P.D.Cole, R.S.J. Sparks, R. Robertson, N.F. Stevens, S.R. Young, G.E. Norton, C. Harford
Montserrat Volcano Observatory, Old Towne, Montserrat, West Indies
The explosive event of September 17th was the first of its kind to occur during the activity at Soufriere Hills Volcano, Montserrat. Explosive activity began after a sustained period of 8.5 hours of pyroclastic flows followed by a pause of ~4 hours. The explosive phase lasted for 46 minutes and distributed coarse tephra products over the southern part of the island. The tephra plume was encountered by an aircraft at a height of 37,000 ft (11.3 km) which gives a minimum estimate for the eruption column height. Studies of lithologies and maximum clast sizes of the coarse-grained tephra deposit formed by the explosive event have yielded a wealth of data regarding the causes and mechanisms involved in this eruption. Isopleth maps of maximum clast sizes of both pumice (87 points) and lithic fragments (61 points) have been constructed and indicate a maximum column height of ~14 km, although slightly lower heights of 10-11 km may have occurred for a large part of the event. Clast dispersal is complex with coarser material being dispersed south-west and smaller clasts sizes to the north-east. These dispersal patterns may be related to the interaction of the tephra with both trade and anti-trade winds. Variable local winds may also have influenced the dispersal of tephra.
Maximum pumice and lithic sizes are not in aerodynamic equilibrium as MP/ML ratios are significantly smaller than expected. Indeed lithic clasts are about 30% larger than required for aerodynamic equilibrium. These data indicate that the lithic clasts may have been the result of an initial vulcanian blast at the onset of the eruption whereas the pumice is derived from the subplinian eruption column generated by a sustained eruption from gas-rich magma within the conduit.
As well as the widespread tephra fall deposit, ballistic clasts were identified preferentially dispersed to the east and blocks up to 1.2 m diameter reached Long Ground 1.8 km from the source. The ballistic deposit is composed of a number of lithologies including dense glassy lavas, tuffisites and moderate density pumice. The ballistic component shows a bimodal density distribution of either pumice (mean density 1600 kg/m3) or glassy component (mean density 2600 kg/m3). The pyroclastic flow deposits of the delta contain predominantly partially vesicular clasts (mean density 2300 kg/m3) considered to be dome material. Such material is notably absent within the blast deposit indicating that it was derived from the conduit system beneath the dome. The pumice of the ballistics are often strongly foliated with flattened vesicles lined with shattered fragments of hornblende. Such a texture is restricted to the ballistic blast deposit and is interpreted as a result of expansion and stretching at high gas pore pressures in the conduit or deep in the core of the dome. This expansion may have occurred during the unloading of the dome by gravitational collapse and the formation of pyroclastic flows. Estimates using the model of Fagents and Wilson (1993) indicate launch velocities of ~180m/s.
The ballistic components also provide important information on the source pressure for the initial vulcanian explosion. Lower limits on the gas content are given by launch velocities of 180 m/s which require 2wt% H2O. Upper limits are indicated by inclusion studies which give results of 4% H2O. The tensile strength of dome rock is between 15 and 25 MPa and a lower limit of 20 MPa just prior to failure is assumed here. Estimated volumes of pyroclastic flow deposits for the activity in the Tar River Valley (~2x106m3), the addition to the delta (~6x106m3) and the associated pyroclastic flow ash cloud deposits formed on the 17th (~1x106m3) are in good agreement with estimations of the volume of the scar in the pre-Sept 17th dome of ~9x106m3. This implies that most of the volume of the scar was excavated prior to the onset of explosive activity. Assuming a conduit diameter of <50 m and discharge rates of 1,300 m3/s (considered to be conservative) then the explosive phase would have excavated magma to a depth of ~ 1.8 km equivalent to a volume of 3x106m3. This is consistent with the average depth of VT earthquake hypocentres in the two months prior to the eruption.