Liq-EBM: data-driven assessment of liquefaction triggering and associated ground damage
Journal
Natural Hazards
Journal Volume
122
Journal Issue
8
ISSN
0921030X
Date Issued
2026-04
Author(s)
Abstract
Soil liquefaction poses significant challenges to geotechnical earthquake engineering, requiring accurate assessment methods for seismic hazard evaluation. This study presents Liq-EBM, a novel data-driven approach using Explainable Boosting Machine (EBM) methodology for assessing liquefaction potential and associated ground settlements. The Liq-EBM model required 6 common parameters as inputs: corrected SPT-N values, fines content, soil depth, groundwater level, peak ground acceleration, and earthquake magnitude. Performance evaluation demonstrated improvements over traditional methods. Subsequently, an EBM-based predictive model for maximum shear strain due to liquefaction enabled the estimation of ground settlement caused by liquefaction through extensive numerical simulations and was validated with the available case history database. Implementation in Taipei Basin demonstrates practical utility for regional-scale hazard assessment, establishing quantitative settlement damage classifications: low (≤ 10 cm), moderate (10–25 cm), and severe (> 25 cm). The Liq-EBM framework successfully bridges advanced machine learning techniques with practical engineering applications, providing geotechnical engineers with reliable, transparent tools for liquefaction risk assessment and mitigation planning while maintaining compatibility with existing workflows.
Subjects
Explainable Boosting Machine (EBM)
Liquefaction assessment
Liquefaction hazard map
Liquefaction-induced ground settlement
Publisher
Springer Science and Business Media B.V.
Type
journal article