Stratigraphic interpretation has largely been done by geological, geophysical, and well-logging techniques. Statistical analysis of geophysical and well-logging data has been rather well accepted during the past 10 to 20 years. Analysis of geochemical data has only been recently accepted as a technique that aids in correlation and interpretation of other stratigraphic problems.

Two statistical techniques which have proven useful in stratigraphic or other types of geological interpretation have been cluster and discriminant analysis. These techniques are generally regarded as investigative tools. However, they can also be used in a predictive sense. This is particularly true of discriminant analysis. In both techniques, one allows the observed data to describe the similarity of formations, produced fluids, or some other measure of interest. These statistical methods have been applied using oil or water chemistry, well-logging data, and various reservoir parameters to solve stratigraphic problems or support a hypothesis. These techniques have been used successfully to recognize bypassed oil, sedimentary environments of deposition, miscorrelated horizons, and e roneous data.

In this paper, several case studies are presented that involve cluster and discriminant analysis to establish or substantiate hypotheses about reservoir continuity. These include one North Sea field, one Middle East field, and one domestic reservoir. A model which was previously proposed is further extended; this model will distinguish whether a produced water has been in contact with either oil or gas.

These techniques can be advantageous in planning future exploration strategy.

End_of_Article - Last_Page 581------------