HR: 14:40h
AN: V52E-05    [PDF]
TI: Inference of Magma Chamber Geometry at Sierra Negra Volcano, Galapagos Islands, Using a 3D Boundary Element Method With Pressure Boundary Conditions and InSAR Observations
AU: * Yun, S
EM: shyun@stanford.edu
AF: Geophysics Department, Stanford University, Stanford, CA 94305-2215 United States
AU: Segall, P
EM: segall@pangea.stanford.edu
AF: Geophysics Department, Stanford University, Stanford, CA 94305-2215 United States
AU: Zebker, H A
EM: zebker@stanford.edu
AF: Geophysics Department, Stanford University, Stanford, CA 94305-2215 United States
AB: We present a new way of implementing a 3D boundary element method with uniform pressure boundary conditions. Uniform opening dislocations have often been used to infer surface deformation due to subsurface processes such as slip on fault planes or magmatic processes under volcanoes. These models require displacement discontinuities as input parameters and quantify the displacement and stress fields at observation points in the continuum. We use the linear relationship between displacement discontinuity and stress to derive a boundary element method subject to uniform pressure boundary conditions. Application of this method to synthetic aperture radar interferometry (InSAR) data over Sierra Negra shows agreement with previous results in terms of maximum sill opening. Amelung and Jonsson et al. [2000, Nature] successfully used a sill model with spatially varying opening boundary conditions. We have improved the model by applying the physical constraint of uniform pressure on the wall of the sill. Given the model geometry, in particular, the periphery of the sill, the pressure boundary condition determines the opening at each model element. Thus the number of degrees of freedom in the problem is reduced by the number of model elements. Determining which elements open, and are thus subject to the pressure boundary condition, is a non-linear problem. We adopt a Bayesian Monte Carlo approach to estimate the sill geometry. The probability that an element either opens or remains closed is assumed to be a Gaussian random variable centered on the peak of the radar range change measured in our interferogram. Elements beneath the largest range change are most likely to be open, those far away least likely. The Monte Carlo estimation is run for an arbitrary time, and the best fitting model is chosen.
DE: 1243 Space geodetic surveys
DE: 3210 Modeling
DE: 3260 Inverse theory
DE: 6924 Interferometry
DE: 8434 Magma migration
SC: Volcanology, Geochemistry, Petrology [V]
MN: 2003 Fall Meeting