Abstract

The subsurface interpretation of an inverted geophysical model is crucial for several reasons. It allows us to understand and characterize the subsurface physical-properties and structures, providing valuable insights into geological features and processes. It aids in identifying and delineating potential targets or anomalies of interest, such as mineral deposits, hydrocarbon reservoirs, or geological hazards. An important aspect of subsurface interpretation involves the classification of different litho-units within the inverted geophysical model. This research explores the application of Support Vector Machine (SVM), a supervised machine learning technique used for regression and classification, to perform lithoclassification. Traditionally, geophysical inversion and litho-classification are considered separate processes. This study introduces the SVM technique, which can be used during inversion or post-inversion processes. We propose integrating this technique directly into the inversion framework, so the need for separate post-inversion lithoclassification analysis can be eliminated. The approach demonstrates two SVM examples for litho-classification. In the first example, SVM accurately delineates the lithological boundaries in the inverted geophysical model, facilitating fault and structure mapping. In the second example, SVM performs litho-classification on a densitysusceptibility cross-plot, which is particularly valuable when density and susceptibility values are available from rock samples of the study area.