42 Geophysics/Geodynamics (Posters)

Added due to amt of discipline abstracts submitted.

Wednesday, 11 March 2026
9:00 AM - 12:00 PM
RCC, Lower Level Hall

AI-Driven Fault and Lineament Mapping from High-Resolution LiDAR in Western Kentucky: Random Forest and U-Net Workflows for Geologic Mapping and Seismic-Hazard Screening

Paper number: 42-3
Program: Technical Sessions
Time: 9:00 AM – 9:00 AM

Authors

  • Sujan Adhikari
    Western Kentucky University, Bowling Green, KY
  • Nahid Gani
    Western Kentucky University, Earth, Environmental & Atmospheric Sci, Bowling Green, KY
  • Owen Blaylock Presenting Author
    Western Kentucky University, Earth, Environmental and Atmospheric Science, Bowling Green, KY
  • Dayvon Woodard
    Western Kentucky University, Earth, Environmental and Atmospheric Science, Bowling Green, KY

Abstract

Western Kentucky lies between the New Madrid and Wabash Valley seismic zones, yet remains poorly characterized for seismic hazard because intraplate faults are commonly concealed by thick sedimentary cover and expressed only as subtle topographic and drainage anomalies. We use 1.5-meter resolution LiDAR digital elevation models (DEMs) to build an AI-enabled, repeatable geospatial workflow for fault/lineament mapping that also leverages hydrologic and geomorphic signals of structure. We construct a multi-scale geomorphometric stack from the DEM (multi-azimuth hillshade composites, slope, curvature, roughness, and topographic position indices) and test three automated fault-detection approaches. These include (1) an interpretable Random Forest classifier trained on geomorphic indices, (2) a U-Net convolutional neural network for multi-channel semantic segmentation of lineament patterns, and (3) a hybrid pipeline that integrate classic edge-detection filters with deep-learning-derived feature maps. Model outputs are evaluated against existing state geologic maps, published lineament interpretations, drainage-network attributes, and historical seismicity to assess both geologic plausibility and geospatial performance. Across test areas, the models improve the continuity and spatial confidence of candidate fault-related lineaments and identify corridors where structural control on drainage is strongest. We translate these results into uncertainty-aware mapping products that can be iteratively refined with targeted field checks and new training labels. Detected structures are incorporated into a preliminary seismic hazard screening layer that combines fault likelihood with earthquake catalogs, site conditions, and exposure layers (infrastructure and building footprints) to highlight zones of elevated vulnerability. Although motivated by seismic-hazard reconnaissance in a data-limited intraplate setting, the workflow and derived products are directly transferable to hydrologic and geomorphic mapping, including delineation of structurally influenced drainage corridors, valley-lineament networks, and terrain units from LiDAR-based surface metrics.

Geological Society of America Abstracts with Programs. Vol. 58, No. 1

doi: 10.1130/abs/2026SE-411578

© Copyright 2026 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.