METHOD AND SYSTEM FOR DETERMINING AN ESTIMATED MODEL PARAMETER FROM A MEASURED FIELD

摘要

The present disclosure is directed at a method and system for determining an estimated model parameter from a measured field. The measured field may be generated using, for example, multiple sources located through a domain of interest, such as a certain volume of land, and measured using, for example, multiple receivers also located within the domain of interest. The method includes casting an inverse problem relating the measured field to the model parameter as a stochastic optimization problem, in which the stochastic optimization problem includes minimizing an objective function having a misfit expressed as an expected value of a function that depends on a random variable; iteratively optimizing the objective function using a stochastic optimization technique; determining when the objective function is reduced to a desired value; and when the objective function is reduced to the desired value, identifying a model parameter that corresponds to the objective function at the desired value as the estimated model parameter. Exemplary model parameters that can be estimated include conductivity and seismic velocity. Beneficially, using stochastic as opposed to deterministic techniques can result in savings in computational resources in the forms of FLOPS and memory.