Method for characterizing subsurface petrophysical properties using linear shape attributes

A method for predicting subsurface petrophysical properties in a subsurface formation from a set of seismic data trace segments for said subsurface formation, said method comprising the steps of:

• (a) obtaining well log information for at least one well in said subsurface formation;
• (b) using said well log information to determine said subsurface petrophysical properties in the vicinity of each said well;
• (c) determining a set of independent waveform shapes which characterize said set of seismic data trace segments;
• (d) selecting a subset of said seismic data trace segments for use in calibrating said set of independent waveform shapes;
• (e) determining a set of weights representing the amount of each of said independent waveform shapes in each of said selected seismic data trace segments;
• (f) determining a relationship between said weights and said subsurface petrophysical properties in the vicinity of each said well;
• and
• (g) using said relationship to predict said subsurface petrophysical properties for at least one other location in said subsurface formation.

A method for determining whether a pseudo-well synthetic seismic trace segment is statistically consistent with a set of seismic data trace segments for a subsurface formation, said pseudo-well synthetic seismic trace segment being derived from well log data obtained from at least one well in said subsurface formation, said method comprising the steps of:

• (a) determining a set of independent waveform shapes which characterize said set of seismic data trace segments;
• (b) determining a first set of weights representing the amount of each of said independent waveform shapes in said pseudo-well synthetic seismic trace segment;
• and
• (c) using said first set of weights to determine whether said pseudo-well synthetic seismic trace segment is statistically consistent with said set of seismic data trace segments.

A method for determining a set of independent geometric patterns which capture substantially all of the geometric patterns in a set of seismic data trace segments corresponding to a geologic zone of interest, said method comprising the steps of:

• (a) datuming said seismic data trace segments to horizontally align said seismic data trace segments;
• (b) determining the maximum length T of any of said seismic data trace segments;
• (c) augmenting each seismic data trace segment having a length less than T by adding a sufficient number of data samples so that the augmented length equals T;
• (d) selecting a plurality of reference trace segments at various locations in said set of seismic data trace segments;
• (e) concatenating each said reference trace segment with a plurality of other seismic data trace segments in the vicinity of said reference trace segment to form a plurality of supertrace segments;
• (f) defining a matrix A.sup.T of said supertrace segments, where said supertrace segments form the columns of said matrix A.sup.T,
• (g) computing the average value of each row of matrix A.sup.T to obtain an average supertrace segment;
• (h) subtracting said average supertrace segment from each column in matrix A.sup.T to obtain matrix A.sup.T ;
• and
• (i) solving the eigenvalue-eigenvector problem ›A.sup.T A!V=.LAMBDA. for V, where matrix A is the transpose of matrix A.sup.T, V is a matrix of the independent geometric patterns in said set of seismic data traces, and A is the singular value matrix for the eigenvalue-eigenvector problem.