The ongoing eruption of the Soufrière Hills Volcano (SHV) on Montserrat provides an unprecedented opportunity to investigate complex magmatic processes at an andesitic volcano. Growth of the lava dome has been unsteady and accompanied by cyclic patterns of ground deformation, seismicity, and explosive eruptions. The cycles include a short-term scale (6-18 hour), a meso-term scale (~7 weeks), and a long-term scale (~30 y) (Voight et al., 1998,1999). They provide insights into eruption dynamics at andesite volcanoes, with the short-term cycles suggesting that degassing, rheological stiffening of the magma, and pressurization in the upper conduits are coupled and control many of the geophysical and dynamical phenomena observed (Voight et al., 1999; Melnik and Sparks, 1999). The meso- and long-term cycles, we propose, reflect deep-seated processes involving the magma reservoir (Mattioli et al., 2000; Voight and Elsworth, in prep.).

We propose to investigate the dynamics of the full system using an integrated array of specialized instruments in 4-6 strategically located 200-m boreholes, and several shallower holes, surrounding SHV. The system is active and dynamic and will remain so for the foreseeable future. Analysis of the continuous perishable data provided by these instruments should provide important new insights and specific constraints to theoretical models, involving the dynamic behavior of the andesite magmatic system. The instrument package for each borehole, designed to have long life at high temperatures, will include a downhole dilatometer and a seismometer, with a surface CGPS station. The borehole observatory will be fully integrated into the surface monitoring network operated by the Montserrat Volcano Observatory (MVO) and will be used to track processes occurring in the active magma reservoir and its associated conduit systems. We will employ a tested version of the Sacks-Evertson borehole strainmeter that will operate with long life (decades) under conditions of high temperature. Dilatational strains from relatively small changes in the reservoir system will be detectable. Recordings of earthquakes by the borehole sensors will provide very broadband data (~2 Hz to 1 kHz) for study of the seismic sources. Passive seismometers at the depths proposed will have very low noise levels and enable detection of very small earthquakes. These instruments will allow probing the changes in the andesitic volcanic system and underlying mafic sources with unprecedented sensitivity. Additional holes will house very-high-gain tiltmeters. Previous work at other volcanoes has shown that such high sensitivity information is essential to our understanding of the physics of magma reservoirs, but little work has been done on andesite stratovolcanoes and the results of our work should be valuable as a model for other sites. Obvious is the societal value of high sensitivity measurements with potential for early warning of unrest.

This is a collaborative protect involving PIs from Penn State University, Carnegie Institution of Washington (CIW), University of Arkansas and Duke University, in collaboration with Montserrat Volcano Observatory (MVO) and our UK colleagues from Bristol and Leeds. An integrated work plan subdivides the project into equipment fabrication and installation, data acquisition, numerical modeling analysis and communication of results. Project Direction is the responsibility of B. Voight (PSU). Simon Young, former Director of MVO, will work as scientific and logistical liaison in coordinating drilling and installations. The downhole installations will be supervised by I. Selwyn Sacks and A.T. Linde (CIW), with contributions by P.E. Malin (Duke). Operation of the CGPS network will be guided by G.S. Mattioli (UPRM). Strainmeter and GPS data analysis emphasizing inverse elastic modeling will be developed by Mattioli at UA with a grad student, and numerical models will be developed by a postdoc working with D. Elsworth and Voight at PSU. The seismic data will be integrated into the MVO database, and various aspects will be analyzed by P. Malin (Duke), and others. All scientists will closely collaborate in modelling.