The storage and transport of magma: The search for the hidden magma chamber
Part 3: The complex interplay between magma accumulation, flank instability, and eruptions

On 22 September 2002, the northeastern flank of Etna was shaken by a sharp earthquake, whose epicenter lay along the westernmost portion of the Pernicana fault system, close to the tourist complex of Piano Provenzana. Field inspection revealed that displacements of up to 0.45 m had occurred along the fault, creating impressive fractures cutting across roads and other man-made structures. Evidently a large portion of the eastern flank of the volcano had begun to slip away from the remaining, stable part of the mountain. This was the first time that such movement occurred at the Pernicana fault system since late 1988, and it would become the best documented episode of flank slip at Etna so far (Neri et al. in press).

Five weeks later, Etna erupted. The eruption occurred simultaneously on the upper southern flank and along the Northeast Rift, a major structure adjacent to the Pernicana fault system. At the same time, the eastern flank moved again. Displacement measured along the Pernicana fault system reached amounts of up to 2 m in the westernmost portion of the fault system and propagated eastward at decreasing rates right down to the coast of the Ionian Sea, over a full distance of 18 km. A previously unknown section of the fault system was activated in its eastern part, belying earlier assumptions that displacement was transferred from the downslope extremity of the Pernicana (as known prior to 2002) to a fault known as Fiumefreddo fault, further to the north. By early November 2002, the Pernicana fault system was seen to constitute a continuous line from the Northeast Rift in the west to the coastline in the east, and probably continuing offshore. About half way down the slope, the fault system was seen to bifurcate in numerous minor fault strands connected by areas of thrusting, where portions of the volcano flank were pushing out in a drawer-like fashion. This transition zone separated the western (and formerly known) part of the Pernicana fault system, where movement was fast and lasted only a few days, from a previously unknown eastern section, which showed much slower but continuous movement - this lower, eastern section continued to move at a slow rate for many months following the initiation of the flank slip.

Three days after the onset of the major flank slip and eruption, on 29 October 2002, a series of earthquakes occurred on the southeastern flank of Etna and caused considerable damage in the area between Santa Venerina and Fornazzo. Like the earthquakes along the Pernicana, these events had a very shallow origin, lying at depths of no more than 5 km below the sea level. In late October and early November, ground rupturing and displacement was observed along numerous fault on the lower eastern and southeastern slopes of Etna. Like in the areas along the Pernicana fault system, roads and buildings were cut by fractures, resulting in widespread damage. One month after the beginning of the eruption, seismicity and ground fracturing extended further south, to the Trecastagni fault, where the same scene of ruptured roads and walls was repeated again. From the directions of movement it was possible to see that the entire area between the Pernicana fault system to the north and the Trecastagni fault was involved in this gigantic flank slip.

Probably the most fascinating aspect of this series of events was the intimate relationship between the flank slip and the eruption on the southern and northeastern flanks of Etna. The eruption itself had come with extremely little warning (only two hours of intense seismicity had preceded the opening of the first eruptive vents), and thus stood in marked contrast with the four days of intense seismicity and ground deformation that had warned of the 2001 eruption. The magma must have virtually shot up through the flanks, especially on the southern flank, where it came from a depth of about 5-6 km below sea level. How was this possible? One explanation might be that after the 2001 eruption the southern flank was so thoroughly fractured, that the magma could rise freely without encountering much resistance. But this would not account for the simultaneous eruption on the northeastern flank, which had not been affected by the 2001 events. It had thus to be another process that triggered the new eruption, and it is likely that this was the massive slip on the eastern flank, which had actually begun five weeks before the eruption.

On that 22 September 2002, the Northeast Crater had been in nearly continuous mild activity since three months, and the rising magma had gradually filled the deep pit present within that crater, eventually arriving 50 m below the lip of the pit. Strombolian explosions were seen on many days even from far away, and provided a clear indicator that the central conduit system had recharged after draining during the 2001 eruption. Immediately after the 22 September earthquake and displacement on the Pernicana fault system, this persistent activity suddenly stopped, and no incandescent magma was seen in the Northeast Crater for the following weeks. Some of us wondered, "where'd that magma go?" The answer came during the night of 26-27 October: very probably it had begun to drain laterally from the central conduits into the Northeast Rift, where the 22 September displacement had opened a new crack, a pathway for the magma into that flank. Unfortunately the message was not understood in time. It might have been interesting to amplify monitoring on the Northeast Rift during the weeks after the 22 September event, in order to see whether there was ground deformation related to magma intrusion, and whether some of the perpetually active fumaroles in the upper portion of the rift showed changes in temperature and composition. But the Pernicana had moved several times in the 1980s without any ensuing eruption on the rift, so why should one assume that this time there would be an eruption? Overall, the dynamics of the Pernicana were still not well understood, and even less were its relationships with flank eruptions. Now, after the events, it is easy to say "things went like this or like that", and that one should have known better. As a matter of fact, there was no urgent reason to assume that the flank slip initiated on 22 September would be followed by as dramatic an eruption as the one that came shortly thereafter.

The second, and larger, movement of the slipping flank on 26-27 October was what unleashed the 2002-2003 eruption. Actually it is difficult to distinguish which of the two processes started first - the movement of magma within the flanks, or the flank slip. The latter is more probably because magma began to move simultaneously on two opposite sides of the mountain, as if some giant key had unlocked the pathways for two different magmas coming from two different sources. This can be hardly imagined to result from magma pressure or movement alone. In an interdisciplinary study (whose preliminary results were presented at a congress on volcano seismicity on the island of Pantelleria, Sicily, in September 2003: Acocella et al. 2003), the patterns of seismic activity and ground deformation related to the eruptions in 2001 and 2002-2003 have been analyzed, and the results are striking. Although in many respects these eruptions were similar, from a purely volcanological point of view (e.g., emission of two compositionally distinct magmas in different areas, strong explosivity on the southern flank), their triggering mechanisms were fundamentally different. In 2001 it was mostly the eccentric magma, which pushed its way to the surface, in leading to the eruption. By the way, it caused extensive fracturing of the upper flank of the volcano, so that the magma stored in the central conduit system since 6 years leaked out through lateral fissures close to the summit area. That is the most plausible scenario for the 2001 eruption.

The analysis of the data from the 2002-2003 eruption revealed no evidence for forceful upward migration of magma below the southern flank before the eruption. While it is true that magma accumulation in the suspected storage area below that flank had resumed several months earlier, the characteristic seismicity that had preceded the 2001 eruption by many months, and especially during the four days before the eruption, was lacking in 2002. Instead, seismic activity had occurred on many of the fault systems that delimit and crisscross the unstable area on the eastern and southeastern flanks, which had not been the case before the 2001 eruption. Slow spreading of the flank had initiated immediately after the 2001 eruption, causing this peculiar distribution of seismicity, but more rapid flank slip began only on 22 September 2002. And it was then when an irreversible chain of events began, which started with the migration of magma into the northeastern flank, a process that probably further enhanced the instability of the adjacent eastern flank and eventually led to the second major slip event on 26-27 October, which allowed magma from both the eccentric storage area to the south and from the Northeast Rift to rise to the surface and produce the large eruption of 2002-2003.

On the northeastern flank the amount of available magma was limited - it was probably equivalent to the quantity of magma stored in the central conduits up to the time of the eruption, plus a certain volume of magma that had begun to intrude into the rift after the 22 September earthquake and flank slip. In the end, the Northeast Rift eruption lasted only one week and extruded some 10 million cubic meters of lava with some minor pyroclastics. On the southern flank, for a long time it seemed that there was no limit to magma extrusion. The eruption there lasted three months, and produced four to five times as much magma as the Northeast Rift eruption, more than half of this in the form of pyroclastics. Only when the magma overpressure was exhausted, the eruption finally ended.

Movement of the eastern and southeastern flank, however, continued even after the eruption was over. Renewed displacement accompanied by shallow seismicity occurred at the Pernicana fault system less than three weeks after the end of the eruption, on 13 February 2003, and slow movement in the lower, eastern portion of that fault system has continued uninterruptedly to the present day (October 2003). This continued movement - although being much less significant than the enormous slip of September-November 2002 - is seen by some of us as an indicator that the next flank eruption might affect essentially the same areas as the 2002-2003 eruption. But it is probably wrong to assume that the next eruption will be a simple repetition of its predecessor. Eruptions at Etna are not all the same, when it comes to their triggering mechanisms. Some cases appear to be similar to that of the 2002-2003 eruptions, while others are definitely not. A review of all eruptions and possible episodes of flank slip in the past 20 years or so shows that there is nothing like a simple repetitive pattern in the processes that eventually culminate in flank eruptions. Instead, things are characterized by such an awful complexity that studying the dynamics of the volcano often drives us to despair.

Digging in an obscure past
Mount Etna is giving us a message, this is clear. But deciphering this message requires at least a reasonable understanding of some fundamental features in its dynamics, which we still do not have. Knowledge is improving, but still far from decent, and each single event - be this an eruption or some process observed with geophysical or other methods - yields new information but brings up still more new questions. Take the example described in the previous paragraphs. It looks like we have a pretty good understanding of the dynamics of the 2001 and 2002-2003 eruptions. There is actually some reason to hope that this is actually the case. Yet these are only two eruptions, and there have been many others in recent years. It might seem a good plan to compare the recent eruptions with the previous ones, and this is actually what many researchers do. Now, with the knowledge that there might be some serious interference between flank instability and magmatism (that is, magma accumulation and eruption), it should be an easy task to find out whether there have been more events like those in 2002-2003 in the past. The problem is, this is anything but easy.

Firstly, the quality and quantity of available data is not the same for the past and present. Today Etna is one of the most thoroughly monitored volcanoes on this planet, but modern geophysical monitoring was begun only sometime during the 1970s, and at that time it was still a faint shadow of what it is now. Monitoring of the area of the Pernicana fault system started in the early 1980s, which means that none of the pre-1980 movements of this structure are documented in any detail. Neither the full extent of the Pernicana nor its potential relationship with flank eruptions were known until 2002. This is not a very good base onto which build comparative analysis.

The Pernicana is known to have undergone at least five major movements during the 1980s. Each time the seismicity was very shallow, concentrated above 5 km below the sea level, and the major events were accompanied by ground fracturing testifying to sub-horizontal (left-lateral) displacements along the fault plane. It is likely that each of these events was actually related to a flank slip episode, like in 2002-2003, but none of these was monitored as closely as the most recent one. Much of the early studies were concentrated on determining the rates of movement along this fault system, from observations of man-made structures of known age, and geological features whose ages are approximately known as well. A connection between the Pernicana fault system and flank instability was suggested but had never been proved before 2002. Many of the other faults involved in the 2002-2003 slip event had moved before, sometimes causing extensive though locally limited damage (due to the shallowness of the earthquakes), but few of the pre-2002 studies placed them into a larger framework of volcano instability.

What is true is that the series of assumed slip events during the 1980s occurred at a time when flank eruptions were quite frequent, and the volcano was thus unstable in a variety of senses. Virtually every time that magma rose again up within the central conduit system, little time would pass and then this magma would drain out somewhere on the flanks, especially in the southern, southeastern and east-northeastern sectors of the volcano. Flank eruptions occurred in 1981, 1983, 1985 (twice during that year, in March-July and again in December), 1986-1987 and 1989. Episodes of strong seismicity along the Pernicana and/or along the numerous faults on the eastern and southeastern flanks (the latter are known as the Timpe fault system) occurred in 1981, 1984, 1985-1986 and 1988, with more isolated episodes in 1982 and 1989. Assuming that all of these seismic episodes reflect flank slip, there might be a true correlation with eruptive activity, for each of them was followed by a flank eruption within less than a year. Two events at the Pernicana were actually followed by flank eruptions within less than one day (December 1985 and October 1986), although none of these occurred on the Northeast Rift (the last eruption there, prior to 2002, had occurred in 1947); instead these eruptions occurred in the Valle del Bove area, on the eastern and east-northeastern sides, respectively.

Work is currently under way to elucidate the times and modes of movement of the unstable flank during the two decades prior to the 2002-2003 eruption. Some early impressions gained during this study are described in the previous paragraph. They indicate that movement of the flank probably occurred several times during the 1980s, but in different areas and in different modes, and the relationship with the numerous eruptive events of that period is highly complex. It is likely that one of the main results of this research will be expressed like this: eruptions of Etna may seem quite similar from one event to the other, but the processes that lead to these eruptions are strongly variable, probably depending on the timing and location of magma ascent and flank slip.

Causes of flank instability, causes of eruptive cycles
While much of the story concerning the relationship between slippage of Etna's unstable flank and flank eruptions remains to be elaborated and written, one thing is clear. Something triggers flank eruptions, and something triggers flank slip. Sometimes it might be flank slip that triggers eruptions, and flank instability can certainly be assumed to increase during periods of frequent magma intrusions into the flanks.

But if you have read all the text until here, you might recall that it basically started with eruptive cycles at Etna. Now, after having received all the information related to flank instability, you might wonder: what has this to do with eruptive cycles? Let's then return to the cycles. As noted at the beginning, such cycles consist of three distinct phases: (1) eruptive quiescence, usually following a major flank eruption; (2) an extended period of summit eruptions, sometimes lasting far more than 10 years; and (3) a sequence of flank eruptions following each other at intervals of a few years, with many periods of intervening summit activity. The last flank eruption in this final sequence is usually the largest and it is followed by the beginning of the next cycle. Four eruptive cycles occurred in 1865-1982, 1892-1928, 1928-1951, 1951-1993. Yet another cycle began in 1993 and has already entered into its third phase, the one with frequent flank eruptions, and two such eruptions have occurred as of late 2003.

The fact that the last eruption of a cycle is commonly the largest (this has been so in three of the four complete eruptive cycles between 1865 and today) is probably fundamental. Three of these cycle-closing flank eruptions, in 1892, 1950-1951 and 1991-1993, emitted significantly more than 100 million cubic meters of lava - in 1991-1992 the magma volume was 235 million cubic meters. The single exception, which is the devastating though volumetrically minor 1928 eruption, will have to be discussed some time, but in the view of other evidence in favor of the cycle concept, it is seen rather as "the exception which confirms the rule". Large flank eruptions like in 1892, 1950-1951 and 1991-1993 drain large volumes of magma from the shallow storage system of the volcano, and significant deflation - a kind of "settling" or sinking - of the Etnean edifice has been directly monitored in the case of the most recent of these large eruptions. It is plausible to assume that all fractures cutting through the flanks of the volcano, which serve as potential pathways for lateral eruptions, are sealed by this deflation, in addition to many of the fractures having been previously filled during magma intrusions related to flank eruptions. Said simply, the volcano becomes structurally more stable. And it is amazing to see how all eruptive activity remains limited to the summit (once magma re-ascends in the central conduits after a period of repose following a major flank eruption) for many years, regardless of the volume of uprising magma. This was impressively illustrated during the prolonged summit eruptions between 1955 and 1971, and once more between 1995 and 2001. In spite of the notable volumes and intensity of these summit eruptions, the flanks of the mountain remained sealed for many years. And then, they gave way. Flank eruptions began to occur one after another. The volcano had suddenly become unstable. Flank slip was frequent during the 1980s, when there were also many flank eruptions, but from the available data it seems that very little movement of the unstable flank had occurred during the 1950s and 1960s, and again verly little movement occurred during the period of prolonged summit eruptions between 1995 and 2001. It thus seems that flank slip is concentrated in periods of frequent flank eruptions.

What is it then that causes flank slip during these periods, and why are there so many flank eruptions in such periods? The answer probably lies below the volcano, in the extensive and complex plumbing system that might actually consist of one or more magma reservoirs. It is well established that much more magma enters into this plumbing system than is actually erupted - the amounts of sulfur dioxide and other gas species emitted from Etna's summit craters indicates that the quantity of magma that is supplied is about ten times as much as the quantity of erupted magma (Allard 1997; Le Cloarec and Pennisi 2001). This means that up to 90 per cent of the magma that enters into the plumbing system remains there and is never erupted. Much of this probably gets stored in a network of dikes (which might be the "plutonic complex" mentioned by Borgia et al. 1992) and gradually cools. But the prime effect of all this magma intruding into the volcanic substratum is GROWTH - growth of an ever larger area where magma accumulates and partially cools. This might be what makes the volcanic edifice unstable and leads to those peculiar clusters of flank eruptions. Since the quantity of non-erupted magma is consistently higher than that of erupted magma, even in periods of frequent flank eruptions, the instability of the edifice will progressively increase until a very large flank eruption brings some temporary relief, and a new cycle starts, with the volcanic edifice being relatively stable for some time. But as soon as the amount of new, non-erupted "excess" magma stored below the volcano reaches a certain quantity, and the overlying volcanic edifice has inflated to a certain degree, the flanks become unstable again, permitting both flank eruptions and episodes of flank slip.

At first sight this might seem a fine model of how Etna works. It might be true that this concept brings us a bit closer to reality. But as a matter of fact it represents only a first, rough approach, and very much work remains to be done and much more understanding is necessary to be able to say "this is how Etna works". There are many aspects that remain enigmatic, and on the other side we see that certain events do not fit into the convenient scheme of eruptive cycles and flank instability, like the 2001 eruption, which was not a direct consequence of flank instability but in turn may have enhanced flank instability. The role of eccentric eruptions in this framework is still mysterious. There is a very high number of "maybe's" and "possibly's", which means that things might be that way or right the opposite. Interpretation of even the recent past is rendered difficult by the lack of data, and the farther we go back in time, the more is this volcano veiled in mystery. In the end the only competent statement that we can make about the way Etna works is that it is incredibly complex.

References

(will be added soon; please visit the Etna reference list)

Copyright © Boris Behncke, "Italy's Volcanoes: The Cradle of Volcanology"