CORRELATIONS AND COMPARISONS OF STRIKE AND DIP SECTIONS USING PROBABILISTIC CLUSTERING OF WELL LOGS TO DELINEATE ROCK TYPES AND STRATIGRAPHIC “TOPS” - LOWER AND MIDDLE PALEOZOIC OF CENTRAL NEW YORK STATE

Correlation of strata using well logs has long been a tool for supplementing outcrop study. In well log studies, the picking of formation “tops” can often be aided by using more than one well log curve to help identify facies changes and formation tops. In this study, probabilistic clustering analyses using four well log curves as clustering variables were used to delineate “rock types” which then were used to identify formation tops. The procedure permits correlations between wells because the clustering work flow permits data from multiple wells to be clustered in the same analysis. For this work, gamma ray (GR), bulk density (RHOB), neutron (NPHI), and photoelectric factor (PEF) curves were used. Focus was on selected Lower and Middle Paleozoic stratigraphic units of current economic interest such as the Marcellus and Utica Shales. A model was used that assumes that the main sediment source areas were to the (current) SE and the depositional basin was to the (current) NW during much of the Lower and Middle Paleozoic. Using this model, it might be expected that facies correlations would be more obvious in strike (i.e., NE-SW) sections than in dip (i.e., SE-NW) sections, and that the thickness of stratigraphic units might be more variable in dip sections than in strike sections. Both small-scale (wells close together) and large-scale (wells far apart) dip and strike sections are shown for comparison.