Kim et al. (2026) Diagnostic method of inundation using a dynamical hydrological model

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2026
• Date: 2026-03-31
• Authors: Minwoo Kim, Eun-Chul Chang, Seok-Geun Oh, Sunghwa Choi
• DOI: 10.1016/j.ejrh.2026.103391

Research Groups

• Department of Atmospheric Sciences, Kongju National University, Gongju 32588, Republic of Korea
• Specialized Graduate School for Integrated Analysis of Meteorological and Climatic Data, Kongju National University, Gongju 32588, Republic of Korea
• School of Earth and Environmental Sciences, Seoul National University, Seoul 08826, Republic of Korea
• School of Civil and Environmental Engineering, Yonsei University, Seoul 03722, Republic of Korea
• Water Resources Satellite Center, K-water Research Institute, Daejeon 34045, Republic of Korea

Short Summary

This study develops and validates the Velocity-based Inundation (VI) method, a non-physics-based post-processing algorithm that calculates overbank flow volume from WRF-Hydro streamflow simulations and distributes it onto a high-resolution Digital Elevation Model (DEM) based on river velocity and topography, demonstrating robust diagnostic performance for flash flood assessment in complex terrains.

Objective

• How does the structural spatial misalignment between coarse 1D hydrological channel routing and actual river geometry limit the accuracy of flood inundation mapping in complex topographical regions?
• To what extent can a decoupled, volume-based spreading mechanism (the VI method) mitigate these grid-resolution constraints and reproduce physically consistent flood wave propagation?
• What quantitative hydrological insights regarding natural flood vulnerabilities can be derived when this simplified framework is validated against operational Synthetic Aperture Radar (SAR) observations during extreme heavy rainfall events?

Study Configuration

• Spatial Scale: Geum River basin (9915 km²) and Nackdong River basin (23,688 km²) in South Korea. WRF domains at 10 km and 9 km, nested to 2 km and 1.8 km. WRF-Hydro resolutions from 1000 m down to 200 m (optimal < 300 m or 200 m). ASTER GDEM at approximately 30 m resolution.
• Temporal Scale: Idealized rainfall experiment (4 days). Real-case extreme flood event: August 7–11, 2020 (simulated from August 1 or 2 to August 19, 2020). SAR observations from August 1/2 and August 7/8, 2020.

Methodology and Data

• Models used: WRF-Hydro (v5.3) with Noah-MP land surface model, WRF (v4.3), Velocity-based Inundation (VI) method (non-physics-based post-processing algorithm).
• Data sources:
  • Reanalysis: ERA5 (ECMWF) for WRF initial/boundary conditions (0.25° horizontal resolution, 3-hour intervals).
  • Satellite/Observation:
    • ASTER GDEM (Version 3) from METI and NASA (approximately 30 m horizontal resolution) for high-resolution terrain.
    • Sentinel-1 Synthetic Aperture Radar (SAR) scenes (Copernicus programme) for flood extent validation.
    • Integrated Multi-satellitE Retrievals for GPM (IMERG) from NASA (0.1° × 0.1° horizontal resolution, half-hourly temporal resolution) for precipitation validation.
    • National Geographic Information Institute (NGII) for South Korean river shapes.

Main Results

• WRF-Hydro resolutions finer than 300 m (or 200 m) are essential to accurately represent the complex river networks of the Korean Peninsula.
• Idealized simulations demonstrated that the VI method accurately captures topographic control on flood propagation: steep upstream terrain confines floodwater, while flat downstream areas permit broad lateral spreading.
• The VI method showed robust diagnostic performance against Sentinel-1 SAR observations for the August 2020 extreme flood event, achieving Accuracies (ACC) of 0.890 for the Geum River basin and 0.855 for the Nackdong River basin.
• Spatial misalignment between the model's channel grid and actual river geometry was identified as the primary source of uncertainty, leading to lower Probability of Detection (POD) in the Nackdong basin (0.372) compared to the Geum basin (0.658), despite comparable False Alarm Ratios (FAR) (~0.48).
• The study highlights that natural flood vulnerability in the Korean Peninsula is governed by the interaction between steep-to-flat geomorphological transitions and the fidelity of the channel network representation.

Contributions

• Development of the novel Velocity-based Inundation (VI) method, a non-physics-based post-processing algorithm that decouples inundation mapping from the host hydrological model's grid resolution.
• The VI method allows excess overbank flow volume to realistically overflow beyond hydrological model grid boundaries, dynamically filling a high-resolution DEM based on local topography and river velocity, overcoming limitations of previous grid-constrained approaches.
• Quantitative validation of a simplified inundation algorithm against SAR-derived observations over complex topography and rapid hydrological response regions (Korean Peninsula).
• Provides new hydrological insights for the Korean Peninsula regarding flood vulnerability and the importance of channel grid resolution.
• Establishes a computationally efficient WRF–WRF-Hydro–VI framework that can be seamlessly integrated into operational weather prediction systems (e.g., KLAPS, HRRR) for pre-emptive, dynamic flood risk assessments.

Funding

• Korea Meteorological Administration Research and Development Program "Enhancement of Convergence Technology of Analysis and Forecast on Severe Weather" (Grant KMA2018–00121).

Citation

@article{Kim2026Diagnostic,
  author = {Kim, Minwoo and Chang, Eun-Chul and Oh, Seok-Geun and Choi, Sunghwa},
  title = {Diagnostic method of inundation using a dynamical hydrological model},
  journal = {Journal of Hydrology Regional Studies},
  year = {2026},
  doi = {10.1016/j.ejrh.2026.103391},
  url = {https://doi.org/10.1016/j.ejrh.2026.103391}
}