Electrical and seismic refraction methods: fundamental concepts, current trends, and emerging machine learning prospects -- A review

TL;DR

This review discusses electrical and seismic refraction methods, highlighting recent advances, challenges, and the promising role of machine learning in improving subsurface imaging and interpretation.

Contribution

It provides a comprehensive overview of traditional techniques and explores emerging machine learning applications for enhanced data analysis and inversion in geophysical surveys.

Findings

ML improves inversion accuracy

ML automates lithological differentiation

ML enhances data interpretation

Abstract

This comprehensive review examines electrical and seismic refraction methods, emphasizing their advanced applications in electrical resistivity tomography (ERT) and seismic refraction tomography (SRT). These techniques are crucial for understanding surface-subsurface crustal dynamics, offering critical insights into resistivity and velocity structures for geological and geohazard assessments. The review also explores the induced polarization (IP) and self-potential (SP) methods as complementary approaches. Despite their effectiveness, ERT and SRT face challenges due to lithological heterogeneities, complex geological processes, and geophysical data uncertainties, necessitating multidisciplinary solutions such as methodological advancements and data integration strategies. Recently, machine learning (ML) techniques have been increasingly applied to joint ERT and SRT analyses, optimizing nonlinear inversion processes and improving the characterization of complex subsurface lithologies. The case studies presented in this review evaluate how supervised and unsupervised ML techniques enhance ERT and SRT by improving data interpretation, refining inversion accuracy, automating lithological differentiation, and predicting seismic velocity from resistivity data. The findings underscore the importance of integrating traditional geophysical methods with advanced data-driven approaches to improve subsurface investigations. Continued innovations in ERT and SRT methodologies, along with emerging computational tools and ML applications, will further enhance their effectiveness in geological, hydrological, environmental, and hazard assessments.