An AI‐Enabled Data Processing Pipeline for Ingesting Borehole Data in Peridotite Environments

Abstract

Researchers analyzing data collected from borehole drilling projects can face dozens of terabytes of seismic, hydrologic, geologic, and rock mechanics data, including complex imagery, physical measurements, and expert-written reports. These diverse data sets play a pivotal role in constraining solid-Earth processes. Ingesting and analyzing such data presents a colossal challenge that typically demands a team of experts and a lot of time. Artificial intelligence (AI) and machine learning have emerged as compelling approaches to tackle volume and complexity of drilling data. This paper presents an AI-based pipeline for ingesting data from the Oman Drilling Project's Multi-borehole Observatory. The study focuses on the alteration of peridotite core segments taken from Borehole BA1B, utilizing a gradient-boosted trees (CatBoost) regression model trained on an integrated data set of machine-learning segmented core images, physical measurements, geological, lithographic data, and AI-summarized expert texts and feature selection. This paper aims to establish a repeatable and efficient pattern for processing such multifaceted data from the well. We present results using the data set generated from BA1B. First, we examine the relationship between fracture/vein networks and peridotite alteration, a stand-in for historical fluid flows. Here we demonstrate that we do not find a strong relationship between these networks and alteration. Then we examine the very strong and also nonlinear relationship between alteration and the magnetic susceptibility and resistivity measured in BA1B.