruption rates of both differentiated magma and lava are likely caused by an overpressure in the reservoir. Such overpressure is however rapidly accommodated by the drainage of magma and elastic deflation of the chamber with an exponential waning of the activity.

A purely elastic model is also used by Stasiuk et al (1993) to account for the variation of flow rate in non-explosive eruptions. They explain departures from the trend of exponential decay of flow rate as due to the relative changes in viscosity, conduit dimensions and thickness of lava over the vent. They estimate that the overpressure necessary to drive the eruptions of several volcanoes is of the order of 10-20 MPa.

Paradoxically, even if the explosive phases of Hekla volcano produced many damages to Icelandic economy, its behaviour is typical of an effusive volcano.


With notable exceptions, the temporal evolution of some major explosive eruptions show a similar succession of phenomena. The main phases are :

1) a waxing phase, during which the mass-flow rate, on average, slowly increases until it reaches a climax. This phase may be further subdivided into a (1a) Plinian stage, and (1b) Pyroclastic flow stage, depending on the rate of discharge.

2) a waning phase accompanied by deep earthquakes, with a fairly rapid decrease of mass flow rate and possibly of all volcanic phenomena.

The different lengths of the phases and their specific character is accounted for by secondary processes that may influence the overall time trends of the eruption.

1) Waxing phase

The paroxysmal phase, which develops with increasing violence, is almost always characterized by the formation of a Plinian column and the emission of pumice. There is a tendency for the mass-discharge rate to increase, even if single episodes may depart from this trend. The phase culminates with the voluminous discharge of pyroclastic flows, derived mostly from non-convecting ash emission because the peak discharge rate erupts too much material to be effectively transported by convection (Wilson et al, 1980). The duration of this phase may be several hours, or even consist of different pulses on separate days. The deposits contain a significant amount of lithics indicating an increasing rate of dismantling of the volcanic structure. The voluminous emission of material is accompanied by a significant release of seismic energy from progressively deeper earthquakes.

The waxing phase is interpreted as indicating the formation of a conduit with a stable shape between the magma chamber and surface. The rigid behavior of the reservoir and the connection with the surface allows the magma to depressurize and vesiculate in a manner similar to the opening of a bottle of champagne. The vesiculation tends to re-equilibrate the magma pressure with the lithostatic load, driving the pressure gradient between the magma and the surface to the lithostatic value. The rate at which the vesiculation process occurs, depends mostly on the silic content of the magma, the degree of saturation, and mainly on the depth of the reservoir. The relatively slow increase of magma volume in the chamber drives the gas-magma mixture out of the chamber at a faster rate which reaches its highest value when the bubbles attain the maximum radius compatible with saturation pressure.

2) Waning phase.

The occurrence of relatively deep, high-frequency earthquakes and the emission of lithic-rich products marks the beginning of a phase characterized by a rapid decrease in the rate of magma discharge. This is interpreted to coincide with the inward collapse of the magma chamber. The collapse has two major consequences: the first is the obstruction of the conduit with lithic debris some of which is eventually erupted. The second is the return to lithostatic pressure in the magma reservoir thus preventing further vesiculation.

The collapse begins when there is a relative large percent of the volume of the reservoir filled with gas. The collapse tends to restore an external pressure acting on the chamber with values comparable with the original lithostatic one.

It is emphasized that this model allows the pressure within the reservoir to vary with time. In some models, the magma flow is driven by a static pressure gradient between the chamber and the surface (given for example by the buoyancy term). The assumption is equivalent to a perfectly elastic medium which, by definition, instantaneously transmits the load of the overlying rocks and give rise to a continuous squeezing of the chamber (tooth-paste model).

In the present model (champagne-bottle model), the chamber walls behave for some time as a rigid envelope. In a simplistic way, as soon as a direct connection with surface is established by, for example, a sudden opening of cracks, the magma "sees" no lithostatic pressure, but attempts to re-establish it by expanding and exsolving bubbles.

The difference with the previous case is that the presssure gradient, necessary to drive the mamga out of the chamber, is not immediately available, but is established by the internal energy of the magma through bubble nucleation. Depending on cases, the volume of the liquid+gas in the reservor grows at a finite rate during the course of the eruption.

The collapse of the chamber does not occur at the beginning of the eruption because of the relative incompressibility of the magma, and as long as the lithostatic load can be counterbalanced by a sufficient exsolution of volatiles.

The fundamental difference with effusive eruptions is in the mechanical behavior of the wall rocks. This difference could be ascribed mainly to two factors: the ratio between the rate of readjustment and drainage, and the shape of the reservoir.

Lines of evidences for the first case are provided by the eruptions of Mt St Helens following the May 18 and May 25 eruptions. These eruptions had a magma drainage less than that of the preceeding events and relatively few deep earthquakes.

The shape of the chamber and of the feeding system (likely in the form of a magma filled crack) may have played a relevant role during the eruptions of Hekla when even high rates of magma discharge were accommodated by an elastic deflation of the feeding system. Alternatively there might have been a very fast bubble growth because of the magma composition.

The rapid initial waxing phase of effusive eruptions may be interpreted as due to the opening of a magma-filled crack as soon as it reaches the surface. The high initial discharge is immediately provided through elastic readjustment by the external stress field. The slow waning is due to the exponential decrease of the overpressure with discharge rate (Machado, 1974, Scandone, 1979, Wadge, 1981, Stasiuk et al, 1993).

The mechanical behavior of the chamber walls plays a fundamental role, in that high tensile strengths and brittle behavior permit more voluminous and intense eruptions as hypothesized in the formation of North American calderas (Walker, 1984, Scandone, 1990). The strong difference in mechanical behavior may be ascribed to the relative maturity of the chamber as, cooling of the liquid and crystallization at the walls may provide a carapace which behaves as a rigid envelope. The occurrence of an high velocity layer at the top of Mt St Helens magma chamber (Lees, 1992) is indicative of this occurence.

Explosive development of an eruption may also be due to how fast a connection between the magma chamber and the surface is produced. A large overpressure in the chamber possibly may cause a catastrophic failure of roof rocks with sudden depressurization of the magma.


The pressure gradient driving magma out of the chamber may be lithostatic or higher. The mechanical properties of the rocks surrounding the magma chamber control the rate at which it is made available. In the case of a purely elastic response, it is made immediately available at the beginning of the eruption and will result in an high, initial discharge rate. This trend is typical of many effusive eruptions which have a rapid waxing phase and a relatively slow waning.

On the contrary the explosive events have, on average a relatively slow waxing and rapid waning. This peculiar trend is due to a delayed bubble growth within the magma chamber which mainly depends on magma composition and depth of the reservoir. Deep chambers possibly favour a slow development of the eruption with a relatively long time span before maximum discharge rates are attained. Shallower reservoir are more likely to cause fast-developing eruptions.

The seismic monitoring of major explosive eruptions provides evidence that in some cases magma chambers behave as a rigid envelope in response to rapid drainage of magma. Collapse of the chamber occurs when the pressure difference between the magma and the lithostatic load becomes sufficiently less than the strength of the rocks. The amount of volatile supersaturated magma, and the strength of the chamber walls control the overall evolution of the eruption and the timing of chamber collapse.

Fig. 6 - Scheme of evolution of an explosive eruptions as a results of varying pressure in the the magma chamber. Pm=magmatic pressure, Pl= lithostatic pressure, Ps = rock strength, rr= rock density, rm= magma density

The cartoon of fig. 6 summarizes the main features that, on average, are observed in the course of major explosive eruptions.

An important consequence of this model is that, as shown by the observations, the paroxysmal stages of the major explosive eruptions are not expected to occur at the beginning of activity. Some warning is thus anticipated for an impending paroxysmal eruption, but when these phenomena are observed there is an increased chance to record a magma reservoir collapse and possibly a caldera formation.

Although this model makes a strong separation between the two modes of behavior of surrounding rocks, it is probable that a mixed behavior be the norm. This fact must not hinder the fact that the primary factor controlling the evolution of an eruption is the temporal variation of the pressure gradient between the magma chamber and surface. The variation of vent geometry, viscosity, and pressure at the vent play a secondary role as demonstrated for effusive eruption by Stasiuk et al (1993).


Aki K., Fehler M., Das S., 1977, Source Mechanism of Volcanic Tremor: Fluid- driven Crack Models and Their Application to the 1963 Kilauea Eruption, J. Volcanol. Geoth. Res. 2, 259-287

Anderson O.L., Grew P.C., 1977, Stress corrosion theory of crack propagation with applications to geophysics; Rev. Geophys. and Space Phys. 15, 77-104

Aramaki S., 1988, The 1986-87 eruption of Izu-Oshima volcano, Publication of Eartquake Research Institute, University of Tokyo 61 pp.

Barberi F., Cioni R., Rosi M., Santacroce R., Sbrana A., Vecci R., 1989, Magmatic and Phreatomagmatic phases in Explosive Eruptions of Vesuvius as Deduced by Grain-size and Component Analysis of the Pyroclastic Deposits, J. Volcanol. Geoth. Res. 38, 287-308

Barker S.E., Malone S.D., 1991, Magmatic system geometry at Mount St.Helens modeled from the stress field associated with posteruptive earthquakes; J. Geoph. Res. 96, 11883-11894

Bianchi R., A. Coradini, C.Federico, G. Giberti, P.Lanciano, J.P.Pozzi, G.Sartoris, R. Scandone, 1987, Modelling of Surface Deformations in Volcanic Areas: The 1970-72 and 1982-84 crises of Campi Flegrei, Italy, Journ. Geoph. Res., 92, 14139-14150

Blake S., 1981, Volcanism and the dynamics of open magma chamber, Nature, 289, 783-785

Blake S., 1984, Volatile oversaturation during the evolution of silicic magma chambers as an eruption trigger; J. Geophys. Res. 89, 8237-8244

Belkin H. E.,and De Vivo B.,1993, Fluid Inclusion Studies of Ejected Nodules from Plinian Eruptions of Mt Somma-Vesuvius, J. Volcanol. Geoth. Res. 58, 89-100

Burnham C. W., 1972, The energy of explosive eruption, Ea. Mineral. Sci. The Pennsylvania State University, 41, 69-70

Burnham C.W.,1985, Energy release in subvolcanic Environments: Implications for breccia formation, Economic Geology, 80,1515-1522

Carey S., Sigurdsson H., 1986, The 1982 Eruptions of El Chichon Volcano, Mexico (2): Observations and Numerical Modeling of Tephra-Fall Distribution, Bull. Volcanol. 48, 127-141

Carey S., Sigurdsson H., Gardner J.E., Criswell C.W., 1990, Variation in column Height and Magma Discharge during the May 18, 1980 Eruption of Mt St.Helens, J. Volcanol. Geoth. Res. 43, 99-112

Cestelli Guidi C. 1980, Geotecnica e Tecnica delle Fondazioni. Vol. 2, Hoepli, Italy, pp. 873.

Criswell C.W., Chronology and Pyroclastic Strtigraphy of the May 18,1980 Eruption of Mount St.Helens, Washington, J. Geoph. Res. 92, 10237-10266

Crosson R.S., Bame D. A., 1985., A spherical Source Model for Low Frequency Volcanic Earthquakes, J. GeopH. Res. 90, 10237-10247

Druitt T.H, Sparks R.S.J., 1984, On the Formation of Calderas during Ignimbrite Eruptions, Nature, 310, 679-681

Eichelberger J.C., and Hays P.B., 1982, Magmatic model for the Mount St Helens Blast of May 18, 1980, J. Geophys. Res. 87, 7727-7736

Einarson T., 1949, Rate of Production of Material during the Eruption, in T. Einarsson, G. Kjartansson, S. Porarinson Editors, "The Eruption of Hekla 1947-48", Societas Scientiarum Islandica IV-2, 1-18

Endo E.T., Malone S.D., Noson L.L., Weaver C.S., 1981, Locations, Magnitudes, and Statistics of the March 20-May 18 Earthquake Sequence, USGS Professional Paper 1250, 93-107

Foxworthy B.L., and Hill M, 1982, Volcanic eruptions of 1980 at Mountè St Helens: the first 100 days, U.S. Geol. Survey, Prof. Pap. 1249, 125 pp

Giacomelli L., Scandone R., 1992, Campi Flegrei-Campania Felix, Liguori Editore, Napoli 154+57 pp.

Giuliani G.B., 1632, Trattato del Monte Vesuvio e de' suoi incendi, Longo, Napoli, pp. 224;

Gronvold K., Larsen G., Einarsson P., Thorarinsson S., Saemundsson K, 1983, The Hekla Eruption 1980-81, Bull. Volcanol., 46-4, 349-363

Gudmundsson A., Oskarsson N., Gronvold K., Saemundsson K, Sigurdsson O., Stefansson R., Gislason S., Einarsson P., Brandsdottir B., Larsen G., Johanneson H., Thordarson T., 1993, The 1991 Eruption of Hekla, Iceland, Bull. Volcanol. 54, 238-246

Gutierrez C., Singh S.H., Havskov J., 1983, Registro Y Analisis de Los Temblores Relacionados con la Actividad del Volcan Chiconal, 1982. Comision Federal de Electricidad, Proyecto 2728, Mexico

Havskov J., De La Cruz-Reyna S., Singh S.K., Medina F., Gutierrez C., 1982, Eruptions of El-Chichon Volcano, Chiapas, Mexico, Geophys. Res. Lett. 10, 293-296

Imbo' G., 1952, Sismicita' del parossismo vesuviano del marzo 1944, Annali Osservatorio Vesuviano, VI Serie, I, 59-268

Lees J.M., The magma system of Mt St.Helens: non-linear high resolution P_wave tomography. J. Volcanol. Geoth. Res. 54, 103-116

Linde A.T., Agustsson K., Sacks I.S., Stefansson R.,1993, Mechanism of the 1991 eruption of Hekla from continuous borehole strain monitoring. Nature, 365, 737-740

Lirer L., Pescatore T., Booth B., Walker G.P.L., 1972, Two Plinian Pumice-fall Deposits from Somma-Vesuvius, Italy, Geol. Soc. Am. Bull. 84, 759-772;

Lipman P.W., Moore J.G., Swanson D.A., 1981, Bulging of the north flank before the May 18 eruption. Geodetica data. USGS Professional paper 1250, 143-146

Lipman P.W., Nortin D.R., Taggart J.R., Brandt E.L., Engleman E.E., 1981b, Compositional variation in 1980 magmatic deposits, USGS Professional paper 1250, 147-156

Machado F., 1974, The search for magmatic Reservoir, In L. Civetta, P.Gasparini, G. Luongo, A. Rapolla Editors "Physical Volcanology, Elsevier, Amsterdam pag. 255-273

Malone S.D., Endo E.T., Weaver C.S., Ramey J.W.,1981, Seismic Monitoring for Eruption Prediction, USGS Professional Paper 1250, 803-813.

McNutt S.R., 1987, Eruption characteristics and cycles at Pavlov Volcano , Alaska and their relation to regional earthquake activity, J. Volcanol. Geoth. Res., 31, 239-268

McKee C.O., Johnson R.W., P.L. Lowenstein, Riley S.J., Blong R.J., Saint Ours P., Talai B., 1985, Rabaul cladera, Papua New Guinea: volcanological hazards, surveillance, and eruption contingency planning, J. Volcanol. Geoth. Res., 23, 195-237

Melson W G., and Hopson C.A., 1982, Petrologic model of the 1981-82 Mount St Helens magma chamber. Eos 63: 1144

Minakami T., Ishikawa T., Yagi K., 1951, The 1944 Eruption of Volcano Usu in Hokkaido, Japan, Bull. Volcanol. 11, 45,154

Neri A., Dobran F., Influence of eruption parameters on the thermofluid dynamics of collapsing volcanic columns. J. Geophys. Res. 99, 11833-11857

Newhall C.F., Self S., 1982, The volcanic explosivity index (VEI): an estimate of explosive Magnitude for historical eruptions, J.Geoph. Res. 87, 1231-1238

Papale P., Dobran F., 1994, Magma flow along Volcanic Conduit during the Plinian and Pyroclastic Flow Phases of the May 18, 1980, Mount St Helens Eruption, J. Geophys. Res. 99, 4355-4373

Parfitt E.A., Wilson L., Head III J.W., 1993, Basaltic magma Reservoirs: factors controlling their rupture characteristics and evolution, J. Volcanol. Geoth. Res., 55,1-14

Pinatubo Volcano Observatory Team, 1991,Lessons from a Major Eruption: Mt Pinatubo, Philippines. Eos , Trans. of AGU, 72, 545-555

Porarinson S., 1949 The Approach and Beginning of the Hekla Eruption, Eyewitness Accounts, in T. Einarsson, G. Kjartansson, S. Porarinson Editors, "The Eruption of Hekla 1947-48", Societas Scientiarum Islandica II-1,1-17

Power J.A., Lahr J.C., Page B.A., Chouet B.A., Stephens C.D., Harlow D.H., Murray T.L., Davies J.N., 1994, Seismic evolution of the 1989-1990 eruption sequence of Redoubt Volcano, Alaska, J. Volcanol. Geoth. Res., 62, 95-136

Prousevitch A.A., Sahagian D.L., Anderson A.T., 1993, Dynamics of diffusive bubble growth in magmas: isothermal case, J. Geophys. Res., 98, 22283-22307

Rosi M., Principe C., Vecci R., 1993, The 1631 Vesuvius eruption. A Reconstruction based on Historical and Stratigraphic Data, J. Volcanol. Geoth. Res., 58, 151-182

Rolandi G., Barrella A.M., Borrelli A., 1993, The 1631 Eruption of Vesuvius, J. Volcanol. Geoth. Res., 58, 183-202

Rutherford M.J., Sigurdsson H., Carey S., Davis A., 1985, The May 18, 1980, Eruption of Mount St. Helens 1. Melt Composition and Experimental Phase Equilibria. J. Geoph. Res. 90, 2929-2947

Ryan M.P., 1987 Neutral Buoyancy and the mechanical evolution of magmatic system; in B.O. Mysen editor "Magmatic processes: Physico-chemical principles", The Geochemical Soc. Special Publication n°1, 259-287

Scandone R., 1979, Effusion rate and energy balance of Paricutin eruption (1943-1952), Michoacan, Mexico., J. Volcanol. Geoth. Res., 6, 49-59

Scandone R., Malone S.D., 1985, Magma Supply, Magma Discharge and Readjustment of the feeding System of Mount St. Helens during 1980. J.Volcanol. and Geoth. Res 23, 239-262

Scandone R. ,1990, Chaotic Collapse of Calderas, J. Volcanol. Geoth. Res., 42, 285-302

Scandone R., 1991, A reply to the comments by I.Yokoyama and S. De la Cruz and some further ideas on the collapse of calderas and the generation of Ignimbrites, J. Volcanol. Geoth. Res., 47, 352-357

Scott W.E., Hoblitt R.P., Torres R.C., Self S., 1993, Pyroclastic flows of the 15 June 1991 Paroxysmal eruption at Mount Pinatubo Volcano, Philippines. Eos, October 26,pag. 667 (abstract)

Scriven L.E., 1959, On the Dynamics of Phase Growth, Chem. Eng. Sci., 10, 1-13

Shemeta J.E., Weaver C.S., 1986, Seismicity Accompanying the May 18,1980 Eruption of Mount St.Helens, Washington. In : Keller S.A.C., Editor, 'Mt St.Helens: five years later'. Eastern Washington University Press, Cheney, WA. 44-58.

Sigurdsson H., Carey S., Cornell W., Pescatore T., 1985, The Eruption of Vesuvius in A.D. 79, Nat. Geog. Res. 1, 332-387;

Sparks R.S.J., 1978, The Dynamics of Bubble formation and Growth in Magmas: A review and Analysis, J. Volcanol. Geoth. Res., 3, 1-37

Sparks R.S.J., Wilson L., 1976, A model for the formation of Ignimbrites by gravitational column collapse, J. Geol. Soc. Lond. 132, 441-452

Sparks R.S.J., Self S., Walker G.P.L., 1973, Products of Ignimbrite Eruptions, Geology 1, 115-118

Sparks R.S.J., Wilson L., Hulme G., 1978, Theoretical Modelling of the Generation, Movement and Emplacement of pyroclastic Flows by Column Collapse, J. Geophys. Res. 83, 1727-1739

Stasiuk M.V., Jaupart C., Sparks R.S.J., 1993, On the Variation of the Flow Rate in Non-Explosive Lava Eruptions. Ea. Plan. Sci. Lett. 114, 505-516

Stefansson R., Bodvarsson R., Slunga R., Einarsson P., Jakobsdottir S., Bungum H., Gregersen S., Havskov J., Hjelme J., Korhonen H., 1993, Earthquake Prediction Research in the South Iceland Seismic Zone and the SIL Project, Bull. Seism. Soc. of Am. 83, 696-716

Swanson D.A., Lipmann P.W., Moore J.G., Heliker C.C., Yamashita K.M., 1981, Geodetic monitoring after the May 18 eruption. USGS Professional paper 1250, 157-168

Tait S., Jaupart C., Vergniolle S., 1989, Pressure, gas content and eruption periodicity of a shallow crystallising magma chamber; Ea. Pl. Sci. Lett. 92, 107-123

Thomas L.J., 1973, An Introduction to Mining , R.Burton Printers Ltd, Sidney, pp. 471

Thorarinsson S., Sigvaldason G.E., 1972, The Hekla Eruption of 1970, Bull; Volcanol., 36, 269-288

Varekamp J.C., 1993, Some Remarks on Volcanic Vent Evolution during Plinian Eruptions, J. Volcanol. Geoth. Res., 54, 309-318

Wadge G., 1981, The variation of Magma Discharge during Basaltic Eruptions, J. Volcanol. Geoth. Res., 11, 139-168

Walker G.P.L.,1984, Downsag Calderas, Ring Faults, Caldera Sizes, and Incremental Caldera Growth. J. Geoph. Res. 89, 8407-8416

Walker G.P.L.,1985, Origin of Coarse lithic Breccias near Ignimbrite source vents, J. Volcanol. Geoth. Res., 25, 157-171

Weertman J., 1971, Theory of water filled crevasses in glaciers applied to magma transport beneath oceanic ridges, J. Geophys. Res. 76, 1171-1183

Wilson L., Head J.W., 1981, Ascent and Euption of Basaltic Magma on the Earth and Moon, J. Geoph. Res. 86, 2971-3001

Wilson L., Walker G. P. L.,1987, Explosive volcanic eruption VI. Ejecta dispersal in plinian eruptions: the control of eruption conditions and atmospheric properties, Geophys. J. R. Astr. Soc. 89, 657-679

Wilson L., Sparks R.S.J.,Walker G.P.L., 1980, Explosive Volcanic Eruptions -IV. The control of Magma Properties and Conduit Geometry on Eruption Column Behaviour, Geophys. J. Roy. Astr. Soc. 63, 117-148

Wolfe E., 1992, The 1991 Eruptions of Mount Pinatubo, Philippines, Earthquakes and Volcanoes, 23, 5-37

Wolfe E.W., Neal C.A., Banks N.G., Duggan T.J., 1988, Geologic observations and chronology of eruptive events, in US Geol. Surv. Prof. Pap. 1463, 1-98

Wood A.W., 1988, The Fluid-Dynamics and the Thermodynamics of Eruption Columns, Bull. Volcanol. 50, 169-193