Work

Deep Learning for Identifying Geological Faults in Deep Waters of the Ceará Basin, Brazilian Equatorial Margin

Year: 2025

Type: conference-paper

Abstract: The Ceará Basin, located on Brazil's equatorial margin, has a complex tectonic evolution, with normal and strike-slip faults controlling structural compartmentalization and hydrocarbon distribution. This study proposes an automated methodology for fault interpretation in the Mundaú Sub-basin using Convolutional Neural Networks (CNNs) applied to seismic attributes (Ant Tracking, Variance, RMS Amplitude, Chaos, and Structural Smoothing). The fusion of RMS Amplitude and Chaos improved fault visualization, while the E-Net architecture ensured computational efficiency. Results revealed predominantly NW-SE-oriented normal faults, subvertical in geometry, extending to depths of up to 6 km. The approach demonstrated higher precision and reduced subjectivity in seismic interpretation, contributing to more detailed tectonostratigraphic models and advancements in offshore exploration.

Source: International Congress of the Brazilian Geophysical Society

DOI: 10.22564/19cisbgf2025.116

Authors: Matheus Edson Mendes Medeiros, Ana Krueger, Narelle Maia de Almeida

Institutions: Universidade Federal do Ceará, Bluware, Inc. (United States), University of Houston

Language: English

Topic: Seismic Imaging and Inversion Techniques

Subfield: Geophysics

Field: Earth and Planetary Sciences

Domain: Physical Sciences

Open Access status: bronze

OpenAlex ID: W7125683055

FWCI:

Cites:

Cited by:

Related to: 10