Introductory Note
Shepherd arrived in Montserrat on July 4 1996 and Stasiuk on July 24. Both are on contract to the BGS to assist in the monitoring of the Soufriere Hills volcano and are required to report on their activities to HE the Governor, who has given his approval for this report to be circulated, and to the BGS. In order to do this properly it is necessary to describe the progress of volcanic activity during the reporting period. Shepherd has mainly been concerned with the ground deformation programme and Stasiuk with geological investigations but each has collaborated in the work of the other.
Much of the basic information is summarised from the daily reports and morning updates of the MVO. Details of numbers of different types of earthquakes are in those reports and are not repeated here.
In compiling this report we have consulted with other scientists working on the volcano, particularly Peter Francis and Mark Davies of the Open University, Clive Oppenheimer of Cambridge Univerisity, Anne-Marie Lejeune of Bristol University, Jean-Christophe Komorowski of the Institut de Physique du Globe de Paris, Allan Smith and Glen Mattioli of the University of Puerto Rico, John Stix of the University of Montreal, and Rod Stewart of the BGS. Jane Toothill and Rob Watts have assisted considerably in the fieldwork and provided useful comments. The views expressed however are those of Shepherd and Stasiuk and do not represent the official view of the MVO or the opinions of the named scientists.
General overview for the period July 4 to August 10
At the beginning of this period activity at the Soufriere Hills volcano was at a low level. Seismic activity consisted mainly of rockfalls but low amplitude broadband tremor occurred regularly from July 4 onwards and hybrid and long period events were reported intermittently. The number of so-called volcano-tectonic earthquakes (VT's) remained low throughout the first half of July. Poor visibility considerably hampered visual inspection of the dome although there were generally short visibility windows from the east (Whites, Long Ground and Tar River) on most mornings. The poor visibility presented a considerable obstacle to the programme of dome volume estimation from photographs and only one quantitative estimate was made. This indicated that the volume of the dome itself had reached 26 million cubic meters on July 7 with about another 4 million cubic meters of talus deposits mainly in the upper part of the Tar River valley. Until about July 20 the growth of the dome was predominantly to the south with talus building against Galways Wall. On the evening of July 20 a period of intense shallow earthquake activity began. From this point onwards the area of growth of the dome and most of the visible rockfall activity was mainly confined to the eastern flank to the north of Castle Peak. These changes signalled a distinct escalation in the level of activity at the volcano and further developments will be discussed after individual measurements have been described..
Ground Deformation
Total Station (EDM) measurements continued on the eastern and northern triangles whenever visibility permitted. The reflector at Chances Peak was obscured by mist and a constant light drizzle of ash throughout the period and the reflector on Gages Wall has long since disappeared. The shortening of the lines from Castle Peak to Whites and Long Ground continued at a slightly accelerating rate but no significant changes were detected on the northern triangle. By July 18 the rate of shortening on the Castle Peak lines had increased to 15-18mm per day which is about one order of magnitude greater than the mean rate measured since December 1995. This rate continues to the present.
GPS and related measurements
Two complementary programmes of GPS measurements were continued throughout this period. The University of Puerto Rico programmme which began in August 1995 measures baseline lengths on a number of lines which are of order 5-6 km long and which cross the whole volcanic edifice. Previously measurements on these lines were made at irregular intervals but two continuously-operating stations have been established and are now in operation.
The MVO GPS programmes measures baseline lengths over shorter distances (average 2 - 3 km) using the Rapid Static method. The primary aim of this program was to link together the existing Total Station (EDM) points into a single network and to provide accurately-located benchmarks for gravity observations and direct measurements to the surface of the dome and to the surfaces of fresh volcanic deposits. A technique has now been successfully tested by which GPS equipment is mounted in the helicopter and provides an accurate three-dimensional location for the helicopter as it hovers near the dome or briefly touches down in safe locations. Simultaneously measurements of distance, elevation and azimuth from the helicopter to the surface of the dome are made using Leica Vector rangefinding binoculars. This technique will enable us to map the surface of the dome and of the deposits from successive eruptions with much greater accuracy than has been possible previously, and on a nearly daily basis. This is an important improvement for the purposes of monitoring since the recent behaviour is characterized by dramatic, rapid changes in the dome and pyroclastic deposits.
In terms of deformation of the volcanic edifice the results of the two programs are in agreement. Some baseline lengths show generally incoherent short-term variations of the order of a few centimetres. It is not yet clear whether these short term variations are of instrumental origin or whether they have a genuine volcanic origin. When the results are averaged over more than a few days there are no long-term deformation trends at levels greater than one standard deviation of the measurements. One exception to this general result involves the UPR measurements to the Chance's Peak site which continue to show displacements above the one sigma level indicating continued outward movement. Taken in conjunction withe the continued displacements on the Total Station lines to Castle Peak (and absence of significant displacements on other lines) these results strongly indicate that deformation associated with this eruption is confined to a region not more than one kilometer from the growing dome.
In conjunction with the MVO progamme a comprehensive gravity survey of the volcano was carried out between July 12 and July 27. A total of over one hundred new gravity stations were occupied and simultaneously located by GPS. These stations are mainly in a number of lines radial to the volcano passing up the steps to Chances Peak, through Plymouth to Gage's Soufriere, from Farrells to the crater rim, from Whites to Hermitage, and from Perche's Estate to Roaches Mountain. A preliminary gravity map of the volcano is in preparation. Twenty-three of these stations were selected as microgravity monitoring stations and will be re-occupied within the next few days
Gas measurements
Half-wavenumber resolution spectra have been acquired in the fingerprint region of the infrared in order to detect and measure the concentrations of volcanic gases in the atmosphere. Data have been collected in several locations using both an artificial source of infrared radiation and the hot lava dome itself as a source of IR. Several measurements were acquired from a helicopter. A preliminary analysis indicates high emissions of gaseous HCl with concentrations of up to 100 ppb in Plymouth, up to 200 ppb in Amersham, and 1 ppm close to the dome (averaged over path lengths of 150-1500 m). HF, CO, COS, SiF4 and HBr have not been detected so far. Surprisingly, SO2 is at or below the detection limit (10-50 ppb over typical path lengths), suggesting an unusually high chlorine:sulphur ratio in the gas phase and/or an unknown mechanism of separation of HCl and SO2 in the atmosphere. Analyses of magmtic gases collected directly from the dome in February did not report HCl concentrations, although evidence for abundant chlorine in fluid inclusions has been noted elsewhere. Data reduction continues and it is hoped that it will be possible to report gas analyses for comparison with earlier measurements soon.
A Minicospec manufactured by Barringer Research and provided through a NERC emergency grant was commissioned in late July and has been operating successfully. Since July 26th, SO2 fluxes have been somewhat elevated above the 200 +/-100 tonnes/day output that characterises the last 3 months of COSPEC observations. However, variability in a single day's measurements is generally of similar magnitude to the scatter expressed in the long-term dataset of averaged daily values. On August 8, two transects of the gas plume yielded fluxes well in excess of 1000 tonnes/day. One interpretation of this observation is that the volcano "puffs" intermittently, and it is recommended that COSPEC SO2 retrievals are compared with contemporaneous geophysical and other observations in order to achieve more useful interpretations of the dataset in future. Analysis of time-series of individual plume retrievals may be of greater benefit to volcano surveillance than interpretation of daily averaged values. For example, measurements taken during a day of major dome collapse showed particularly high values (500 tonnes/day) just prior to the major collapse event whereas earlier in the day and after the collapse event the values were between 100 and 200 tonnes/day. COSPEC data have also been obtained at variable distances from the active dome. These appear to show evidence for cloud-processing of SO2 to form aerosols.
Samples of fumarolic gases have been taken at Galways Soufriere and levels of Sulphur Dioxide in the atmosphere have been made regularly but the detailed results are not yet available. Water samples taken at Amersham have been tested for sulphate, fluoride and chloride and show no significant changes.
Geological Observations
>From approximately 20 July onward, the area of growth on the lava dome shifted to the east, close to Castle Peak. Until 20 July the growth rate of the dome was apparently similar to the previous months and fairly low relative to other lava domes worldwide, at about 2 cubic meters per second of magma reaching the surface. At about that time the growth rate increased dramatically. The increase was signalled by increasing numbers of hot rockfalls from the steepening, blocky surface of the dome in the eastern and northeastern sectors. There continues to be no evidence that the lava dome is capable of significant magmatic explosive behaviour, but there is clear evidence that at least one minor phreatic explosion occurred on about10 July. The ash deposits on Chances Peak were showered with small hydrothermally altered fragments on about that date, to a distance of about 500 m from the crater rim. The blast was witnessed by Glen Mattioli and Jean-Christophe Komorowski from Old Town. The blast apparently occurred on an inactive flank of the dome and shows that there are significant hazards associated with all parts of the dome.
Vigorous Activity of 24 July to Present
On 24 July the volcano showed a clear increase in its level of activity, and this has developed into a distinct phase which has had a major impact on the Tar River Valley and has important implications for hazards on Montserrat. The initiation of this phase was marked by a sharp increase in rockfalls off the NE dome flanks and in the level of seismic signals associated with subsurface processes. For the first time since November 1995, the seismographs recorded swarms of impulsive events (VTs) and long periods of almost harmonic tremor. The seismicity developed a strong pattern, in which the seismic energy output peaked sharply at almost exactly 4 hour intervals, which lasted until a few days ago. On 26 July, brief glimpses of the eastern sector of the dome through breaks in cloud cover revealed a new spine with an upward tapering form, about 50 m high and 50 m in diameter at its base. Clear conditions during the night of July 27 allowed extensive observation of the actively extruding spine. Incandescent rockfalls of metre-sized blocks were produced by almost continuous spalling from the spine shoulders. Throughout Juy 28 almost continuous rockfalls and small pyroclastic flows were generated in the upper Tar river valley. During the early part of the day the small pyroclastic flows did not travel further than the disused north-south road in the Tar River valley but they generated moderate-sized ash clouds which were blown to the west and resulted in moderate ash fall in most regions to the west of the volcano. From about 17:45 onwards the pyroclastic flows appeared to increase in size although they were not under direct observation. A continuous series of ash columns was generated which drifted to the west and resulted in light ash fall in Plymouth and surrounding regions.
During the period 25-29 July the rate of growth of the dome was estimated to have risen to 10-15 cubic meters per second, amongst the highest sustained rates recorded in historical dome-growth eruptions. During this period the vigour of collapsing material from the dome escalated. Small, regular rockfalls were observed to have runouts of perhaps 200 metres and to generate minor, dilute ash clouds at the start of the period, but these progressed to nearly continuous block and ash flows having runouts of 800 metres, velocities of about 70 km/h, with associated active surge clouds and generated columns about 1 km high. Seismic activity during this period had a distinct periodicity of about four hours. Relatively quiet periods were punctuated by increasing numbers of hybrid-type earthquakes which eventually merged into periods of continuous tremor which lasted for up to one hour before settling back to background level.
A major series of pyroclastic flows generated ash columns from 05:30 on July 29. During the morning of July 29 the new intrusion through the eastern flanks of the dome was observed briefly. The shape had changed from a truncated cone to a flat-topped plateau. Beginning at about 12:00 and again at 16:00 major periods of pyroclastic flow activity generated major ash columns which were carried by the wind in a southeasterly direction and deposited ash to a thickness of a few centimeters in St.Patricks. The period of 16:00 produced flows which reached the sea and were observed from the helicopter. These flows moved much more rapidly than the shorter flows observed earlier, and were much more mobile. The associated ash clouds clung to and followed nearly vertical cliffs.
The pyroclastic flows which reached the sea at the mouth of the Hot River built a fan which, by Sunday August 4, when it was surveyed in detail, extended about 400 meters into the sea and was about 550 meters wide at its base. The fan was built by a long series of pulsed flows rather than one major event but include blocks up to several meters in diameter. The flows caused extensive denudation of the Tar River valley in a swathe which was 500 meters wide in places. None of the pyroclastic flows overtopped the rim of English's Crater nor the main valley wall of the Tar River Valley, although significant surge deposits overtopped at the head of White's Ghaut and proceeded about 50 metres down the Ghaut, leaving scorched vegetation and knocking down small trees. Vegetation was scorched to the top of the north Tar River Valley wall in almost all locations. Almost all vegetation was removed from the Tar River Valley by the flows, and only minor scorched trees on steep leeward slopes and blown down or bent, charred trees high on the valley walls were left rooted. Brief views at night revealed that the NE sector of the dome continued to grow and showed a similar behaviour to that on 27 July, with nearly continuous incandescent rockfalls from the upper margins of the extruding material. On 30 July partial clearing allowed helicopter observations of the NE sector and revealed that the collapses of the previous day had not removed the entire spine (or it had been replaced by continued extrusion). The NE sector appeared to be only the jagged root of the recently extruded spine and was composed of a thick sheaf of dark-coloured slabs about 1 metre thick, with intervening scoriaceous material. This material was surrounded by fresh talus ridges which joined at the spine base and led continued rockfall material to a new erosional gully in the talus along the south margin of the Castle Peak. The helicopter altimeter indicated that the top of the new material was at a height of 2800 feet abve sea level.
The four-hour seismic periodicity continued through July 30 and the rate of extrusion continued at the elevated rate established around July 24. Again the rockfall behaviour escalated rapidly in vigour and on 31 July at about noon there were more large pyroclastic flows which reached the sea, extended the pyroclastic fan, further denuded the Tar River Valley and generated ash plumes up to about 5 or 6 km altitude with associated lightning. The entrance of the flows into the sea generated vigorous boiling of the sea and a large steam plume which mixed with the ash plumes from the pyroclastic flows. It appeared that this mixing of steam produced rains of accretionary lapilli and hence premature ash deposition. The collapses removed a major part of the dome, including more than the recently extruded spine and probably much of the material which began to be extruded in early July. The collapses left a spoon -shaped depression which cut mainly through the eastern talus slopes but extended upwards above and behind the old Castle Peak dome. The volume of material lost was estimated to be in the range 3-5 million cubic meters.
Profiles across the Upper Tar River Valley show that very little material was deposited except along the narrow axes of the deep gullies. The surge deposit thickness reached local maxima of 70 cm and the blocky pyroclastic flow deposits in the gullies had thicknesses up to about 15 m. The greater changes to the valley were a planing off of pre-existing material. Some small ridges, hills and irregularities were removed, and in one location the pre-existing ground surface was eroded down by about 25 m. These observations are consistent with preliminary volume estimates of the pyroclastic fan at the shore which have reasonable maxima somewhat greater than 5 million cubic metres of dense rock on 4 August. That is, as much or more material reached the sea than collapsed from the dome. The implications are that the pyroclastic flows were very mobile, probably stopped by entrance into the sea, and were highly erosive. Similar flows overtopping the rim of English's Crater would be likely to cut down rapidly through the crater rim structure and could trigger continued collapse.
As of 1 August, major collapses stopped temporarily and the dome became relatively quiet, although growth clearly continued at an elevated rate. On 2 August relatively good visibility allowed detailed inspections of the dome. The slip scar marking the limit of major collapse was still clearly visible but the void left by the collapses was almost completely refilled by a fresh extrusion of dark, slabby lava. Extremely vigorous fumarole activity at the highest part of the new material was noted and created a deep booming sound noted by scientists and locals of the Long Ground area. On this day the pyroclastic fan and Upper Tar River valley deposits were sampled. Many of the blocks were found to be highly scoriaceous, which is a distinct difference from the character of the previous year. By the night of 2 August the 4 hour periodicity in seismic energy output was re-established, and by the afternoon of 4 August, only 4 days after the climactic collapse, rockfalls from the NE dome sector and associated dilute ash plumes began again. The new NE flank continued to swell visibly during this period, based on brief and partial visual observations from the ground and helicopter. The rockfall behaviour escalated through the period 4 to 11 August.
On 9 and 10 August, partial visibility of the lower dome flank in the NE sector allowed GPS-vector binocular helicopter surveys of the dome and flanking talus shape. These data are still being processed but will define a baseline for comparison as the behaviour continues. On 10 August nearly continuous, pulsing small pyroclastic flows were observed in the first half of the day. The slip scar created in the 28 to 31 July period was almost completely obscured by the newly extruded material and its flanking talus. On 11 August, the pyroclastic flow behaviour escalated again, with the first flow reaching the sea at about 9 am and observed from the Police Launch, and continuous flows moving down the valley to distances of about 800 m throughout the morning. During the afternoon, the flows escalated again and climaxed at about 17:00 with a 1 hour period during which numerous large flows reached the sea and extended the pyroclastic fan. These flows were preceded by very loud continuous sounds of rockfalls from the dome (obscured by cloud and ash) and initially followed paths along the south and north sides of the Tar River Valley. The associated ash plumes again generated lightning and a wet rain of accretionary lapilli, to a maximum thickness in Plymouth of 3 cm.
Possible future activity
Prediction of the pattern of long-term future activity at the Soufriere Hills volcano remains extremely difficult. Since this eruption began in July 1995 it has continued to escalate although at a slow rate. There have been lulls in activity but each successive renewal of activity has been at a higher level than the previous one.
At the time of writing (August 14) the accelerated rate of dome growth which began in mid-July shows no clear signs of slowing down so that the pattern established over the past few weeks is likely to continue. If this happens further pyroclastic flows will continue to affect the Tar River valley, and ash falls with thicknesses of up to a few centimeters will continue to affect the western side of the volcano. The areas of maximum ash fall are determined by more-or-less random wind patterns in an arc around from St. Patricks in the southwest to Salem in the northwest but exceptional wind patterns could produce ash falls almost anywhere in the western half of the island. The correlation between pyroclastic flows entering the sea and the fall of wet accretionary lapilli on the western side of the volcano is now clearly established. The particular hazard implication of this phenomenon is that wet accretionary lapilli fall more rapidly than dry ash so that a greater proportion of the total ash produced falls on the island rather than in the sea. Wet ash is heavier than dry ash so that the hazard of accumulated ash on roofs becomes greater. Roof collapse has not yet become a significant hazard but it may become one in the next few weeks if the rapid rate of dome growth continues. An immediate obvious hazard is that driving conditions rapidly become extremely difficult.
So far no pyroclastic flows have escaped from English's crater or from the Tar River valley although surges have entered the upper part of Whites Ghaut and scorched vegetation along almost the entire northern edge of the Tar River valley and the northeastern rim of English's crater to a depth of a few metres.There seems to be no immediate hazard of pyroclastic flows entering any of the ghauts which lead to inhabited areas but the situation must be kept under careful observation with immediate aerial inspection as soon as it is safe after each collapse episode. The immediate threat of pyroclastic flows to the western side of the island is also low and will remain so as long as the present pattern of collapse to the east continues. This conclusion would change if rapid growth continued without significant collapse to the east or if the region of growth moved to a different sector of the dome.
Hazards from mudflows in the western sector will continue to increase as long as ash continues to accumulate in the upper reaches of the western ghauts. These hazards relate as least as much to the weather as to the volcano but Fort Ghaut and Aymer's Ghaut are already extremely dangerous in the event of heavy or prolonged rain
Genuine magmatic explosions remain a remote probability although the probability increased slightly with the appearance of vesicular material in the products of the most recent eruptions. The hazard can be tracked by prompt and regular sampling of the products of each successive phase. At least one small phreatic explosion has been observed, and its products sampled, but the ballistic fragments from this explosion travelled no more than a few hundred meters from the source and they are not a significant hazard outside the areas already known to be highly dangerous.