B. Derroux1, C. Harford2, A.-M. Lejeune2, J.B. Shepherd3, G. Skerritt1, M.V. Stasiuk1, N.F. Stevens1, J. Toothill3, R. Watts5, S.R. Young5
1Montserrat Volcano Observatory, Montserrat, West Indies
2Department of Geology, University of Bristol, Bristol
3E.B.S., University of Lancaster, Lancaster
4ESSC, University of Reading, Reading
5British Geological Survey, Edinburgh
The period of activity late July through mid-September 1996 was characterized by relatively high extrusion rate and cyclical, rapid dome growth followed by catastrophic collapse of the dome flank into the Tar River valley and the sea. The final collapse on 17 September triggered a vulcanian explosion followed by a vertical eruption column. The repetitive behaviour and data on the dome thickness during this time suggest that simple, nonexplosive collapse of the dome is likely when a thickness of lava not significantly greater than about 130 m is involved, whereas collapse of thicknesses greater than about 150 m can generate decompression-related explosions. This result may provide a simple gauge for predicting explosive events.
On 20-21 July 1996 the lava dome on Montserrat showed an important change in its rate and locus of growth, accelerating from 1-2 m3s-1 to nearly 10 m3s-1, accompanied by renewed extrusion on the northeast side of the dome. From that point until 17 September, the dome continued to extrude rapidly to the northeast, with dome expansion dramatically punctuated by wholesale gravitational collapse of the dome flank every 10-15 days: major collapses occurred on 28-31 July, 11 August, 21 August, 2 September and 17 September, prior to the explosive event of 17/18 September. Each collapse was heralded by escalating pyroclastic flow activity. This corresponded to ever larger quantities of material slumping from the dome flank to produce increasingly frequent block and ash flows of increasing runout distances. In the 12-24 hours preceding a major collapse, pulsing, "quasi-continuous" pyroclastic flows with runouts of 400-1000 m carried material from the dome into the upper Tar River valley. Major collapses produced block and ash flows and accompanying surge clouds which swept down the Tar River valley in a swath nearly 1000 m wide, where the blocky deposits were strictly confined to the narrow axes of the gullies. The largest flows reached the sea, where they were halted, and built up a pyroclastic apron. The succession of collapses removed almost all vegetation in the valley, as well as stripping much of the topsoil and significant bedrock in locations close to the gully axes. None of the collapses showed any evidence of an explosive component, either in the form of a propelling lateral blast or a convective column. During periods of growth, the active flank of the dome was observed to be the site of vigorous degassing in the form of distributed gas flow as well as focused, jet-like, ash-bearing fumaroles. Each major collapse excavated a distinct landslip-like scar or canyon in the northeast dome flank, directly adjacent to Castle Peak. In each case, this canyon was nearly refilled with dark, scoriaceous, slabby lava within about 5 days of the previous collapse. During such periods of initial regrowth the fresh extrusion changed in appearance on an almost hourly basis, as new slab-like spines up to 10 m high appeared and collapsed away. Once close to being refilled, the fresh material bulged outward but changed more slowly in appearance, and at this point rockfalls and block and ash flows restarted. The collapse of 17 September appeared similar to those previous, but triggered explosive activity.
This period was unfortunately accompanied by frequent cloud cover. Under such conditions the photogrammetric methods of dome volume measurement are less workable and so a new method was developed to supplement the photo technique: helicopter-based binocular surveys. This technique involved mounting the Rover antenna of the Leica GPS equipment on the helicopter with the Base antenna at a known location, and perform an airborne kinematic survey. When processed using only the code of the satellite signal, this yields the location of the helicopter every second to within 1 m in all dimensions. During a survey, the helicopter hovers within 1 km of the dome or deposit surface while Leica Vector Infrared Laser Rangefinding binoculars are used to determine the distance, inclination and azimuth from the helicopter to points on the surface. The combination of GPS and binocular data allows calculation of coordinates on the surface and hence reconstruction of deposit topography. Surveys were carried out on the pyroclastic deposits in the valley on 4,11,14,17,25 August and 6 September to determine the volume of material slumped from the dome, and on the dome itself on 9,12,16,17,25 August and 4,15 September. Visibility directly before and after a major collapse was typically at its worst due to excessive suspended ash, although one exception to this was on 12 August, the day after a collapse.
Dome survey results allow determination of the thickness and volume of the deposits and how they changed during the period of vigorous collapse activity, as well as the eruption rate in this period. The maximum measured volume of the dome prior to each collapse increased with time from 29.1 to 32.5x106m3; the volume of material shed in each collapse varied from 2 to 5x106m3. The rate of dome growth in the period 12-17 August was determined to be 10.6 m3s-1, and from 2-15 September was 2.9 m3s-1. The general trend of decreasing eruption rate is consistent with the lower values of October.
The survey data on the dome show that during the July-September period it developed a strongly two-lobed form. It's maximum thickness of 200-220 m occurs along a ridge close to the southwest side of English's Crater and is unchanged from mid-July. The lobes emanate from the thick ridge and part around Grouch's Hump. The major collapse events excavated the lobe on the south side of Grouch's Hump. Each collapse removed much of this lobe, eroded bedrock in the valley, and also probably scoured dome substrate below the collapsing lava. Data from 12 August, one day after a major collapse, clearly define a collapse scar carved through a maximum of about 130 m of dome lava and reveal that up to 80 m of Grouch's Hump had been removed by this time. The erosion of the substrate was caused by the collapse of 11 August or a combination of that collapse and previous ones. Re-growth of the dome occurred by extrusion into this new low, flat area. Thereafter, the thickness of the active lobe of the dome over the eroded area, measured a few days prior to collapses, was close to that on 9 Aug (about 130 m). Visual observations directly after collapses showed that this was the location of main collapse. On 15 September, however, the dome reached its greatest volume and the extruding lobe its greatest thickness, slightly more than 155 m in the area of previous collapses and directly over the eroded Grouch's Hump. This increased thickness may have been a result of reduced eruption rate, allowing increased bulk viscosity of the lava. On 17 September the thickened lobe collapsed and triggered a significant explosive eruption. It is not certain how far into the dome the 17 September collapse excavated prior to the vulcanian explosion, but it seems likely from the previous cyclical behaviour that it removed essentially the same unstable lobe of lava. The fact that previous collapses were produced from thinner lobes which did not lead to explosions is consistent with the idea that the initial (vulcanian) explosive event was generated by rapid decompression of relatively volatile-rich, but still highly degassed, magma. This suggests that a critical thickness of collapsing lava for generating explosive behaviour for the Montserrat dome may be between 130 and 155 m. The difference in pressure in the two cases corresponds to a difference of only about 0.2 wt% dissolved water in the melt and implies a remarkable sensitivity of the behaviour to water content.