Gamma-ray log derivative and volatility attributes assist facies characterization in clastic sedimentary sequences for formulaic and machine learning analysis

Rate of change, second derivative and volatility gamma-ray (GR) well-log curves provide useful indicators with which to characterize lithofacies in clastic sedimentary sequences. Rolling averages these variables, as they change depth, are also able distinguish certain features. These attributes make it possible accurately key facies by using only data, both formulaic calculations employing machine-learning (ML) algorithms. This is the many wellbores for basic logging suites available. As well enhancing classification more generally GR can be used forecast real time based on logging-while-drilling data. The application demonstrated simple formula synthetic logs featuring common their transitions. Seven widely ML methods each trained validated a curve (1450 data points) displaying six distinct facies. ability model those seven compared further tested an independent set (800 points). random forest algorithm outperforms other models this prediction task, achieving mean absolute error 0.25 (on class range 1 6) testing dataset. results highlight benefit technique providing reliable analysis Random forest, support vector eXtreme Gradient Boost that most from defined. Annotated confusion matrices assist revealing details accuracy precision achieved formulas. Cited as: Wood, D.A. Gamma-ray log characterization sequences machine learning analysis. Advances Geo-Energy Research, 2022, 6(1): 69-85. https://doi.org/10.46690/ager.2022.01.06