Thin-bed reservoir characterization using integrated three-dimensional seismic and well log data: A case study of the central Boonsville Field, Fort Worth basin, north-central Texas

This dissertation is designed to resolve two problems using seismic attributes to: (1) delineate thin-bed reservoirs, and (2) distinguish thin-bed sandstone reservoirs from thin-bed non-reservoir carbonates where they appear similar at seismic scale. I evaluated some widely used techniques, developed new approaches for better imaging of thin-bed reservoirs, and found optimal attributes for thin-bed reservoir characterization. These techniques were then tested on the Pennsylvanian Caddo sequence of the Boonsville Field, Texas using public domain data. The main results found through this study are: (1) A new pattern recognition model has been developed to recognize the subtle geological and geophysical features of a thin-bed sequence based on cross-correlation of seismic traces with one or more traces believed to represent specific depositional environments. This algorithm has been proven, via the case study, to be robust and promising in defining seismic facies for subtle geological features and predicting thin-bed reservoirs. (2) Examination of the conventional thin-bed tuning model reveals that it works well only if one single thin-bed is developed or multiple thin beds are widely spaced in the sequence of interest. In other words, the model does not work for multiple closely-spaced thin-beds because of significant destructive interference. (3) A statistical inversion method was developed using the generalized regression neural network (GRNN). For a comparison study, two commercial packages were applied to the Boonsville Field data set. This study shows that all three models were able to identify the thicker reservoir sandstones and non-reservoir limestones. However, the resulting details for the thin beds vary. The GRNN method predicted the thin beds at 13 out of 20 wells with less noise and can be very useful in detecting thin-bed reservoirs in existing fields where a number of wells are available.