The 26 December (Boxing Day) 1997 sector collapse and debris avalanche at Soufriere Hills Volcano, Montserrat

Abstract

The southern sector of Soufriere Hills Volcano failed on 26 December 1997 (Boxing Day), after a year of disturbance culminating in a devastating eruptive episode. Sector collapse produced a c. 50 x l0 6 m 3 volcanic debris avalanche, and depressurized the interior of the lava dome, which exploded to generate a violent pyroclaslic density current. The south-directed growth of a lava lobe and build-up of lava-block talus, since early November 1997. brought the hydrothermally weakened sector to a condition of marginal stability. Limit-equilibrium stability analyses and finite-difference stress-deformation analyses, constrained by geomechanical testing of edifice and debris avalanche materials, suggest that the sector collapse was triggered by a pulse of co-seismic exogenous lava shear-lobe emplacement. Slip-surface localization was influenced by strain-weakening. The source region fragmented into avalanche megablocks, and further disruption generated a chaotic avalanche mixture that included variably indurated and coloured hydrothermally altered material, and much talus. The avalanche consisted of several flow pulses that reflected complexities of source disruption and channel topography. In the proximal zone, within 1.5 km from source, many megablocks preserve pre-collapse stratigraphy. At major bends the avalanche separated into channelled and overspill flows. In the distal region, >2.5 km from source, stacked sets of the main lithologies occur with a hummocky surface and abrupt flowage snouts, beyond which sparse hummocks occur in a thinly spread deposit. Textures suggest emplacement by laminar mass transport of partly saturated debris riding on a frictionally sheared base. Three-dimensional numerical simulations of emplacement governed by a Coulomb-type (Pouliquen) basal friction law imply low values of friction (< 15 o ), consistent with geotechnical test data and the localized presence of pore-water pressures. The best-fit model suggests an emplacement time <3 minutes and a typical maximum velocity of about 40 ms -1 , which are consistent with field estimates.