Interpretation Of Shallow Stratigraphic Facies Using Artificial Neural Networks And Borehole Geophysical Data

A study has recently been conducted to assess the extent of hydrocarbon impacts to groundwater<br>and soil resources at a petroleum refinery site in Billings, Montana. To accomplish the study,<br>forty-six groundwater monitoring wells were installed at the site. Data collected from the wells<br>included detailed lithologic descriptions from split-spoon samples, cutting returns from air rotary<br>drilling, and suites of geophysical well logs. Because the quality of the lithologic descriptions<br>from the borings was erratic, our approach was to produce lithofacies interpretations based on<br>gamma ray logs input into a neural network classifier system.<br>The type of neural network used was a self-organizing map. This type of network does not<br>require user interpretations, instead, the network categorizes each input vector into a class based<br>on similarity to other input vectors. The number of output classes is determined by the user. The<br>output classifications were then plotted as ‘pseudo-logs’ and correlation performed using these<br>pseudo-logs.<br>Cross sections constructed using conventional well log interpretation and the neural network<br>classifications show good, general agreement. A significant advantage of the neural network<br>approach over a conventional interpretation approach is that all of the well log data are analyzed<br>concurrently preventing inconsistencies that frequently occur with conventional methods.<br>Another major benefit to the neural network approach is the choice of the number of classes<br>which correlates with the level of lithologic detail that can be resolved.