Monte Carlo method for estimating lithology from seismic data

Claim 1

A method for modeling a continuous profile of the petrophysical properties of subsurface earth layers, comprising: establishing the presence of, and measuring the petrophysical properties of a plurality of geologic rock classes in an earth layer, said rock classes including discrete ranges of petrophysical properties; assigning one of at least two possible colors to each said rock class; measuring the spatial transition statistics between said rock classes in said earth layer between a plurality of spaced-apart geological control points; providing a continuous array of seismic stations between said plurality of spaced-apart geological control points, and recording seismic attributes that are indicative of the petrophysical properties of said rock classes; calculating the means and variances of said seismic attributes for each said rock class; creating an initial model of the substrate earth layers consisting of an N-dimensional array of pixels and randomly assigning one of at least two possible colors to each said pixel of said initial model; choosing a pixel at random from said array; for each possible color of said pixel, calculating the likelihood energy and the prior energy; summing the likelihood energy and the prior energy to evaluate the global energy; selecting the one of said at least two possible pixel colors that minimizes the global energy; repeating the steps of choosing, calculating, summing and selecting for every other pixel in the array that forms the initial model thereby completing a first global iteration cycle; executing additional global iteration cycles by perturbing the pixel coloring over all pixels until the global energy converges to a minimum to provide a final model; and displaying said model.

Claim 2

A method for modeling a continuous profile of the petrophysical properties of subsurface earth layers, comprising: establishing the presence of, and measuring the petrophysical properties of a plurality of geologic rock classes in an earth layer, said rock classes including discrete ranges of petrophysical properties; assigning one of at least two possible colors to each said rock class; measuring the spatial transition statistics between said rock classes in said earth layer between a plurality of spaced-apart geological control points; providing a continuous array of seismic stations between said plurality of spaced-apart geological control points, and recording seismic attributes that are indicative of the petrophysical properties of said rock classes; calculating the means and variances of said seismic attributes for each said rock class; creating an initial model of the substrate earth layers consisting of an N-dimensional array of pixels and randomly assigning one of at least two possible colors to each said pixel of said initial model; choosing a pixel at random from said array; for each possible color of said pixel, calculating the likelihood energy and the prior energy; summing the likelihood energy and the prior energy to evaluate the global energy; selecting the one of said at least two possible pixel colors that minimizes the global energy; repeating the steps of choosing, calculating, summing and selecting for every other pixel in the array that forms the initial model thereby completing a first global iteration cycle; executing additional global iteration cycles by perturbing the pixel coloring over all pixels until the global energy converges to a minimum to provide a final model; and displaying said model.