28 Geoscience and Hydrogeology in the AI Era: From Predictive Models to Real-Time Applications (Posters)

This session offers a unique platform to showcase advancements in Artificial Intelligence (AI), Machine Learning (ML), and physics-informed methodologies, focusing on topics in hydrogeology, subsurface flow and transport, real-time geophysical inversion, critical minerals, and energy applications.

Sunday, 19 October 2025
8:00 AM - 5:30 PM
HBGCC, Hall 1

From Pixels to Pores: Advancing Formation Characterization through Deep Learning Analysis

Paper number: 28-3
Program: Technical Sessions
Time: 8:00 AM – 8:00 AM

Authors

Abstract

Precise characterization of rock formation physical properties, such as porosity, permeability, and mechanical strengths, is crucial for assessing reservoir quality and forecasting subsurface fluid flow connecting hydrocarbons or groundwater. These parameters control fluid movement, affect hydraulic conductivity, and are critical for optimizing petroleum extraction and planning environmental cleanup in contaminated aquifers. Their accurate estimation depends on the rocks detailed particle size, shape, and distribution data, which are traditionally obtained through labor‑intensive, time‑consuming, and subjective manual analysis of petrographic images. Existing microscopy image analysis tools (e.g., ImageJ) for sediment particles often lack scalability, require manual input, hinder reproducibility, and reduce efficiency in high-throughput petroleum and environmental geoscience research. This study aimed to develop an AI‑deep learning framework for rapid and accurate sedimentary particles identification, quantification, and morphometric 2-dimensional analysis in petrographic optical microscopy images. We collected petrographic sedimentary samples from Sphagnum Swamp owned by Morehead State University in Kentucky. For this preliminary study, we curated a dataset of more than 100 high-resolution images of heterogeneous sedimentary quartz particles using Leica optical microscopy. We adopted deep learning-based convolutional neural network, U-Net for fully automated grains segmentation. We developed in-house algorithms for particle shape (e.g. size, roundness, and sphericity etc.,) determination using Wentworth’s circle formulae, and finally classified the particles (very angular, angular, sub-angular, sub-rounded, rounded, and well rounded) following Power’s scale of roundness method. Future work includes further improvement of our detection method for heterogeneous particle types, and incorporating multimodal features (e.g., mineralogical, geochemical, and textural information) to enable real-time, scalable petrographic image analysis for better reservoir characterization and supporting decision making in hydrogeological studies.

Geological Society of America Abstracts with Programs. Vol. 57, No. 6

doi: 10.1130/abs/2025AM-414816

© Copyright 2025 The Geological Society of America (GSA), all rights reserved. Permission is hereby granted to the author(s) of this abstract to reproduce and distribute it freely, for noncommercial purposes. Permission is hereby granted to any individual scientist to download a single copy of this electronic file and reproduce up to 20 paper copies for noncommercial purposes advancing science and education, including classroom use, providing all reproductions include the complete content shown here, including the author information. All other forms of reproduction and/or transmittal are prohibited without written permission from GSA Copyright Permissions.