Montserrat Volcano Observatory

Overview of the Eruption
at Soufriere Hills Volcano, Montserrat
29 March 1997

Introduction | Observations | Expert Elicitation | Concluding Comments


A general review of the activity of the Soufriere Hills Volcano, Montserrat has been carried out with the principal purpose of providing the Governments of Montserrat and the UK with a synopsis of the state of the volcano and its scientific evaluation. The overview is based on a meeting at Bristol University on the 3rd March, a meeting at the Montserrat Volcano Observatory on the 6th March and an expert elicitation involving 5 scientists in the UK and 10 MVO scientific staff on Montserrat. The major findings to date are summarised. Some concluding remarks are given on the extent of success amongst the scientific team in anticipating the course of the eruption and the prognosis.


Since the onset of dome growth in mid-November 1995 approximately 77 x 106 m3 of andesite magma has erupted as of 27th March 1996.. Experimental and analytical studies of the magma constrain its origin from a source region at between 5 and 6 km depth with an initial temperature of between 820 and 850oC. The magma has been intermingled with higher temperature magma, likely derived from a deeper source and responsible for a heating event that has affected the andesite magma. The magma lost much of its gas on the way to the surface and crystallised extensively. As a consequence the lava erupts with very high viscosity. However it is clear that the lava does not lose all its gas efficiently and that high pressures can develop in the erupting magma. Evidence for high pressures include September 17th explosive eruption and the long period earthquakes are attributed to movement of pressurised fluid along fractures.

Since the beginning of the eruption the volume of erupted magma has been documented with determinations of volume being accomplished at intervals varying from one week to a month. The rate of eruption has shown marked variations on time scales of a few hours to a few weeks. Some of the largest pyroclastic flow events and the 17th September 1996 explosive eruption were associated with extrusion rates of up to 700,000 m3/day. Despite marked fluctuations, inspection of the volume data since magma first reached the surface indicates that the rate of eruption built up in the first few months and has then maintained a fairly consistent rate of about 200,000 m3/day ever since. Furthermore the height of the dome at the time of this assessment is 942 m asl, which is approaching the greatest height of 970 m asl in July 1996. A major conclusion is that the rate of activity has remained at a high level for almost a year with no signs of diminishing and no signs of any significant decay in the driving pressure. These results are the basis of a consensus amongst the scientists that the eruption shows no signs of finishing and is likely to continue for a considerable time to come.

Seismicity has been a primary monitoring tool throughout the eruption and the seismic network was enhanced in October 1996 when the broad-band network was successfully installed. The seismicity has shown a rich diversity of phenomena. Earthquakes have been predominantly shallow (ie less than 2 km depth) and located within or beneath the dome, apart from deeper earthquakes in the early phase of the eruption and for a few isolated instances after the 17th September explosive eruption. Earthquakes early in the eruption were more widespread, with locations beneath St Georges Hill and along a broadly NE-SW line extending from Englishs Crater to Long Ground and out to sea. Types of earthquake include classic volcano-tectonic earthquakes, long period events, continuous tremor, events related to rockfalls from the growing dome and the unstable Galways Wall and shallow earthquakes known as hybrid events. The hybrid events have dominated activity since the 17th September explosive eruption and are characterised by both high frequency components similar to volcano-tectonic earthquakes and long period components. The MVO is currently revising the classification of the earthquakes. The earthquakes described here as hybrid are reported as VT earthquakes in the daily and weekly scientific reports. The hybrid earthquakes occur in swarms and in the October to early December 1996 period a remarkable anti-correlation developed between swarms and dome growth. The earthquake patterns have been more complex since the New Year began with the reappearance of banded tremor and episodes of increase in the frequency of hybrid earthquakes merging into continuous tremor followed by generation of rockfalls and pyroclastic flows activity on the dome. These observations demonstrate that magma production, dome growth and earthquake activity are intimately related. While present understanding of these relations is very limited and inhibits their use as predictors of short-term activity, they are the most important diagnostic that activity is ongoing at depth.

Deformation studies, using the EDM and GPS methods and supplemented by studies of fracture movements and a tilt station on Chances Peak, have continued to generate important information on the eruption. The most significant finding is that the deformation is almost entirely confined to areas close to the dome. The GPS surveys show that locations further than about 1km from the dome have shown almost no relative horizontal movements since 1972, when the area was first surveyed, and during many repeated measurements during the eruption. There is some evidence emerging of slight subsidence on these distal locations which could prove significant. The minor horizontal motions on some lines between measurement stations were largest in the early phase of the eruption. Various measurements show large movements of the ground close to the growing dome. The old Castle Peak site on the SE margin of the dome has been monitored by EDM throughout the eruption until 2 months ago when it was finally destroyed by the encroachment of the dome. It typically expanded outwards from the dome at several mm/day with a total movement of over 1.3 m. Other sites on Farrells crater wall and Chances Peak have also shown evidence of outward motion. The lack of a widespread deformation field is not consistent with a large relatively shallow magma chamber deforming the crust elastically. The most dramatic deformation of all, other than the dome itself, has been that related to the deformation of Galways Wall, which began to crumble in early October with development of extensive and deep fractures on the southern shoulder of Chances Peak. Much of the upper parts of the wall have now disintegrated and recently rockfalls from the still incandescent pre-September 17 dome have moved over the wall down towards the Galways Soufriere. Movement on the Chances Peak cracks have continued up to the time of writing with movement rates being largest during earthquake swarms. When the Galways wall first started to deform the most likely outcome for the next three months was assessed to be slow crumbling and this has proved to be the case. There still exists a fairly wide spectrum of scientific opinion on the likelihood of a substantial failure of the wall and the threat of a lateral blast. Generally there is a consensus that the threat of a large-scale sudden failure has diminished. The most recent development is minor signs of instability developing at the Gages Wall. Gas studies have involved the now routine surveying of the plume by the COSPEC method, some exploratory work with the FTIR method and information on the gases in glass inclusions preserved from the deep source chamber. These studies have confirmed that this is a very sulphur poor volcano and that the major gas species is water with a concentration of about 4% in the original magma. Studies in the early stages of the eruption indicate a correlation of SO2 flux with seismicity. High rates of dome growth and pyroclastic flow generation also approximately coincide with higher SO2 fluxes. Work by the Open University group has shown that the concentrations of SO2 in the plume are affected by chemical reactions within the plume as it moves away from the volcano, introducing added uncertainty to the measurements.

Expert Elicitation

The expert elicitation method has been used to assess the opinions of the scientific team and help decision making. The results reflect the matters on which there is a consensus and those where there are some differences of view as well as providing an indication on the level of uncertainty on any single issue. However results must be viewed with some circumspection. The Soufriere Hills eruption is the first time that the method has been used during a crisis. It should be regarded as experimental and the results as just part of the information that the Chief Scientist on duty uses to advise the authorities.

There is no rational basis for predicting the length of the eruption, but there is a strong consensus that the eruption will continue for some time to come. The mean duration in the elicitation of 2.4 years should not be taken literally as a prediction. Interestingly a total duration of the dome eruption of 4 years would be comparable to the length of previous volcano-tectonic crises on Montserrat and similar to the average world-wide length of lava dome eruptions. The upper end of the range (about 50 years) reflects the fact that some lava dome eruptions develop into activity lasting for tens of years. The relaxation in view of a catastrophic failure of the Galways wall is reflected in the significant reduction of the percentage probability estimates of a catastrophic failure from the elicitation carried out in November 1996. The wide range of opinions on various scenarios for explosive activity reflect both some teething problems now being recognised in the elicitation approach and considerable uncertainty in the scientific understanding of the exact circumstances which lead to an explosive eruption and on the conditions which determine the intensity of such an eruption.

Concluding Comments

It is worthwhile to assess how well the scientists have done in forecasting the eruptive activity. From the administrators point of view, the most obvious and important conclusion is that the eruption shows no signs as yet of stopping. The scientific consensus, articulated more than a year ago, remains that this is very likely to be a long-lived eruption that could last for years. The scenarios that have been identified as most likely in earlier scientific assessments have proved to be fairly accurate. The generation of pyroclastic flows on the eastern unconstrained side of Englishs Crater has been the dominant hazard of the eruption and further pyroclastic flows moving down the Tar River are still seen as the most probable activity over the next few months. Early assessments of the probability of a significant explosive eruption were 10%. The fact that the build-up to the 17th September explosive eruption was not recognised as such at the time it was happening, shows that assured interpretations of transient, marginal and often ambiguous manifestations of deeper, hidden processes can be difficult. There is a consensus that the probability of movement into an explosive eruption can be increased by sustained periods of pyroclastic flow production, which decompresses the deeper parts of the dome and upper parts of the conduit. The Galways wall remains a threat but has crumbled significantly, and the level of concern amongst the scientists is now reduced. Small avalanches of the dome over the Galways Wall have already occurred so that the area around the White River towards St Patricks is now viewed as at increasing risk from pyroclastic flows. The dome is currently growing in the south adjacent to the Galways wall. The pre-September dome has largelt disappeared in this area.and the dome has no barrier to avalanching over the Wall. Plymouth remains vulnerable from a switch in dome activity or pyroclastic flows associated with explosive eruption.

Montserrat Volcano Observatory