S045-0004
Force and moment tensor uncertainty quantification with 1D and 3D Earth models

Monday, 14 December 2020
Poster
Ryan Modrak, Los Alamos National Laboratory, Los Alamos, NM, United States, Carl Tape, University of Alaska Fairbanks, Geophysical Institute, Fairbanks, AK, United States, Celso R Alvizuri, University of Alaska Fairbanks, Fairbanks, AK, United States, Vipul Silwal, University of Alaska Fairbanks, Geophysical Institute and Department of Geosciences, Fairbanks, AK, United States, Julien Thurin, ISTerre Institute of Earth Sciences, Saint Martin d'Hères, France and Michael Cleveland, Los Alamos National Laboratory, EES-17, Los Alamos, NM, United States
Abstract:
Point forces and moment tensors provide generality to describe indigenous and non-indigenous seismic sources with or without net momentum transfer. Underground explosions with net mass movement or nonlinear surface interaction, as well as complex surface processes such as landslides, may benefit from such joint force and moment tensor descriptions. Choosing simple, low-dimensional spatial and temporal representations and quantifying tradeoff between parameters is a first step toward joint modeling. For robust uncertainty quantification, we present uniform parameterizations for point forces and moment tensors that are bounded in all but one parameter. Searching over these bounded spaces is readily feasible with efficient parallel misfit evaluation codes (e.g., github.com/uafgeotools/mtuq) and avoids complex random walk implementation details. We will apply such uncertainty quantification methods to landslides and declared underground nuclear tests using both one- and three-dimensional Earth models. Preliminary findings suggest that 3D models can reduce misfit multimodality, with striking effects on credibility estimates.