Montserrat Volcano Observatory, Montserrat, West Indies

Scientific Report 64
10 May 1997

This report covers the 14-day period to 24:00 on 10 May


Visual Observations

Visual observations have been hampered by poor weather conditions which have meant that the volcano has been in cloud for most of the reporting period. The period was characterised by a moderate level of rockfalls, but no major pyroclastic flows. Between 27 April and 2 May no views of the dome were possible. During brief clearances on 2 May, it was confirmed that dome growth continued in the same area, above the Galway's Wall on the southern side of the dome. There was no sign of recent pyroclastic flows to either the south or east of the crater, but some rockfalls and falling large blocks were observed.

On 3 May, the talus slope beneath the Galway's Wall had widened slightly, and new material was seen against the south crater wall to the east of the wall.

On 5 May, more observations of the dome were made in poor conditions. Material was spilling from the area of active growth to the south and east, with the eastern material forming pyroclastic flows as far as the base of Perches Mountain. There was no evidence of flows beyond the base of the talus slope in the White River. No detailed dome survey was possible, but it was estimated that the growth rate had decreased.

No more views were possible until 8 May, when most of the dome was observed from the Tar River Estate House. Rockfalls were heard from the south side of the Tar River valley, and several large, bouncing blocks descended towards Perches Mountain. No major changes were seen on the eastern face of the dome, although there were several fumaroles.


The seismicity has continued in the same pattern, with rockfall signals and long-period earthquakes. The level of activity was about the same as previous weeks.

The only significant departure from this was a swarm of volcano-tectonic earthquakes on 7/8 May. This swarm comprised 28 earthquakes that triggered the broadband network. Of these, 13 were located, with hypocentres beneath the crater at between 2.4 and 3.6 km depth BMSL. VT swarms have been rare since the start of dome growth, and all previous swarms have been accompanied by shallow hybrid earthquakes (e.g., the swarms of 2 November 1996 and 3 April 1997). While the swarm on 8 May was less intense than these swarms, it marks a departure in the nature of the seismic activity observed since the explosion in September 1996. During the swarm, the level of rockfall and long-period earthquake activity dropped, and rose again to previous levels when the swarm ended. This pattern of an anti-correlation between hybrid or VT swarms and rockfalls has been observed many times before at this volcano.

Table 1: Earthquake types

These earthquake counts are of events that triggered the broadband network's event recording system between 0000 and 2400 each day (local time).

Date		VT	Hybrid	LP	Dome RF	LPRF*
27 April 97	0	2	37	78	30
28 April 97	1	0	18	78	10
29 April 97	0	0	24	58	12
30 April 97	1	0	18	55	6
01 May 97	2	0	27	70	9
02 May 97	6	1	15	101	5
03 May 97	4	1	20	87	17
04 May 97	1	0	25	89	20
05 May 97	1	4	64	98	29
06 May 97	8	4	39	87	19
07 May 97	2	0	31	77	20
08 May 97	26	0	29	77	17
09 May 97	0	0	22	62	7
*LPRF: LP earthquake followed by rockfall signal. The LPs and rockfalls in an LPRF signal are also counted in their respective columns.

In the last few weeks, further analysis of recent hybrid and long-period earthquakes has been carried out. For the hybrid earthquakes, the initial onset is impulsive and followed by a complex high amplitude wave-train which apparently consists of several different phases. Frequency amplitude spectra from the same event recorded at different stations all have a narrow bandwidth with energy mainly in the 1-3Hz range. Higher frequencies are attenuated at the more distant stations. Distinctive spectral peaks are observed, and there are clear differences in the position of the peaks both between individual stations and between the three components at each station. Chouet et al. (JVGR 62, p95, 1994) show how the resonance of a fluid-filled crack can generate directional variation in spectral amplitude.

Horizontal plane particle motion plots have been used to determine the polarisation direction for a few hybrid earthquakes. The initial onset is linear and polarised in a direction radial to the crater, suggesting a P-wave arrival. However, this phase is followed after about half a cycle by a much higher-amplitude orthogonal phase, with prograde elliptical particle motion, which degenerates into a complex signal with further phase arrivals. Energy is greater on the horizontal components at all stations. These large amplitude secondary arrivals show some of the characteristics of Rayleigh waves.

Each station shows a large amplitude spectral peak on the transverse component, except Galway's Estate, where the dominant peak is on the radial component. Again the peaks are at different positions at different stations, which may suggest some kind of directional variation in the source radiation.

The spectra for radial and transverse components for a particular station are almost identical for earthquakes in the same swarm, and for earthquakes in different swarms. Spectral peaks show very similar alignment, and particle motion plots are in good agreement. The similarity between both the spectral characteristics and particle motion indicates a consistent source location and mechanism. However, some hybrid earthquakes have different amplitude spectra, and so the source is not always the same.

The apparent horizontal velocity is about 3.3 km/s for a selection of hybrid earthquakes. This is calculated from the difference in the onset of the P-wave arrivals at two stations. This fairly high velocity suggests a deeper source than for the long-period earthquakes.

For long-period earthquakes, the signal is quite monochromatic, with energy contained in modulated envelopes or packets similar to the eigen oscillations which can occur due to resonance. The LPs have fairly similar spectral bandwidths to the hybrid events, and the differences in the waveforms may result from much lower signal-to-noise ratios as the amplitudes are, in general, much lower. This means that it would be much more difficult to pick out the initial P-wave from the background noise. Measured apparent horizontal velocities of around 1.9 km/s for the events analysed suggests a shallower source depth than the hybrids. Particle motion analysis indicates that the long period wave-train consists of several different phases with energy much greater on the horizontal components than on the vertical.

Further analysis of the LP / Rockfall signals shows that the LP is a separate event which extends beyond the start of the rockfall. The spectral characteristics and particle motion of the long period signal are similar to those of the isolated long period events.

Our working hypothesis is that the LP earthquakes are caused by gas escape at shallow depth, within the dome or the upper part of the conduit. In a leaky system, the pressurised gas forces its way up to the surface along cracks and fractures. Hydraulic fracturing may increase the permeability and may also be a possible source of seismic energy.

The LP events probably trigger rockfalls when the ground acceleration exceeds a critical amount needed to dislodge unstable material on the surface of the dome. This threshold value would be dependent on a number of factors, not just the amplitude of the long period energy. For example a smaller amplitude LP could generate rockfall if the dome material were highly unstable.

Long-period energy might propagate as guided waves along low velocity channels with behaviour similar to surface waves. Amplitudes may be very high close to the channel, but decay rapidly as distance increases. So observed amplitudes at the stations may be much smaller than those on the dome. This initial analysis suggests that the signals are non-dispersive.

The LP events typically occur 10-20 s before the start of a rockfall. So far, no visible manifestation of the long-period signals has been seen, although due to cloud cover, visual observations of the source area of the rockfalls are difficult to make.

Ground Deformation

EDM measurements were made of the lines on the western side of the volcano running between Lower Amersham to Upper Amersham and Lower Amersham to Chance's Steps. The slant distance between the instrument site at Lower Amersham and the two target sites continues to show erratic jumps. It is possible to recognise a very slow shortening trend of the slant distance. There is also some very slow subsidence of the target sites relative to the instrument site. This may reflect relaxation due to removal of magma from depth below the volcano but will need further occupations to confirm.

There were two occupations of the northern triangle, which showed that the slow outward and downward movement of the Farrell's target seems to be continuing. Individual measurements are fairly erratic, but the trend can be identified in the long-term data.

There have been several occupations using the GPS equipment at sites on the crater wall close to the dome. A long occupation of around 20 hours was used in each case. Table 2 summarises the results

Table 2: Motion of GPS sites on the crater wall.

In each case the location of the rover site is calculated with reference to a fixed point at M18 Harris Lookout.

Date	Rover Site		Motion Since 18/1/97	Last Occupied	Distance to Dome (m)
28/4/97	Chance's Peak		77mm toward WSW		18/1/97		80
29/4/97	FT3 (High Farrell's)	183mm toward NW		06/3/97		300
30/4/97	Hermitage		36mm toward NNE		17/3/97		600
02/5/97	Perches Mountain	small movement to N	26/3/97		450

These data indicate that there is significant movement of the crater rim close to the lava dome complex. Movements are radially outward from the dome and diminish very quickly with distance from it - the wider field GPS networks have recorded very small movements since they were established in June 1996. EDM reflectors will be installed soon probably near the FT3 and Hermitage GPS sites.

The GPS network EASTNET (Harris, Whites, Long Ground, Windy Hill, Farrells) was occupied on 10 May. The significant result from repeated occupation of this network is the slow movement of the Farrells site towards the NNW. The baseline to Harris has shortened by 4cm since last June and the total displacement of the site is 5 cm.

Measurements of the cracks on Chance's Peak and on the eastern side of the Galway's Wall were made on 28 April and 3 May, respectively. The Chance's Peak crack continues to show both extension across the crack and dextral shear along the crack. There has been 6 millimetres of extension and 7 millimetres of dextral shear since the last measurement on 6 April. The rate of deformation has slowed significantly in the time between the two measurements. The crack on the eastern side of Galway's Wall showed the first movement since the measurements began in late March. A few millimetres of opening and 25 millimetres of sinistral shear had occurred on this crack. The implication from the movements on the Chances and Galways cracks is that the area between the cracks which contains the remains of Galways wall is moving slowly to the SSW away from the dome complex. Further measurements will be made to monitor the deformation taking place in these areas.

Dome Volume Measurements

No dome measurements were possible in this period, because of the cloud conditions.

Environmental Monitoring

The problems with the MiniCOSPEC have been sorted out by the manufacturer and the machine should arrive back at the MVO in the next few days.

Sulphur dioxide diffusion tubes were collected from five sites around the volcano on 4 May and were sent to the UK for analysis.

Routine weekly collection of ash is continuing at 12 sites west and north west of the volcano. The greatest thickness of ash collected during this period was 2mm at Upper Amersham on 4 May. Two bulk samples of ash, thought to have been relatively undisturbed since the start of the eruption, were collected from ash fall deposits at the American University of the Caribbean, Plymouth on 8 May. These bulk samples and a sample of ash collected during the Easter weekend activity have been sent for analysis. The aim is to determine the time-averaged cristobalite content and composition of the ash throughout the eruption.

Rainwater was collected from 4 sites on 29 April after a night of heavy rainfall. Further samples were collected from 3 sites on 11 May. The rainwater continues to be highly acidic and certain anions are present in high concentrations (Table 3).

Water samples were also collected from the Trials reservoir overflow tap and from a tap supplying a cattle trough at Amersham on 11 May. Analysis shows pH and anion concentrations to be well within WHO guideline levels.

Table 3: Rainwater geochemistry

Location		pH	Cond.	TDS	Flrd	Chlrd	Slpht
				mS/cm	g/l	mg/l	mg/l	mg/l
Weekes			5.21	0.223	0.111	1.20	54	8
Police Headquarters	3.26	0.622	0.311	1.25	*	*
Upper Amersham		2.60	1.902	0.953	0.60	*	*
Lower Amersham		2.90	1.166	0.584	>1.5	*	*

Location		pH	Cond.	TDS	Flrd	Chlrd	Slpht
				mS/cm	g/l	mg/l	mg/l	mg/l
Weekes			5.56	0.051	0.025	0.25	11.8	3
Police Headquarters	2.98	*	*	*	*	*
Upper Amersham		2.69	1.525	0.764	0.85	166	37
Trials overflow		7.78	0.776	0.388	0.4	103	38
Amersham cattle		7.74	0.284	0.141	0.2	38	nd

* insufficient quantities for analysis

Summary Of Recent Petrological Work On Montserrat Magma

Samples of dome rock collected by MVO staff, mainly from pyroclastic flow deposits, are routinely sent to two laboratories for petrological and geochemical investigation. These two labs, one at Bristol University in the UK and the other at Brown University in Providence, Rhode Island, USA, are undertaking a collaborative research programme aimed at providing information to MVO on the history of and recent changes in the magma reservoir which is driving the current eruption.

Whole rock major element composition has remained largely unchanged throughout the current eruption. Temporal changes in groundmass textures show a broad correlation with effusion rate in that groundmass crystallinity is lowest in samples erupted at high ascent rates. For example, glass content is higher (25-30%) in samples erupted between July and mid-September 1996 when effusion rates were high whereas samples from the early stages of the eruption when effusion rates were low have between 5-10% glass. However the correlation is not simple as groundmass crystallinity also depends on the time spent in the near-surface environment prior to deposition in pyroclastic flows. Samples erupted during January 1997 show a wide variety of groundmass textures and represent different batches of magma which spent varying times at the surface prior to deposition together in single pyroclastic flows.

The average An content of plagioclase microlites has increased over the course of the eruption. The cause of this is not fully understood but it may reflect greater degassing efficiency during ascent in the earlier part of the eruption as the An content of plagioclase depends on the water content of the melt. Maturation of the conduit system with time may inhibit degassing during ascent. Alternative possibilities are that the erupted magma has become hotter or more volatile rich over the course of the eruption perhaps due to tapping of deeper levels of the magma chamber.

Hornblende phenocrysts show a complex variety of breakdown rims, reflecting destabilisation due to both heating and decompression. Decompressional rims were thickest in the earliest-erupted magma and are completely absent from hornblendes erupted during 17 September explosion and the 19 December pumiceous pyroclastic flow.

Melt inclusions are hosted in both quartz and plagioclase phenocrysts which have measured H2O contents of ~ 4.5wt% (CO2 <50 ppm in quartz). Low S (<150 ppm) and Cl (2000-2500 ppm) are also measured in the quartz inclusions. The Cl contents of quartz melt inclusions are lower than those of melt inclusions trapped by plagioclase phenocrysts.

Water-saturated experimental phase equilibria have shown hornblende to become unstable in the Soufriere Hills magma at pressures <120 MPa and temperatures between 875-880 degrees centigrade. At the relatively low temperature conditions (prior to any reheating) where quartz might be stable pH2O are constrained to lie between 115 and 130 MPa; equivalent to chamber depths of approx. 4.5 to 5.5 km, although this may only be the upper part of the chamber.

There is strong evidence for the periodic influx of hotter, more mafic magma throughout the evolution of the Soufriere Hills volcano. The earthquake swarms that have occurred in the last 100 years may have been caused by influxes of mafic magma into a pre-existing magma chamber which has fed Soufriere Hills volcano eruptions for the last ~24,000 years.

MVO Scientific Staff (at 11 May)
Lloyd Lynch (Chief Scientist), SRU
Paul Jackson, SRU
Sayyaddul Arafin, SRU
Ouchi Osuji, SRU
Simon Young, BGS
Angus Miller (Deputy Chief Scientist), BGS
Richard Herd, BGS
Sue Loughlin, BGS
Anne-Marie Lejeune, BGS / Bristol University
Barry Voight, Penn State University
Matthew Watson, Cambridge University

Departures during this period: Gill Norton, BGS
Brian Baptie, BGS

Mr Lloyd Lynch will be Chief Scientist in post until 6 June.
The next CS will be Dr Willy Aspinall.

Montserrat Volcano Observatory