Team Seismic, Montserrat Volcano Observatory
Team Seismic: W.B. Ambeh, W.P. Aspinall, W. Balgobin, V. Bass, O. Butler, L. Cabey, T. Christopher, J. Ewert, G. Ford, D. Galloway, R. Hoblitt, J.L. Latchman, A.B. Lockhart, R. Luckett, L.L. Lynch, A.D. Miller, E.J. Morris, L. Pollard, J.A. Power, C. Ramsingh, G. Ryan, D. Silcott, B. Simpson, R.C. Stewart, D. Seupersad, G. Thompson & R.A. White
The first visible signs of unrest during the present volcanic eruption at the Soufriere Hills volcano, in southern Montserrat, was a small phreatic steam and ash eruption on 18 July 1995. This followed a 3.5 year period of occasional earthquake swarms, which started in January 1992 (Ambeh and Lynch, 1996). One of the swarms, in 1994, was sufficiently intense to prompt the Seismic Research Unit (SRU) of the University of the West Indies to increase the number of seismographs in Montserrat from two to six. Earlier, notable, volcano-seismic crises had occurred in 1966-67, 1933-37 and 1897-98 (see Shepherd et al., B. Volcanol. (35), 1971).
After the initial outbreak, seismograph coverage was rapidly expanded by the addition of further instruments from SRU and from the USGS Volcano Disaster Assistance Program. A network with a minimum of eight short-period seismograph stations has been maintained in continuous operation throughout the eruption. At times, there has been need to relocate some stations and, during the explosive eruption of 17/18 September 1996, one was destroyed and its site abandoned. In October 1996, the network was supplemented by the addition of a new array of state-of-the-art broadband, wide dynamic range digital seismographs, intended for long-term seismic monitoring as part of a permanent volcano observatory in Montserrat (Stewart et al., this volume).
In common with experience at other volcanoes in eruption, a wide variety of seismic signals have been observed. At the MVO, these signals have been provisionally classified using a scheme which generally accords with categorisation schemes used elsewhere. The categories used are:
As is often the case, there are sometimes only small differences between one event type and another, so some attributions have been necessarily arbitrary and analyst-dependent. In consequence, routine counting of different event types has not been wholly consistent. Routinely processed hypocentre solutions include locations of some hybrid events, derived from the first arrival times.
By the end of October 1996, more than 50,000 individual seismic events had been detected and recorded, of which almost 7,000 have been processed for hypocentral locations. The majority of these have originated under the Soufriere Hills region of south Montserrat, at depths between 0-8 km, although a few others have been located further afield and at slightly greater depths. There has been significant temporal and spatial clustering of seismic energy release, with different types of event dominating at the different stages in the eruption. The initial, phreatic stage was accompanied by VTs, mainly as swarms, with some reaching felt intensity (e.g., the 12-13 August 1995 sequence of 38 felt earthquakes in one night under St. George's Hill). There is some suggestion of a general shallowing of hypocentres with time during the first few weeks of the crisis, but this requires careful reappraisal.
When extrusion of a lava dome into English's Crater started in September 1995, the numbers of VTs diminished considerably and intense swarms of hybrids developed, often associated with episodic changes in rate of dome growth and changes in the locus of growth within the crater. Hybrid swarms accompanied the arrival of the first incandescent magma at the surface in late November 95 and occurred again at the end of January, coinciding with an increase of the effusion rate. Thereafter, and throughout much of the first half of 1996, dome growth proceeded with only low, variable seismicity. However, there was a short outbreak of VT's near Windy Hill on 10-11 February, some felt, and there were also episodes of vigorous rockfall activity and pyroclastic flow generation interspersed in this period of dome accumulation, notably on 3-8 April, 12 May and 15 June. There then followed a lull in seismic activity of all kinds until a sudden burst of more than 500 small VTs took place between 20-21 July 96.
In the two months prior to the explosion of 17/18 September 1996, almost 2,500 volcano-tectonic earthquakes triggered the short-period recording system. Most of the locatable VTs were at shallow depths beneath the crater, and occurred in swarms. A period of intense seismic activity occurred between 20 July 1996 and 11 August 1996, with VT and hybrid earthquake swarms accompanied by episodic tremor (G. Thompson/Team Seismic, this volume). The rockfall activity reached a maximum on 11 August, and then generally declined until the beginning of September. During early September, there was possibly a tidal influence on the VT swarms, and many of the swarms were followed by an increase in rockfall activity (MVO Special Report #1). Prior to the 17/18 September explosion, the swarms became shorter but more frequent, and there was an increase in the number of rockfalls. The number of LPs and the amount of tremor remained low. Since the explosion, the level of seismic activity has been lower, and there have been few rockfall signals, principally because of the morphology of the October 1 dome. However, frequent VT swarms re-commenced on 21 October 1996, with the swarm of 1-2 November being the longest and most intense recorded since dome building began.