Geological
Patterns and Rock Physics for a New Seismic Inversion Method
Gonzalez, Ezequiel
F.1, Tapan Mukerji2,
We present a novel inversion technique
that constructs solutions based on expected geological patterns in a study area
and honors local rock physics transformations. The inversion is posed as an
inference problem, with prior geological and rock physics knowledge,
represented as probability densities (pdfs), being
updated by seismic data to obtain solutions. Using probabilities is a way to
accept and account for imperfect and limited data as well as imperfect models.
The technique can be extended to invert different types of geophysical data.
Solutions from this inversion are realizations of predefined groups with common
lithology, fluids, or other reservoir property of
interest. Thus, e.g., shales, oil sands, and brine
sands might constitute three different groups in a particular study. Rock
physics principles help to establish the link between multivariate pdfs of elastic properties and the reservoir
characteristics defining the groups. Multiple-point geostatistics
(MPG) is used to define and explore the space of geological patterns. The
pattern-based MPG algorithm (a modified version of SIMPAT, i.e. sequential
simulation with patterns) first extracts geologically consistent spatial
patterns of groups from a geological model (“training image”). Then, using
sequential simulation principles, candidate local solutions are proposed that
are accepted only if, on applying the forward modeling operator, they generate
synthetic data reproducing the observed seismic data within acceptable
tolerances. The concept and implementation of the inversion were validated
using a synthetic reservoir (Stanford VI), consisting of a system of fluvial
channels. Moreover, applicability to real problems was demonstrated by
inverting a field seismic section. We believe that this technique lays the
foundation for innovative inversions of geophysical data tied to geological
patterns.
AAPG Search and Discover Article #90063©2007 AAPG Annual Convention, Long Beach, California