Montserrat Volcano Observatory, Montserrat, West Indies

Scientific Report 34
14 September 1996

This report marks the restarting of weekly scientific reports from MVO. The last weekly report (No. 31) is dated 3 July; the remainder of July will be covered by Scientific Report 32 and August by Scientific Report 33. This report covers the period from 1 to 14 September inclusive, and weekly reports will henceforth be dated on the Saturday of each week, covering the 7 days prior to 16:00 on the given date.


The early days of the reporting period included several periods of intense rockfall and pyroclastic flow generation from the eastern flank of the lava dome. These were caused by over steepening of the active flank and subsequent partial gravitational collapse of a segment of the dome. A scar was left on the eastern flank which had refilled by the end of the reporting period following continued rockfall activity but an apparent lowering of the growth rate of the dome. Seismicity was dominated by short swarms of shallow (< 2 km) volcano-tectonic earthquakes occurring on average twice per day over most of the reporting period.

Visual observations

Visibility was somewhat variable for the reporting period, with some good observations possible early in the month but low cloud covering the volcano for much of the remainder of the period. The passage of Hurricane Hortense through the northeastern Caribbean caused several days of strong winds and heavy rain.

Almost all of the activity on the dome has been on the eastern flank, although steaming has been occurring consistently from many areas on the dome. Indications on 1 and 2 September were that the scar left on the eastern flank after the mid-August collapse events had filled by spalling of new blocky lava, and additional new material was in an unstable state high on the western flank. These indications proved correct when a partial collapse occurred in two main pulses between 14:00 and 16:00 on 2 September and between 00:00 and 03:00 on 3 September. Both pulses included generation of pyroclastic flows which, certainly in the earlier pulse and probably in the later pulse, reached the sea at the mouth of the Tar River valley. Both of the pulses were preceded by two or more hours of increasing rockfall activity and comprised semi-continuous generation of pyroclastic flows. Visual observations of the earlier pulse showed that pyroclastic flows increased gradually in size and runout length, and the single flow to reach the sea was almost the last event. Flows were generally confined to the southern part of the valley although they sourced from north of Castle Peak. Significant erosion occurred in the middle part of the valley, with most deposition occurring in the lower part of the valley and on the pyroclastic flow delta built up since late-July at the mouth of the Tar River. Excavation of a deep (> 10 m) channel from the base of the new dome through the upper part of the talus fan enabled later flows to be more channelised and thus have greater run-out potential.

Initial rockfalls were accompanied by loud avalanching noise but pyroclastic flows themselves were absolutely silent. Most of the flows did not move particularly fast (probably < 50 km/h) although the last flow in the first pulse of the collapse was markedly faster and occurred as a single rather than compound flow. Both pulses produced ash clouds to at least 6 km in height by convection from the flows - there was no evidence for generation of vertical columns from the summit of the dome.

Sampling of the pyroclastic flows at the sea shore a few days after deposition showed that they contained less vesicular material than other flows since late-July, and were typical ash-rich, very poorly sorted deposits with juvenile lava blocks to at least 50 cm diameter. The deposits were too hot to do any detailed studies, but samples have been dispatched for geochemical and petrological analysis.

All pyroclastic flows during this collapse were confined to the Tar River valley, and ash clouds were distributed northwestwards, with 1 to 2 cm of ash deposited in the Cork Hill area, thinning but still more than 5 mm in the Old Towne area. Estimates of the volume of ash deposited during the partial dome collapse early in the month were made using ash mass and thickness from the network of ash collectors. Due to the size of the on-island deposition area, any estimates can only be very approximate, although account is taken of ash deposited out to sea by extrapolation assuming an exponential decay of thickness. The total mass of ash deposited on 2 and 3 September was c. 170 x 106 kg, which equates to a dense rock equivalent volume of c. 70 x 103 m3.

Visual observations for the remainder of the period were hampered by generally poor visibility. Rockfall and associated pyroclastic flow activity was confined almost exclusively to the eastern flank of the dome, with the scar left after 3 September gradually filling with talus spalling from steep new dome faces being extruded higher on the eastern flank. Rockfalls and pyroclastic flows continued to be generated at a higher than normal level for several days after the main collapse events. By the end of the reporting period, the scar was full and considerable unstable material looked liable for collapse from the upper eastern flank. Small rockfalls were noted on the northeastern flank and several small pyroclastic flows were generated to the south of Castle Peak. Ash generation has been at a moderate level after the major falls of 2 and 3 September; most has been deposited to the west of the volcano.

Several periods of heavy rain produced sediment-laden flash floods in Fort Ghaut during this period. These are recorded as high amplitude broadband tremor on the Gages seismometer and deposited about 60 cm of sediment in lower Fort Ghaut.


Rockfalls continue to dominate the seismicity although short swarms of volcano-tectonic earthquakes have been a feature of this reporting period. Long period and hybrid events remain at background levels and tremor has been generally low. Wind noise during the passage of Hurricane Hortense through the area dominated the seismic records on 8 and 9 September.

Earthquake types: 01 to 14 September 1996

These earthquake counts are made manually from the helicorder record for the Gages seismic station.

Date     VT        LP        Hybrid         RF        Tremor
01/09         107       4         21        92        Low
02/09         N/A       N/A       N/A       151+      N/A
03/09         110       1         25        52        N/A
04/09         4         2         3         84        Low
05/09         179       1         8         47        Low
06/09         74        1         39        49        Low to intermediate
07/09         46        0         25        49        Low
08/09         25        N/A       N/A       73        N/A
09/09         270       1         105       65        N/A
10/09         159       4         12        35        Low
11/09         152       12        32        86        Low
12/09         192       12        18        49        Low
13/09         32        12        25        129       Low
14/09         103       12        37        123       Low

Volcano-tectonic earthquakes during this period occurred exclusively in short swarms lasting between one and six hours. The earthquakes were all located at depths of less than 2 km (below sea level) beneath the crater. They are generally too small to be felt in currently occupied areas, although they have been felt in areas close to the crater by scientific teams, and some on 9 September were reportedly felt in the Richmond Hill area. On average, two swarms occur each day.

Seismic signals from rockfalls and pyroclastic flows continue to dominate the seismic records apart from during the VT swarms. Signals are best recorded on the eastern part of the network and comprise high frequency signals with a slow build-up to maximum amplitude and a slow decay to background. Duration of the signal is a good indication of the scale of the event, although the relationship between maximum amplitude and event duration is not always a simple one, with rockfalls containing large blocks producing higher amplitude signals than those only containing smaller blocks. The signal produced by the pyroclastic flows on 2 and 3 September saturated the record on most of the seismic network.

Background levels of long period and hybrid events have been noted throughout this period, and the level of tremor on the Gages seismometer has generally been low.

Ground Deformation

Few EDM measurements were possible during this period due to the generally low level of the cloud and due to ashing of various reflectors. The line from White's Yard to Castle Peak was measured on 11 September and showed a shortening of 15 cm since the previous occupation in late August. This c. 1 cm/day shortening trend is slightly higher than the trend established since mid-July but is not of immediate concern. The permanent loss of the Castle Peak reflector is expected in the next few weeks as rockfalls erode the eastern flank of Castle Peak in the area where the reflector is located; alternative locations for a reflector are being considered. The Galway's to Chance's Peak line was measured on 13 September and showed a continuation of the long-established trend of less than 1 mm/day shortening.

No GPS surveys were undertaken during this period, although the results from a survey late in August showed no measurable deformation of the volcano. The GPS survey does not utilise stations as close to the summit as the EDM network so that continued shortening on certain EDM lines is not inconsistent with the lack of movement over the GPS network.

Gas Measurements

No COSPEC measurements were undertaken during this period due to technical problems with the instrument. SO2 tubes continue to be used to monitor the concentrations of gas at ground level beneath the volcanic plume although results are not yet available for the period being reported here.

Additional measurements

Estimates of dome growth rates and deposit volumes continue using a variety of techniques. Fixed photographic and theodolite methods are becoming more automated and a new technique utilising kinematic GPS and range-finding binoculars from the helicopter is proving very effective. Ground-based surveys of the Tar River valley are also being undertaken to semi-quantitatively assess the areas of net deposition and erosion in the valley. A plan to utilise ship-borne sonar to map the offshore limits of the new delta at the mouth of the Tar River valley had to be aborted due to the coincidence of ship availability with the passage of Hurricane Hortense.

Staff changes
Nicki Stevens, Reading University/BGS
Richard Robertson, SRU Desmond
Supersud, SRU
Wilkie Balgobin, SRU
Dr Bill McGuire, University College London/BGS

Rob Watts, BGS
Dr William Ambeh, SRU

Montserrat Volcano Observatory