--> Abstract: Dynamic Time Warping: a Flexible Framework for Stochastic Stratigraphic Correlation, by F. Lallier, G. Caumon, J. Borgomano, S. Viseur, and C. Antoine; #90090 (2009).

Datapages, Inc.Print this page

Dynamic Time Warping: a Flexible Framework for Stochastic Stratigraphic Correlation

Lallier, Florent 1; Caumon, Guillaume 1; Borgomano, Jean 2; Viseur, Sophie 2; Antoine, Christophe 3
1 CRPG, Nancy-Universite, Vandoeuvre les Nancy, France.
2 Laboratoire de Géologie des Systèmes et des Réservoirs Carbonatés, Université de Provence, Marseille, France.
3 ASGA / Gocad Consortium, Nancy, France.

The stratigraphic correlation of well cores and logs is a fundamental step in reservoir studies, since it is the basis for the static and dynamic model and seismic processing, and is yet subject to many uncertainties.

Several authors have discussed about the approach to achieve well correlation, but the proposed methods mainly lead to a unique correlation path. Therefore, uncertainties due to stratigraphic well correlations cannot be handled and integrated in existing subsurface uncertainty assessment workflows.

We propose a new method of stochastic stratigraphic well correlations: the aim is to generate automatically several possible sets of well correlations accounting for interpretations made along well paths. This method is based on the Dynamic Time Warping (DTW) algorithm, used in speech recognition and bioinformatics. In practice, the probability of correlation of each stratigraphic marker and interval identified along the well path is computed using: (1) information available on the interpreted well (e.g. paleobathymetry, lithology); (2) some reference lines coming from clear, first-order stratigraphic boundaries; (3) sedimentological scenario about the studied area; (4) sedimentological concepts deemed applicable. Then, using these probabilities, several possible sets of well correlations are stochastically generated with a modified version of the DTW algorithm.

Moreover, the method acts hierarchically. For example, low level stratigraphic events or high confidence data (e.g. biostratigraphic boundary, regional MFS) will be correlated first, and will constrain the correlation of higher-order events.

This method is applied to the sequence-stratigraphic well correlation in a carbonate ramp system. Correlation probabilities are obtained by: (1) evaluating the consistency of the paleo-angle formed by correlated couple of well markers and the paleo-horizontal; (2) comparing depositional environment trends interpreted along each well interval (e.g. more distal or proximal).

 

AAPG Search and Discovery Article #90090©2009 AAPG Annual Convention and Exhibition, Denver, Colorado, June 7-10, 2009