Lithocube Analyses with Uncertainty using Rock Physics, High Resolution Seismic and Bayesian Classification Scheme: Applications to Clastic Basins

A successful seismic based lithology prediction effort depends on reliable rock physics<br>transformation from seismic attribute space into the lithology space. This problem is non-unique and<br>thus uncertainty associated with the different predicted lithologies and fluids (i.e. confidence interval<br>and estimate of misclassification probability) should be addressed. We address the problem using rock<br>physics relationships in conjunction with well known statistical tools and Bayesian Classification<br>technique to characterize various lithofacies and fluids in subsurface. Our final output is most likely<br>lithology unit, with most likely fluids and the probability associated with it. The procedure is<br>illustrated with examples from clastics basins.