Non-Linear Inversion of Seismic Data by Successive Approximation of ModelParameter Probability Distribution Functions

摘要

Inversion of seismic data from a 1-D elastic earth is accomplished by choosingmodels randomly from a distribution which is the reciprocal of the objective function. Minima in the objective function correspond to peaks in distribution, thereby enhancing the probability that more models will be chosen from this area. Results show good recovery of dominant model parameters values and their uncertainties.