Poststack stochastic seismic inversion method based on gaussian-markov prior constraints

Abstract

The application discloses a poststack stochastic seismic inversion method based on Gaussian-Markov prior constraint. The method can comprise the following steps: setting an inversion parameter, inputting a wavelet, and establishing a low-frequency model for inversion; determining a to-be-inverted trace in a three-dimensional work area; calculating a conditional mean and a conditional variance according to a kth sampling point in the to-be-inverted trace and a Gaussian-Markov prior distribution; constructing a candidate wave impedance vector of the to-be-inverted trace according to a Gaussian distribution determined by the conditional mean and the conditional variance; performing probability transition on the candidate wave impedance vector of the to-be-inverted trace to realize the update of the wave impedance; and updating the wave impedance of all sampling points of all traces in the three-dimensional work area. The application improves the vertical resolution of the wave impedance inversion result by extending the Metropolise-Hastings algorithm, further introduces a Gaussian-Markov field to describe the spatial correlation of the three-dimensional wave impedance, and improves the lateral continuity of the inversion result.

Images