12 MARCH 1999

CONTRIBUTED BY DR P J BAXTER

SCIENTIFIC AND HAZARDS ASSESSMENT OF THE SOUFRIERE HILLS VOLCANO, MONTSERRAT: REPORT OF A MEETING HELD IN TRINIDAD, 24-25 FEBRUARY 1999

Executive Summary

A

Growth of the lava dome at the Soufriere Hills Volcano ceased in early March 1998 when the magmatic eruption stopped. There has been residual activity and instability over the last year, but no sign of the magmatic eruption starting again. The post-eruptive residual volcanic activity has included occasional collapse of the lava dome to generate pyroclastic flows, vigorous venting of ash and gas sometimes accompanied by small explosions and the generation of small but mobile pyroclastic flows, occasional swarms of volcano-tectonic earthquakes, and continued ground deformation on the eastern flanks of the volcano. There is also a large amount of loose volcanic debris and ash on the slopes of the volcano which has been remobilised to form volcanic mudflows during periods of heavy rain.

B

The residual activity over the last year is thought to have been caused by the continued release and ascent of volcanic gases from the source of the magma deeper in the earth. The rise of pressurised gas makes the dome unstable, is associated with the continued low level earthquake activity and is responsible for minor explosions, some pyroclastic flows and the ash venting. Such activity is known at other volcanoes as an eruption is ending and may continue for many years. However, the scale and magnitude of this activity and the associated hazards are expected to diminish with time.

C

The residual activity together with the occurrence of volcanic mudflows during intense rainfall and the substantial amounts of volcanic ash deposits will continue to pose a number of hazards in southern Montserrat.

D

The Tar River valley, Plymouth and the north east flanks of the volcano are particularly dangerous. Further collapses of the lava dome and pyroclastic flows generated during periods of ash venting threaten major valleys on the flanks of the volcano. However, the reduction in dome volume since 3 July 1998 and the change in its morphology and configuration within the crater, led the meeting to conclude that the potential for big gravitational collapses (and hence large pyroclastic flows) is now much reduced. The collapses which could be generated during periods of ash venting are most likely to be confined to the Tar River valley and to Gage’s Valley-Fort Ghaut, although the possibility of avalanches in other directions cannot be completely ruled out. Because it is thought unlikely that pyroclastic flows will extend as far as the largest pyroclastic flows of the eruption, the areas already inundated by flows define a logical zone of high hazard (fig 1). Within this zone pyroclastic flows, which can be generated with no warning, could be lethal.

E

It is now a year since magmatic activity stopped and so some of the areas in the current exclusion zone are no longer considered under direct threat, provided that a magmatic eruption does not restart. However, many of these areas have substantial amounts of fine volcanic ash and further minor ash falls are likely. The health risks to potential residents from volcanic ash in these areas should be subject to expert medical assessment (see annex provided by Dr P J Baxter).

F

The meeting reconsidered the issue of the likelihood of no further magmatic eruption in the next 6 months, (previously assessed at about 95% probability in the July 1998 study) and, on the strength of the evidence gathered in the intervening period, concluded that there was no significant new information or basis for revising this estimate.

G

The results of the July 1998 elicitation on the chances of magmatic activity resuming over the longer term were also reviewed, and no changes were proposed to the estimates provided previously (i.e. 15% probability of reactivation within 5 years; 25% probability within10 to 30 years).

H

The present assessment, taking account of the possibility of both magmatic and non-magmatic activity, indicates a slight reduction in overall risk compared with the July 1998 appraisal. The results of the latter indicated that the risk levels in the populated areas of Montserrat were no worse than those to which populations on other Caribbean islands with dormant volcanoes have been historically exposed. The risks in the currently populated areas of Montserrat remain LOW to NEGLIGIBLE on the CMO’s Risk Scale, with IRPA values (annualised individual risk exposure) ranging from about 1 in 2,500 in parts of Area 4 marginal to the Belham valley, to less than 1 in 200,000 for Area 1. However, the increase in population numbers on Montserrat in recent months suggests that the societal risk of multiple casualties in the event of renewed magmatic eruption may be slightly increased.

I

I The threat to marginal areas of the current exclusion zone along the north side of the Belham valley has come down as a consequence of the reduction, since July 1998, of the volume of material comprising the dome edifice in the crater and the change in its morphology. Under the present circumstances of no magmatic activity, the chances of major flows or surges affecting even the margins of these occupied areas are lessened, and the risk to individual life and limb is now assessed to be MINIMAL (IRPA ~ 1 in 200,000) for people residing on the north side of the Belham valley.

J

A risk analysis has been undertaken for partial re-occupation of areas south of the Belham River (Iles Bay; Cork Hill; St George’s Hill; Richmond Hill and Fox’s Bay) on the basis that, for the near future, only limited numbers of returning residents would be involved in each case. Under present conditions at the volcano (i.e. no magmatic activity), the annualised individual risk of fatality for residents in one of these areas from dome collapse is estimated to be of order 1:50,000, or VERY LOW

on the CMO’s scale. However, if magmatic activity were renewed, the risk exposure for individuals would immediately become MODERATE (IRPA ~ 1 in 250) requiring an immediate decision regarding re-evacuation.

K

St. George’s Hill has been identified as a possible vantage point for tourists and other visitors to view the volcano and the effects on Plymouth: given the relatively short timescale for which they would be present there, and their presumed mobility by taxi, the annualised individual risk of fatality must be considerably lower than that of any permanent resident (above reported to be about 1:50,000 or VERY LOW on the CMO’s scale), while non-magmatic conditions exist.

L

The risk for workers undertaking remedial work in and around the Bramble Airport area under present non-magmatic conditions has been assessed from a model based on 25 workers operating 8 hours per day, 7 days a week, for a range of scenarios. The two end members of this range are: 1] If measures are taken to monitor the volcano very closely and there is assumed to be a 90% chance of successfully detecting the start of a dome collapse and giving immediate warning, it is estimated that the individual fatality risk exposure (IRPA) for alerted workers at or near Bramble airport would be about 1:16,000 p.a. 2] For the worst case scenario in which the probability of a big collapse is treated very conservatively AND no effective warning of sudden collapse is given, the IRPA for workers may reach 1:650 p.a. (i.e. comparable to offshore oil workers in the North Sea). A best estimate risk value would fall between these extremes, but requires further scientific input.

M

For flight operations at Bramble airport, a revised risk analysis in which the possibility of magmatic reactivation occurring sometime within the next 30 years is taken into account (not just the present non-magmatic conditions) has been completed. This indicates that if ground activities were implemented well away from the airstrip (as discussed in more detail in the September 1998 report), the reduced danger of large dome collapses produces a moderate drop in overall risk exposure: over the long term, the risk to a group of people going onto the airstrip for a very short time to board or leave a flight, or to perform ground support duties, is estimated to be fractionally lower than that for permanent residents living on the margins of Area 4, for example. For the individual going to the airstrip for a short time only, the estimated IRPA is about 1 in 7000, which exposure is similar to someone living full-time in Area 3 (Woodlands), for instance.

N

Heat dissipation in a recently extruded dome is very slow and, given the relatively large volume of material involved in the 1995-98 eruption, it is considered likely that, if further major collapses do not occur, many decades (or even a century or two) will have to elapse before the heat energy contained within the dome is sufficiently reduced to effectively remove all hazard potential from that source.

O

Volcanic mudflows are generated only during periods of intense rainfall and dangerous situations can be easily identified. The situation at the Belham River crossing is no different to many parts of the world where periods of intense rainfall can cause mudflows and floods and populations routinely cope with the situation. In terms of risk to individuals from dome collapse pyroclastic flows and surges when crossing the river, the limited exposure time and small numbers involved at any one time suggest the risk levels should be no worse than for residents living continuously in a re-occupied Cork Hill, for instance.

P

Continued careful monitoring of the volcano is required to help minimise exposure to residual hazards and to ensure identification of any signs of resumption of magmatic eruption.

Q

MVO is confident that renewed magma ascent can be detected and that escalation to dangerous levels of activity can be recognised. In some circumstances, escalation might take place in a very short period of time (a matter of hours). However, any warning helps to significantly reduce the risk exposure of the population below that estimated for an unprepared society, and risks will be continuously reassessed by the MVO.

Introduction

1. A meeting of MVO scientists took place at the Seismic Research Unit of the University of the West Indies, Trinidad, on 24, 25 and 26 February 1999 for the purpose of evaluation of the status of the Soufriere Hills Volcano. The meeting was commissioned by HMG to take place approximately six months after the previous evaluation on July 1998.
2. The assessment of July 1998 recorded that the dome of the Soufriere Hills Volcano had stopped growing in early March 1998. Since that time there is no evidence of any further magma ascent from depth to the Earth's surface. Thus the primary magmatic eruption has stopped. However, there is residual volcanic activity which still causes hazards in southern Montserrat.
3. This report evaluates the residual volcanic activity in the period following magmatic eruption. The activity since July 1998 is summarised, the potential hazards associated with this activity are identified and the implications of this assessment for the vulnerability of different areas on Montserrat are given. The hazards to specific places are considered in the context of this analysis as requested by DFID. An update of the formal risk analysis for Montserrat is presented.
4. In the following report it is very important for the reader and user to understand the terminology that scientists use to describe the state of a volcano. From a scientific point of view an eruption can be said to have stopped when magma (molten rock) is no longer rising to the earth's surface. However, there can still be activity at the surface even when magma is no longer rising and this activity can be impressive and hazardous. In this report such post-eruption phenomena are described as residual volcanic activity. Following a major volcanic eruption like the Soufriere Hills in 1995-1998 the residual volcanic activity can last for a considerable period of time (often months or years) before gradually returning to the background low level activity of a dormant volcano.

 

Activity July 1998 to February 1999

5. The residual post-eruption activity has included several collapses of the andesite lava dome accompanied by occasional minor explosions, vigorous ash venting and gas release. The collapse of 3 July 1998 formed a major pyroclastic flow down the Tar River valley. On 12 November a deep canyon-like depression on the southern side of the dome developed. The canyon was enlarged by further collapses, which extended the deep canyon to the WNW and made the depression open both to the west and east. About 30% of the dome has been removed. The new canyon cut the dome in two and the precipitous walls continue to be very unstable.

6. The July 3 1998 collapse revealed vents on the floor of the new depression from which ash, steam and volcanic gases discharged. The vents are aligned along approximately east-west fractures. There have been several periods of strong ash venting and gas discharge over the last year. Some episodes were initiated by explosions and collapses of parts of the dome to form pyroclastic flows. Some of the more vigorous explosions and ash venting has formed fine-grained but highly mobile pyroclastic flows which have flowed down the flanks of the volcano. On November 12 a large collapse of the dome caused pyroclastic flows to go down the Gages valley into Plymouth as well as down the Tar River and White River. Temperatures of over 350ºC were measured in the deposits, demonstrating that the interior of the volcano is still very hot and capable of producing very dangerous flows with no warning.
7. Seismic activity has consisted largely of volcano-tectonic earthquakes and some signals from rockfalls and pyroclastic flows. There are very few hybrid earthquakes, which were the dominant seismicity during active dome growth. Volcano-tectonic earthquake swarms accompany the major episodes of collapse and ash venting. Swarms can both precede and follow major dome collapses and explosions that initiate ash venting periods. The earthquakes have been located at depths of 1 to 3 km below sea level. The activity waxes and wanes over periods of a few weeks with very low activity and periods of enhanced activity such as October through to December 1998.
8. Ground deformation may still be occurring on the eastern flanks of the volcano. A new fracture has also been observed on a road on the eastern flanks. The continued ground deformation on the eastern flanks should be watched carefully for any signs of larger scale instability.
9. The fluxes of the volcanic gases sulphur dioxide and chlorine have been monitored. Although the rise of magma has stopped, the SO₂ flux has remained high with typical values between 500 and 1500 tons per day. The gas flux fluctuates substantially partly reflecting uncertainties in the measurements and partly reflecting real fluctuation. The periods of high flux correlate with the periods of vigorous ash venting. The ratio of sulphur to chlorine gases has increased by more than an order of magnitude in comparison with measurements made in 1996..
10. The scientific group agreed that the following explanation is consistent with the observations of these residual activities. The considerable gas fluxes and high ratio of sulphur to chlorine imply a relatively deep source for the volcanic gas. The eruption is thought to have been caused originally by a different kind of magma known as basalt being intruded at depths greater than about 5 or 6 km and pushing out the andesite magma. The basalt is now thought to be cooling and solidifying at depth and releasing gases which rise along deep fractures to the surface. These gases increase the pressure in the interior of the volcano which is released in explosions and vigorous venting of ash and gas. The fractures feeding the pressurised gases to the surface are oriented east-west and have developed underneath the dome and destabilised it.
11. Residual activity, particularly continued gas release and earthquakes related to post-eruption adjustments of a disturbed volcano, is typical after a major volcanic eruption. Such activity can continue for many months to years. However in general such activity diminishes with time as the volcanic system settles. In the case of the Soufriere Hills the residual activity has already lasted a year and the gas fluxes have not yet shown the decline that has been observed on other volcanoes.
12. The magmatic eruption emplaced about 100 million cubic metres of loose ash and volcanic debris on the flanks of the volcano. During the hurricane season in 1998 there were intense periods of rainfall, in particular during the passage of Hurricane Georges. Volcanic mudflows were formed in the hurricane and discharged large amounts of debris into the lower reaches of the major river valleys. In particular mudflow debris inundated Plymouth further burying and damaging buildings in the centre of town, buried the Belham river bridge and spread debris in the vicinity of Bramble airport. In the case of the Belham river much of the debris in the catchment area has already been swept to the lower reaches of the valley and into the sea at Old Road Bay. However, there still remain large amounts of loose material on the upper flanks and valleys of other catchments.
13. Despite heavy rainfall over the last several months thick ash fall deposits still drape much of the western flanks of the volcano with the ash thickening substantially from Cork Hill to Plymouth.

 

Residual Hazards

14. . Pyroclastic flows remain a major hazard. The dome is precipitous and unstable, particularly along the newly formed east-west canyon. The interior of the dome is also still very hot and is likely to remain so, potentially for decades. The modifications in the dome and reduction in its size have changed the degree of hazard around the volcano since the July 1998 assessment. The dome has been consistently collapsing into the new east-west canyon, making collapses into this feature the favoured area for future instability. Pyroclastic flows therefore can be expected down the Tar River and Gages valley with the Plymouth area being very vulnerable. The dome has reduced in size so the capacity to cause very large pyroclastic flows is decreasing. For example the probability of a collapse three times larger than the reference collapse of 21 September 1997 has reduced greatly and is considered negligible in some directions, thus reducing the hazard. There remains a large mass of dome material on the northern side over the valleys to the north-east towards Bramble airport and towards the Belham valley. However, this material has remained stable for over a year. With time it is expected that continued cooling, consolidation, degradation and collapse of the dome will reduce the hazard further.
15. The most vigorous ash venting episodes are initiated by explosions and can form another type of pyroclastic flow. These flows are mobile and hot and can travel down several valleys simultaneously. As is the case for pyroclastic flows from dome collapse these flows can occur with no warning. Particular attention is drawn to the vulnerability of Plymouth.
16. The boundaries of the exclusion zone have been considered during this assessment in the light of the reduction in the hazard now that a year has passed since magmatic activity stopped. As long as magmatic activity does not start again then the scientific consensus is that pyroclastic flows are very unlikely to extend beyond the largest flows that occurred during the eruption. Thus an objective and conservative boundary for a zone of high hazard can be defined as the limit of areas already inundated by pyroclastic flows and surges. The accompanying map (Figure 1) shows the boundary of the zone based on this criterion with the boundary defined as 100 m beyond the extent of deposits in areas of gentle relief and 50 m above the limit of surge clouds in areas of rugged relief. The group recommend that the area defined by this boundary is used as a guide for identifying a revised exclusion zone.
17. Volcanic mudflows will continue to occur during periods of intense rain. Mudflows should not be a concern provided the susceptible valleys are avoided during periods of high rainfall. The situation at the Belham River Crossing is no different to many other parts of the world where high rainfall causes flash floods and mudflows. The populations in countries like Iceland and Chile routinely deal with such situations.
18. It is likely that minor ash falls will occur from time to time as further ash venting episodes are expected. Due to the prevailing wind directions at Montserrat the likelihood of ashfall increases significantly as the western flanks of the volcano are approached. There are still large amounts of fallen ash in some of those areas south of the Belham valley which are now regarded as at low risk from other volcanic hazards. This report refers to the July 1998 report where the issues of the health effects of the ash are considered and the requirements for clean-up and monitoring are stated. A brief annex to the current report has been contributed by Dr P J Baxter

Probabilistic assessment of current volcanic hazards

19. The objective of the present risk assessment is to update earlier calculations of the potential loss of life from direct volcanic hazards and make adjustments which arise from changes in perceived likelihoods of occurrence for the various identified threats. The timeframe for the assessment exercise covers the next six months, as in previous assessments. The outlook for general eruptive activity levels over longer defined future time periods (i.e. 5 years; 30 years) remains unchanged from the last assessment in July 1998. (The present analysis does not include estimates of numbers of injured persons - which, for emergency planning purposes, might be inferred from fatality numbers by medical specialists and volcano emergency specialists - or long-term health risks from ash).

20. For the present risk assessment, some modifications have been made to previous versions of the map defining areas of population clusters (Figure 2). The boundaries of the Population Area 4 have now been redrawn to enclose only Old Towne, Salem and Frith, where formerly it extended across the Belham River to encompass the coastal strip down the western side of Garibaldi Hill (in previous assessments, there were assumed to be no residents on the south side of the river, anyway). Five new (un-numbered) subdivisions are introduced on Figure 2: Iles Bay; Cork Hill and Devlins; St George’s Hill and Weekes; Richmond Hill, and Fox’s Bay, to allow estimation of risk exposure in these areas if limited re-occupation is contemplated. A separate sub-area for Bramble Airport is also included.

21. In order to take account in the risk calculations of the changing population numbers on Montserrat, the most recent population total (about 4,400 persons, according to the Statistics Department of the GoM in February 1999) has been used to provide updated estimates of populations in the four currently-occupied geographical zones 1 - 4 (see also Figure 2), with the corresponding partitioning for risk assessment modelling purposes as indicated on Table 1. These approximate figures reflect the slight increase in overall numbers when compared with those used in the earlier assessments, but individual area totals have no real claim to absolute accuracy.

Table 1: Numbers of persons in each Population Zone used for risk analysis calculations, compared with values used in the three previous assessments.

22. It is required to consider risk exposure levels if partial re-occupation of some areas south of the Belham River were allowed in the near future. For risk modelling purposes, limited numbers of returning residents are assumed to be involved in each case, with upper limit population numbers for the time being of 50; 200; 100; 100; 50 into Iles Bay; Cork Hill; St George’s Hill; Richmond Hill and Fox’s Bay, respectively. These numbers are arbitrarily assumed to be drawn mainly from Population Zones 1 and 2, and from people returning to Montserrat from abroad, and are intended for guidance only.

23. The hazard and risk assessment approach and methodology follows that described in detail in the December 1997 report, subsequently validated by the UK Government’s Chief Scientific Adviser’s consultative group. The approach has been to assemble all plausible volcanic scenarios for the Soufriere Hills volcano on a logic-tree framework (Figure 3), with branching to accommodate a hierarchy of related hazards. A Monte Carlo technique is then used to re-sample the probability distributions for each branch of the logic-tree, for the given population distribution, to determine the corresponding levels of risk exposure in different population areas, arising from such hazards. Within the Monte Carlo simulations, statistical distributions have been used, wherever appropriate, to represent the spread of scientific uncertainty associated with individual factors, estimates of frequency of occurrence, or other inputs.

24. These risk analyses are based on the current scientific assessment of potential volcanic hazards. If volcanic activity changes, a significant event occurs, or important new information arises (e.g. from monitoring), then the assessed probabilities of occurrence of hazardous events may also change. Regular updates of risk analyses therefore contribute an essential element to the on-going process of hazard management in Montserrat.

25. The meeting considered all the scientific evidence which had been gathered throughout the crisis and, in particular, that which had accumulated since the last meeting in July 1998. A review of the relative probabilities of important scenarios was undertaken, and the principal changes are summarized on the event-tree of Figure 3. Because of the requirement to estimate risk in specific areas south of the Belham River, such as Richmond Hill and Fox’s Bay, a pair of branches have been added to the event tree (Figure 3), comprising representative instances of two magnitudes of dome collapse out of the Gage’s sector of the volcano.

26. On this occasion, the changes to the probability tree comprise reductions to the assessed probabilities of dome collapses in key directions, reflecting the reduction in dome volume since 3 July 1998 and the change in its morphology and configuration within the crater. The total volume of dome material in the crater, as at February 1999, was reported by MVO to be 77 x 10⁶ m³, of which two-thirds is contained in three heaps on the north side of the crater, the remainder sitting over the former Galway’s wall area. The triggering of dome collapses by external events, such as a major nearby regional earthquake, was discussed by the meeting, and this was identified as the most likely cause of significant avalanching from the stagnant dome remnants. However, it has to be recognised also that considerable uncertainty surrounds the setting off of dynamic avalanching processes from big domes containing high heat energy contents and high, possibly non-uniform internal pressurisation. Initiation of failure may not necessarily take place at the point on the dome where simple gravitational instability is greatest. Thus, although collapses into the 3 July 1998 scar are considered most likely, the possibility of an outburst in a different direction cannot be precluded.

27. Given these circumstances, the meeting concluded that the potential for a really big gravitational collapse, such as a 3x reference event, or larger, (and hence associated large pyroclastic flows) affecting the Belham River valley is now reduced to a negligible level, and a conditional probability of occurrence in the next six months not exceeding 10^-6 was adopted. The present dome configuration and local topography above Tuitt’s Ghaut suggests that, if provoked by a large earthquake, for instance, such a collapse towards Bramble airport has a slightly greater chance of being triggered, for which the conditional probability of occurrence is set at about 10^-5 in the same interval. The possibility of a big collapse out of the Gages area of the crater, perhaps involving some pre-eruptive material as well (as in the Boxing Day 1997 event), is also considered conceivable, and incorporated conservatively in the model at a conditional probability of 10^-3.

28. Smaller collapses, represented in magnitude by the 1x reference event, if they occur, are likely to be confined to the Tar River valley and to Gage’s valley-Fort Ghaut, although for the reasons given in para 24 above, the possibility of avalanches of this magnitude spilling out in other directions cannot be completely precluded. The meeting felt there was about a 5% chance of a dome collapse of this magnitude in the next six months, so in the risk model, the probability of a 1x reference collapse down Gage’s is ascribed a likelihood of about 3.3% in the next six months, twice that for one in the Bramble direction (1.6% probability), and about seven times that for a similar event occurring into the upper Belham River valley (0.5% probability).

29. The meeting determined that the assessed likelihood of magmatic reactivation of the volcano in the next six months (see lower branches on Figure 3) should remain unchanged from the last elicitation, at a 5% chance in the next six months. The chances of a major life-threatening magmatic eruption, which could impact outside the present exclusion zone in the next six months (assessed last time at about 0.7%), also remains unaltered: the equivalent annual probability is just over 1%, which is close to the long-term average occurrence rate for an Eastern Caribbean volcanic island, as indicated by historical experience. In other words, the threat from the volcano in its present