Non-Linear Seismic Velocity Estimation from Multiple Waveform Functionals and Formal Assessment of Constraints.

摘要

Our goal is to develop optimal procedures for the use of multiple datasets. Due to the inherent variability, inconsistency, and peculiarities of disparate datasets and the well-known nonlinearity and non-uniqueness associated with geophysical modeling, such procedures must include methods for evaluating the performance and contribution of each dataset to the final results. We use quantitative assessment tools and a formal Bayesian approach to explore and evaluate each step of the modeling process, rather than to simply toss all constraints into a simultaneous fitting procedure to find the single solution that satisfies particular criteria. The procedure we propose is best characterized as velocity analysis via optimization; it is analogous to velocity analysis in exploration seismology, rather than inversion . It provides quantitative error measures of structural parameter estimates that can then be translated to earthquake location errors, and thus guide seismologists toward the most effective and efficient ways to improve model reliability.