Cho et al. (2026) Toward transferable flash flood oriented high flow prediction in ungauged basins using a decoupled physics–DL framework

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-09-03
• Authors: Hyunjin Cho, Hyojun Bak, Gunhaeng Lee, ChangHwan Kim, Hyejin Park, Dae‐Hong Kim
• DOI: 10.1016/j.jhydrol.2026.136346

Research Groups

• Department of Civil Engineering, The University of Seoul
• Han River Flood Control Office, Ministry of Climate, Energy and Environment

Short Summary

This study proposes a decoupled physics–DL framework to predict flash floods in ungauged basins by combining physics-based surface runoff routing with deep-learning based rainfall loss estimation.

Objective

• Investigate the feasibility of predicting flash floods in pseudo-ungauged basins using a decoupled physics–DL framework

Study Configuration

• Spatial Scale: Regional scale, focusing on South Korea
• Temporal Scale: Event-scale (hours to days)

Methodology and Data

• Models used:
  • Dynamic wave-based instantaneous unit hydrograph (DIUH)
  • Multilayer perceptron (MLP) for rainfall loss estimation
• Data sources:
  • Topographical and land-cover information
  • Basin descriptors and rainfall event characteristics

Main Results

• The decoupled physics–DL framework achieved a median critical success index (CSI) of 0.50 under strict warning criteria, with CSI increasing under relaxed tolerance settings.
• Median Kling–Gupta efficiency (KGE) of 0.75 for stage-hydrograph reproduction.

Contributions

• This study provides an effective and interpretable approach for flash flood oriented high flow prediction in pseudo-ungauged basins without local calibration.

Funding

• This research was supported by the Ministry of Climate, Energy and Environment (Han River Flood Control Office)

Citation

@article{Cho2026Toward,
  author = {Cho, Hyunjin and Bak, Hyojun and Lee, Gunhaeng and Kim, ChangHwan and Park, Hyejin and Kim, Dae‐Hong},
  title = {Toward transferable flash flood oriented high flow prediction in ungauged basins using a decoupled physics–DL framework},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136346},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136346}
}