Machine Learning Mineralogy Classification Comparison to Empirical Log Relationship and Implication for Physics Informed Modeling

David J. Emery, Marcelo Guarido, Daniel O. Trad

Determination of seismic lithology, porosity and pore fluid requires detailed modelling of petrophysical logs to improve the correlation with a seismic AVO response. Unfortunately, acquiring a complete set of logs for all wells in a seismic survey is unpractical, and estimating sonic, shear and density using empirical relationships is the standard approach. While these empirical relationships have worked for recon analysis, they have generally not given the details needed for accurate geophysical analysis. Machine Learning has given us a new way of investigating these relationships. By analyzing over 138 wells with DT, Vs & RHOB logs from the North Sea, Australia, and Canada, we could generate synthetic Vp, Vs, and RHOB using traditional and the XGBoost regressor, where the latter showed to work better in this data.