[M-GI28] Data-driven geosciences

[MGI28-P01] Estimation of lithology and geofluid parameters from seismic velocities and electrical resistivity using Bayesian inference

*Tatsu Kuwatani¹, Kenji Nagata², Toshimoto Sakai³, Hikaru Iwamori⁴ (1.Japan Agency for Marine-Earth Science and Technology, 2.National Institute for Materials Science (NIMS), 3.Chesterford Co. Ltd., 4.Earthquake Research Institute, The University of Tokyo, Tokyo)

Keywords:Geofluid, Bayesian estimation

It is important to convert geophysical observations of the earth's interior into geological information, such as rock type and geofluid amounts. In this study, we develop a Bayesian inversion method that estimates lithology, geofluid types, amounts, and related parameters from seismic-velocity and electric-conductivity data. The proposed method utilizes a marginalization technique to first select the lithology and geofluid types, then quantify the geofluid parameter values. Through artificial data tests, we demonstrate that our method can evaluate the possibility of various combinations of lithology and geofluid types from over a hundred candidate sets, providing approximate estimations for other unknown parameters. In our presentation, we will introduce the methodology and results, as detailed in the recently published paper by Kuwatani et al. (2023, JGR: Solid Earth), and discuss the challenges and potential applications.