Wang et al. (2026) Unraveling uncertainty in compound flood modeling: sensitivity of simulations to forcings and model parameters

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-03-29
• Authors: Chen Wang, Francisco Javier Gómez, Mohammad Mosavat, Soheil Radfar, Hamed Moftakhari, Hamid Moradkhani
• DOI: 10.1016/j.jhydrol.2026.135424

Research Groups

• Department of Civil, Construction and Environmental Engineering, The University of Alabama, Tuscaloosa, AL, USA
• Department of Computer Science, The University of Alabama, Tuscaloosa, AL, USA

Short Summary

This study evaluates the sensitivity of the SFINCS model to forcing and parameter uncertainties during the Beryl compound flood event in Houston. It reveals strong spatial variability in forcing sensitivity and consistent parameter sensitivity, concluding that parameter-induced uncertainty generally outweighs forcing-induced uncertainty.

Objective

• To evaluate the sensitivity of the SFINCS reduced-physics hydrodynamic model to external forcings and internal parameters during a compound flood event, aiming to improve prediction and risk management.

Study Configuration

• Spatial Scale: Coastal regions, specifically focusing on the area around Houston, USA.
• Temporal Scale: The Beryl compound flood event, which occurred in July 2024.

Methodology and Data

• Models used: SFINCS (Super-fast INundation Code for Surge) hydrodynamic model.
• Data sources: Forcings (rainfall, river discharge, storm surge, wind, pressure) were perturbed within physically realistic ranges to represent measurement and reanalysis errors. Nine model parameters were systematically varied.

Main Results

• Forcing sensitivity showed strong spatial variability: river discharge dominated upstream stations, while boundary water surface elevation controlled downstream levels. Wind and pressure had negligible influence under the event's hydro-geomorphic conditions.
• Parameter sensitivity was consistent across stations, with water-level accuracy primarily governed by the numerical stability parameters theta and alpha, and Manning’s roughness (N).
• Fixing theta at its recommended value (1.0) rendered alpha insensitive, indicating its effect was numerical and coupled with theta.
• After constraining numerical parameters, Manning’s N emerged as the main physical control on water surface elevation (WSE) accuracy, while GridSize primarily governed runtime.
• Parameter-induced uncertainty was generally greater than forcing-induced uncertainty at most stations, although localized forcing biases could occasionally modulate the total response.

Contributions

• Provides a practical framework for improving compound flood prediction and risk management by jointly quantifying forcing and parameter uncertainties.
• Offers specific insights into the relative importance of different forcings (discharge, boundary WSE, wind, pressure) and model parameters (theta, alpha, N, GridSize) for the SFINCS model.
• Highlights the dominance of parameter-induced uncertainty over forcing-induced uncertainty in compound flood modeling for the studied event.

Funding

Not explicitly mentioned in the provided text.

Citation

@article{Wang2026Unraveling,
  author = {Wang, Chen and Gómez, Francisco Javier and Mosavat, Mohammad and Radfar, Soheil and Moftakhari, Hamed and Moradkhani, Hamid},
  title = {Unraveling uncertainty in compound flood modeling: sensitivity of simulations to forcings and model parameters},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.135424},
  url = {https://doi.org/10.1016/j.jhydrol.2026.135424}
}