The storage and transport of magma: The search for the hidden magma chamber
Mount Etna erupts almost continously, but where does the magma feeding its activity come from, and how does it move into the volcano and up to the surface? Does Etna, like many other volcanoes, have a "magma chamber" where the mixture of gas and molten rock accumulates and evolves before rising to the surface?
At many volcanoes, modern seismological methods have helped revealing the location of what is popularly named "magma chambers" at relatively shallow depth (a few kilometers) - since these areas where magma accumulates before eruption are probably much more complex than anything having the simple shape of a "chamber", it is more appropriate to talk of "magma storage areas" or "magma reservoirs". At volcanoes like Kilauea on Hawaii and Vesuvio in Italy, magma reservoirs have been located at depths of 2-10 km. In the case of Etna, no permanent shallow reservoir appears to exist in this moment, at least none of significant dimensions, but the latest eruptions in 2001 and 2002-2003 indicate that this might be changing.
At Kilauea volcano on Hawaii the presence of a shallow magma storage area has been suspected for many decades, and intense studies have led to the evolution of a dynamic model of Kilauea's plumbing system that is amazingly clear and detailed. The magma reservoir of that volcano probably lies between 7 and 2 km below the surface, and secondary magma reservoirs lie below the two rift zones extending from the summit to the E and SW. In recent decades it has been possible to trace the pathways of magma in all eruptions (and several intrusive events) of Kilauea. The destabilizing effect of magma pressure in the shallow plumbing system on the S flank of Kilauea has been recognized as one of the most serious hazards threatening life and property on Hawaii. Similarly, other basaltic volcanoes like Piton de la Fournaise on the island of Réunion in the Indian Ocean have been thoroughly studied and can be said to be fairly well understood.
In contrast to these volcanoes, which are located on oceanic islands built on relatively thin oceanic lithosphere, Etna has kept many of its secrets to the present day. This is in part due to the fact that it is located in a much more complex tectonic setting, which in itself is not yet fully understood. Another reason is that Etna is the result of an eventful and highly changeable history, and last but not least the history of geophysical monitoring is much shorter on Etna than on Kilauea. Nonetheless the past two decades have brought many new findings - at times contrasting - about the internal dynamics of Etna.
Seismic studies carried out during the late-1970's revealed the existence of a vast zone containing about 14 per cent molten material, located at a depth of about 20-25 km below the volcano (Sharp et al., 1980). According to their model the storage area was 22 x 31 km in diameter, and about 4 km thick in its central portion. Based on these dimensions the zone contained about 1600 km3 of magma, more than four times the current volume of the volcano (about 350 km3).
More recent studies by Hirn et al. (1997) indicate an unusually thin crust below Etna, and thus the boundary between the lithosphere and the upper mantle is much closer to the surface than in the surrounding regions. Their model envisages a kind of "lens" of magma on top of this raised, "upwarped" mantle, at a depth of 15-20 km.
Closer to the surface, within the edifice of the volcano itself, magma was believed to be stored mainly along the axis of the central conduit system feeding the summit craters (see next section, below), and in its possible lateral extensions along the main tectonic trends (Murray, 1990). The assumption that storage of magma over periods of up to several years may occur in dikes under the flanks of the volcano is regarded little realistic by Murray (1990), since a dike with a thickness of a few meters is expected to chill and solidificate within a few weeks. However, if dikes are envisaged to be larger than those exposed, for example, in the Valle del Bove on the E side of Etna, especially when lying close to the central conduit system in an environment that is probably fairly hot, chilling might take much more time.
During the 1990's many new data were gathered with significantly enlarged seismic and geodetic networks and much more sophisticated instruments, and the internal dynamics of Etna were assuming clearer, but also more complex shapes. The results of new seismic surveys carried out by Murru et al. (1999) differ somewhat from those of the early studies, mainly in the postulation of a possible minor magma storage area at a depth of 4-8 km below the summit, and a more significant storage area about 10-16 km depth. The observed contraction of an ellipsoidally shaped body during the most recent flank eruption, in 1991-1993, at a depth of about 5 km below the summit (Bonaccorso et al., 1996) has been interpreted by Murru et al. (1999) as further evidence for some magma storage in that area. Deformation data presented by Massonnet et al. (1995) and Lanari et al. (1998) provided additional evidence for the lower storage area postulated by Murru et al. (1999).
It may be of some importance to point out that the data used by Murru et al. (1999) are based on a catalog of 450 seismic events recorded between 1990 and 1997, whereas the model of Sharp et al. (1980) was based on seismic data gathered 15-20 years before, and it cannot be entirely excluded that some changes have occurred in the plumbing system of Etna during the period separating the two data sets. We will see later that the behavior of Etna has changed in that period, and this change may in part be related to the dynamics of the plumbing system.
In an analysis of radar data (interferograms) from the European Remote Sensing Satellites (ERS 1 and ERS 2) Lanari et al. (1998) find a transition from deflation (subsidence) of the Etnean edifice during the 1991-1993 eruption to inflation after that eruption, which was particularly strong just before eruptive activity resumed in the summer of 1995. Interestingly the source of post-1993 inflation was located at greater depth than the 1991-1993 deflation source, and the latter was not affected by renewed inflation, that is, the post-1993 deformation source did not move upwards. This may be interpreted in terms of a highly changeable pysical regime in the shallow plumbing system in a relatively short time span, and it is very probable that this has also existed in the past.
The most fascinating moment for those interested in the dynamics of magma storage and transport came with the eruptions of 2001 and 2002-2003. Before, during and after the 2001 eruption a flow of data of unprecedented quantity and quality was produced by the network of geophysical monitoring instruments on Etna. Never before has an eruption of this volcano been so well observed in any sense, and never before were the data as rapidly available. As I told in an interview to a national broadcasting station of the U.S. in late July 2001, while the activity was at a climax, "this eruption will teach us more about Etna than we have learned before during 3000 years of history". During the months after the eruption, when the need of rapid - and often rough and not exactly accurate - interpretations was replaced by careful analysis and evaluation, a stunningly complex image of the volcano's inner dynamics appeared with surprising clarity. Etna seemed to have become much more "transparent", although it seemed that the higher the precision of the "views into the volcano", the more complicated and changeable were the things that we saw. In the end, the lessons from the 2001 eruption did help very little when it came to the next eruption, in late October 2002.
During the months before that eruption it had been observed that the volcano was recharging. Inflation of the southern flank started about half a year after the 2001 eruption, and magma reappeared in the summit craters in June 2002. Nearly everybody was convinced that little time would pass until there would be another flank eruption. Yet the timing and the extremely limited amount of premonitory signs came as a complete surprise. In fact, the mechanisms that triggered this new eruption were completely different from those in 2001. While it seems that in 2001 it was largely the uprise of magma from a newly formed reservoir below the southern flank that led to the eruption, the 2002-2003 eruption was probably a result of the movement of a large sector of the eastern flank of the volcano, which had begun five weeks before that eruption. The flank movement - this is called spreading or slip - was, in turn, a result of increased instability of the edifice, caused by the rapid growth of at least one of the two magma reservoirs below the volcano (Neri et al., in press).
Research carried out by scientists of the Palermo section of the Istituto Nazionale di Geofisica e Vulcanologia has served to define in more detail the dimensions and dynamics of the magma storage system lying below Mount Etna (Caracausi et al. 2003a, b). Among the most important findings are the sheer size of the supposed storage area, which these authors believe to extend far beyond the surface area of the volcano, especially to the south. Evidence for the presence of magma storage is found up to 40 km from the southern boundary of the volcano (Caracausi et al. 2003a). Geochemical signals observed before and after the 2001 eruption at numerous surveillance stations mark different surges of magma uprise from a deeper (around 10 km below the sea level) reservoir into a shallower (less than 5 km below sea level) reservoir, confirming the existence of two distinct zones of magma storage as suggested by Murru et al. (1999). The critical geochemical signals of uprising magma batches are seen several weeks before the magma itself arrives at the surface and are interpreted by Caracausi et al. (2003b) as useful tools for eruption forecasting. One problem, however, lies in the fact that signals of magma ascent are also observed without flank eruptions following, as in the months following the 2001 eruption - Caracausi et al. (2003b) interpret this as the incapacity of the volcano of producing another eruption immediately after a flank eruption. Furthermore, it remains to be established what relationship there is between the supposed storage areas and the fact that two compositionally distinct magmas have been erupted during both the 2001 and 2002-2003 eruptions. Do the two storage areas corrrespond to these different magmas?
There are a few further open questions, which for the moment will be difficult to answer. Has this constellation of two distinct reservoirs existed for a long time before it was detected with modern geophysical and geochemical methods? If not, when did these reservoirs form? Are they currently growing? Is there a possibility that the presently existing reservoirs will eventually merge to form a single, more or less homogeneous storage area, as was probably the case during the 17th century? To discuss these questions, it is necessary to consider what can be directly observed, both in the topmost portion of Etna's magma plumbing system (the central conduits), and in the long-term eruptive behavior of the volcano.
central conduit system
Detailed observations and descriptions of the activity and evolution of the summit craters over the past few decades indicate that the central conduit system is far from stable in the long term. While the central conduit system is the only area on Etna where magma is rising to the surface continuously, it does so through a variety of pathways whose number and location change with time. Moreover, the central conduit system has become significantly more complex since the beginning of the 20th century.
If you try to imagine what the central conduit system of Etna is like, in terms of geometry, you must not think of one or more cylindrical shafts extending vertically from the craters towards the Earth's interior. Probably the main conduits feeding the four summit craters do have some of these characteristics only for their uppermost few hundred meters. However, at depth they are possibly interconnected and may merge or split into a network of conduits or magma-filled fractures; there is very little means to get clear ideas of how the conduits look like at depth. Even at a surficial level, the conduits are far from simple. While the conduit perimeters immediately below the summit craters correspond to those of the craters, most of the conduits are filled with a mixture of congealed magma, lavas and pyroclastics deposited on the crater floors, and material that has fallen onto the crater floors by collapse of the crater walls. Much of this mixture is probably rigid and therefore it is not to be considered something like a "boiling lava lake" or something similar. Active magma rarely occupies the entire width of the conduit, but rather rises to eruption within the craters through much smaller pathways piercing the fill of the principal conduits. At times there are several vents erupting on the crater floor, with their number increasing as activity intensifies, and more magma occupies a network of narrow conduits within the principal conduit.
The "rising" and "dropping" of the floors of Etna's summit craters is described in many publications, and has also been described at other volcanoes with similar eruptive behavior. This is generally thought to reflect the rising and subsidence of the magma column(s) within the conduit(s). However, at least as far as the "rising" is concerned, the process is not the simple rise of fluid magma within the shaft of the conduit. During the eruptive activity initiated in 1995, the summit craters were simply filled by the products of activity that took place on the crater floor. This could be observed particularly well at the Bocca Nuova, which was the deepest of the four craters at the onset of the activity (150 m below the W rim). Eruptive activity took place from only one vent, or a group of very closely spaced vents, in the NW part of the crater floor for the first 15 months, and then also from a second vent in the SE part of the crater. By mid-1997 relatively large pyroclastic cones had developed around the vents, and lava had covered large parts of the surrounding crater floor, so that the deepest points of the crater lay only about 100 m below the W rim. Very rapid growth of a cone at the NW vents took place in the second half of 1997, and in November its summit stood only about 50 m below the W crater rim. Similarly, the pyroclastic edifice at the SE vents grew notably at the end of that year.
During the activity of 1995-1997, no uplift of the crater floor by magma pushing from below was observed, the gradualy shallowing of the crater was entirely due to the filling by material erupted from the vents on the crater floor. However, in January 1998 there occurred a sudden withdrawal of the magma in the conduit, causing dramatic subsidence that was most pronounced at the large cone in the NW part of the crater. At an early stage the summit of the cone collapsed, but the entire cone was seen to subside, as a spectacular system of concentric, inwards-stepping faults developed around its base. During the following four months subsidence was most pronounced in the S part of the cone, but it appeared as though the whole crater floor was sinking.
In the summer of 1998 the magma rose again beneath the Bocca Nuova, and the partially destroyed NW cone began to rebuild. This was accompanied by the slow uplift of the whole crater floor which may have lasted a few weeks. In this way most of the volume lost during the early 1998 subsidence was recovered, simply by being pushed up by the reascending magma from below. While the 1998 activity at the Bocca Nuova was significantly less intense than that of 1997, the crater produced the most spectacular activity of its 30-years-long life in late 1999. This activity, which is here named "the October-November 1999 Bocca Nuova eruption" showed a number of surface phenomena related to magma dynamics in the uppermost part of the conduit that were unexpectedly complicated and varied (Behncke et al. 2003; Calvari and Pinkerton 2002; Harris and Neri 2002). Among the most impressive features was the formation of an endogenous lava dome (a dome that grows by internal magma supply, with no surface extrusion of magma) on the western rim of the Bocca Nuova, which was pushed onto the steep outer slope of the Bocca Nuova where it collapsed and generated small pyroclastic flows (see the photos and video clips of that event at Stromboli On-line).
of flank eruptions
Eccentric eruptions had their big return on the scene in 2001 and 2002-2003. Those eruptions were unique in the modern history of Etna, because they incorporated both eccentric and lateral activity at different but closely spaced sites. Evidently the magma erupted during these eruptions came from different storage areas, one being the central conduit system, and the other lying below the southern flank, probably at a depth of about 5 km below the sea level. This latter reservoir was a new feature, for no eccentric eruption had occurred in the area since 1892, and its growth had been well visible from the strong inflation of the southern flank during the months before the 2001 eruption. It is tempting to assume that this reservoir coincides with the shallower of the two reservoirs observed by Murru et al. (1999) and Caracausi et al. (2003b). This reservoir rapidly recharged after the 2001 eruption, for which the geochemical evidende presented by Caracausi et al. (2003b) and renewed inflation of the southern flank, first noted at the beginning of 2002, were intriguing indicators. But when the next eruption came (the one that lasted from late October 2002 until late January 2003), it was seen that the triggering mechanisms of flank eruptions are not alike, and the storage, uprise and eruption of magma is set in a complex framework of regional tectonics and volcano instability. In addition, there are impressive long-term fluctuations in the rate at which magma arrives at the surface, which somehow are governed by all of these factors, and eruptions (of all types) are actually only the final manifestations in a series of hauntingly complexly interwoven dynamic processes.
It seems that at least the output rate of Etna varies significantly in time, both in the short and in the long term. What is especially amazing is that there is a very clear increase of the activity in all senses since about 1950, which is not an artifact of improved observation and reporting, but it is a genuine growth of the number, intensity, volume and complexity of eruptive events. We will return to this in another section. Here we will have a brief look on how the volcano behaves during what may be called an "eruptive cycle" and what may be inferred from this on the dynamics and rates of magma uprise in such a cycle. A preliminary report on our findings regarding eruptive cycles is given by Behncke and Neri (2003b). There we report that since 1865, flank eruptions do not occur at random intervals, but are rather coming in clusters, or series. Each of these series is preceded by a long (up to 16 years) of summit eruptions without intervening flank eruptions, which in turn begins after a period of eruptive quiescence that lasts up to a few years. The pattern of repose-summit eruptions-flank eruptions constitutes an eruptive cycle, and most cycles end with particularly voluminous flank eruptions.
Typical examples are the 1865-1892 cycle, which started with three years of repose following the voluminous 1865 flank eruption, then summit activity resumed in 1868 and continued, with fluctuations, until 1874 when there was the first in a series of five flank eruptions over the next 18 years. The last of these, in 1892, was quite large, with a total volume of about 150 million cubic meters of lava and pyroclastics. Again, the most recent complete eruptive cycle, began after a huge flank eruption (1950-1951, volume: about 150 million cubic meters), and its first 3.5 years were characterized by eruptive quiescence. Then, in 1955, summit activity started at the Northeast Crater, and for the following 16 years there were virtually continuous eruptions at the Northeast and Central Craters, interrupted only by 1.5 years of relative quiet in 1964-1966. Flank eruptions started in 1971 and occurred at a mean interval of 1.5 years until 1993, when the last, and most voluminous, eruption (235 million cubic meters) ended that cycle. A new cycle started in 1993, after the large 1991-1993 flank eruption, with its first two years being characterized by the absence of eruptive activity anywhere on the volcano; vigorous summit eruptions occurred between 1995 and 2001, and then a new series of flank eruptions started in 2001, with two such eruptions accomplished by October 2003. If this new cycle behaves like its predecessors, these two eruptions mark just the beginning of a long series of flank eruptions at relatively short intervals (1-2 years). That means, for the next 10-15 years at least we will see frequent eruptions on the flanks of Mount Etna, and some of them will probably be more voluminous than those in 2001 and 2002-2003.
So if there are cycles, how do they work? Why are there cycles, and why do they occur only since 1865? Do they constitute a distinct feature in a long-term evolution of the eruptive activity of Mount Etna? Can they be used to draw a "big picture" of how this volcano works? Continue
Copyright © Boris Behncke, "Italy's Volcanoes: The Cradle of Volcanology"
Page set up on 28 September 1999, last modified on 11 October 2003