Liang et al. (2026) Advancing Urban Flood Risk Mapping: A Hybrid Framework Integrating Interpretable Machine Learning and Uncertainty-Aware Expert Judgment

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Water
• Year: 2026
• Date: 2026-08-30
• Authors: Shuoyuan Liang, Tsuyoshi Kinouchi
• DOI: 10.3390/w18172141

Research Groups

Not specified

Short Summary

This study develops a hybrid framework integrating CatBoost machine learning and Z-number-based Fuzzy Analytic Hierarchy Process (Z-FAHP) to map urban flood risk in Tokyo, Japan, by combining physical susceptibility with socioeconomic exposure and vulnerability.

Objective

• To create a comprehensive urban flood risk mapping framework that integrates interpretable machine learning for susceptibility and quantifies expert judgment uncertainty in socioeconomic assessments.

Study Configuration

• Spatial Scale: Urban scale (Tokyo, Japan)
• Temporal Scale: Not specified

Methodology and Data

• Models used: CatBoost (Machine Learning), SHAP (SHapley Additive exPlanations), Z-number-based Fuzzy Analytic Hierarchy Process (Z-FAHP), and simple additive weighting for risk synthesis.
• Data sources: High-resolution digital elevation model (DEM) derived from airborne LiDAR, public geospatial datasets, and socioeconomic indicators (3 for exposure, 5 for vulnerability).

Main Results

• CatBoost was identified as the optimal model for flood susceptibility mapping, validated through non-spatial and spatial 4-fold cross-validation.
• Approximately 32.9% of the study area is classified as having high-to-very-high flood risk.
• High-risk zones are primarily concentrated in the eastern lowlands and along river corridors.

Contributions

• Integrates socioeconomic dimensions into flood risk assessment, moving beyond simple susceptibility mapping.
• Enhances the quantification of expert judgment by incorporating confidence levels via Z-numbers, reducing the subjectivity inherent in traditional FAHP.
• Provides a transferable, interpretable framework for urban flood risk management in other global cities.

Funding

Not specified

Citation

@article{Liang2026Advancing,
  author = {Liang, Shuoyuan and Kinouchi, Tsuyoshi},
  title = {Advancing Urban Flood Risk Mapping: A Hybrid Framework Integrating Interpretable Machine Learning and Uncertainty-Aware Expert Judgment},
  journal = {Water},
  year = {2026},
  doi = {10.3390/w18172141},
  url = {https://doi.org/10.3390/w18172141}
}