Railway flood and landslide vulnerability, a bibliometric review of AI applications and research gaps

Authors

  • Surachai a/l Airak Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia , University of Technology Malaysia image/svg+xml https://orcid.org/0000-0002-9538-9041
  • Sitti Asmah Hassan Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia , University of Technology Malaysia image/svg+xml https://orcid.org/0000-0001-7196-3793
  • Nor Eliza Alias Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia , University of Technology Malaysia image/svg+xml
  • Nordiana Mashros Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia , University of Technology Malaysia image/svg+xml
  • Kojiro Matsuo Dept. of Architecture & Civil Engineering, Toyohashi University of Technology, 441-8580 Japan , Toyohashi University of Technology image/svg+xml
  • Himawan Hadi Sutrisno Faculty of Engineering, Universitas Negeri Jakarta, 13220 Indonesia , State University of Jakarta image/svg+xml
  • Muralia Hustim Environmental Engineering Department, Faculty of Engineering, Hasanuddin University, 92171 Sulawesi Selatan, Indonesia , Hasanuddin University image/svg+xml

DOI:

https://doi.org/10.15282/construction.v6i2.13631

Keywords:

Railway Infrastructure, Flood Vulnerability, Landslide Risk, Artificial Intelligence (AI), Bibliometric Analysis, Decision Support

Abstract

Railway networks are increasingly exposed to flood and landslide disruption. Despite the growing existing literature, it still remains fragmented and not sufficiently railway centred. Therefore, this review attempts to map and synthesise peer-reviewed research on railway vulnerability to floods and landslides, with particular focus on artificial intelligence applications. SCOPUS database was used as the bibliographic data source, spanning from the year 2000 to 2025. A total of 546 articles were obtained analysed via bibliometric mapping with narrative synthesis. Results show that research in rail-flood and rail-landslide is geographically concentrated, with China dominating the field. In addition, international collaboration is limited which reduces transferability across different railway contexts. The field is better described as method-led rather than railway-led as the keyword structure suggests that generic artificial intelligence and hazard terms are much more prominent than railway-centred descriptors. Machine learning, deep learning, remote sensing, susceptibility assessment, and forecasting dominate the field. The synthesis of top cited railway-related studies indicates that AI becomes more useful when model outputs are linked to practical management decision. Cross-study comparability remains limited as substantial variations occur in outcome definitions, validation strategies, and local operating assumptions. Based on the findings, future research should better connect AI-related models with exposure, sensitivity, adaptive capacity, and operational decision-making in railway systems. Overall, the rail-flood and rail-landslide field in relation to AI is still insufficiently aligned with the practical realities of railway vulnerability reduction.

Downloads

Download data is not yet available.

Published

2026-11-17

Issue

Section

Review Article

How to Cite

[1]
S. Airak, “Railway flood and landslide vulnerability, a bibliometric review of AI applications and research gaps”, Constr., vol. 6, no. 2, p. In-Press, Nov. 2026, doi: 10.15282/construction.v6i2.13631.