Li et al. (2025) Comprehensive flood risk identification and assessment in small mountainous watersheds using GF-7 satellite imagery and hydrological-hydrodynamic modeling

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2025
• Date: 2025-12-13
• Authors: Jiqing Li, Ke Huang, Liang Wu, Qiumei Ma, Yutao Xie, Yuqing Cao, Wei Zheng, Yi Wang, Junbin Gao
• DOI: 10.1016/j.ejrh.2025.103037

Research Groups

• School of Water Resources and Hydropower Engineering, North China Electric Power University, Beijing, China
• Suzhou Institute of North China Electric Power University, Suzhou, China
• School of Economics and Management, North China Electric Power University, Beijing, China
• Beijing Tidelion Science and Innovation Group Co., Ltd, Beijing, China

Short Summary

This study developed a comprehensive flood risk indicator (HRS&T) by integrating high-resolution Gaofen-7 (GF-7) satellite imagery and hydrological-hydrodynamic modeling to assess flood risk in small mountainous watersheds. The research demonstrated that incorporating land-surface types, terrain slope, and flood duration significantly improves the accuracy of flood risk assessment, particularly for areas prone to secondary hazards like debris flows and landslides.

Objective

• To develop and demonstrate a comprehensive flood risk indicator (HRS&T) suitable for small mountainous watersheds.
• To evaluate the applicability of Digital Elevation Model (DEM) and land cover data derived from high-resolution GF-7 imagery for improving flood risk assessment.
• To provide scientific support for flood prevention, management, and decision-making by governmental authorities.

Study Configuration

• Spatial Scale: Liucun Town watershed in Beijing, China, covering an area of 269 square kilometers.
• Temporal Scale: GF-7 satellite imagery captured on May 13, 2023. Observed rainfall event from July 29–July 30, 2023. Five design rainstorm scenarios with 2, 10, 50, 100, and 1000-year return periods, each with a duration of 1440 minutes (24 hours). Flood duration thresholds for risk classification, notably 3 hours.

Methodology and Data

• Models used:
  • Hydrological-hydrodynamic model based on 2D Saint-Venant equations, solved using the Finite Volume Method (FVM) with Total Variation Diminishing (TVD) schemes and Riemann solvers.
  • Horton infiltration model for pervious areas.
  • Fixed runoff model for impervious areas.
  • Maximum Likelihood Classification (MLC) for land cover classification.
  • Rational Polynomial Coefficients (RPC) model for DEM generation from stereo imagery.
• Data sources:
  • Gaofen-7 (GF-7) satellite imagery (panchromatic 0.8 m, multispectral 3.2 m, pan-sharpened to 0.65 m resolution) from China Platform of Earth Observation System.
  • Observed rainfall data from the Hezijian rain gauge station for the July 2023 event.
  • Design rainstorm scenarios generated using the rainfall intensity formula for Zone II of Beijing (DB11/T 969–2016).
  • Publicly available 30 m DEM for comparison.
  • Field survey data and flood marks for model calibration and validation.

Main Results

• GF-7 satellite imagery successfully generated a 1 m resolution DEM (GF-7 DEM) with a root mean square error (RMSE) of 26.88 m and a mean error (ME) of 9.40 m compared to a 30 m DEM, demonstrating superior fine-scale terrain representation.
• Land cover classification from GF-7 imagery achieved an overall accuracy exceeding 0.82 and Kappa coefficients greater than 0.72, identifying Cultivated Land, Forest Land, Residential Areas, Road Networks, and Water Bodies.
• The hydrological-hydrodynamic model, calibrated with the July 2023 rainstorm event, showed high consistency with observed flood marks, achieving a Mean Absolute Error (MAE) of 0.04 m for simulated water depths.
• Under increasing rainfall intensities (from 2-year to 1000-year return periods):
  • Maximum inundation depth increased from 9.68 m to 29.43 m.
  • Maximum flow velocity increased from 1.90 m/s to 6.61 m/s.
  • Inundation area (depth ≥ 0.15 m) expanded from 245.24 hectares to 595.50 hectares.
  • The time lag between the rainfall peak and the discharge peak at downstream section D shortened from approximately 170 minutes to 45 minutes.
• The "23⋅7" rainstorm event exhibited flood characteristics (inundation area, peak flow) comparable to a 1000-year return period design rainfall scenario.
• The proposed HRS&T indicator, incorporating land-surface type, terrain slope, and flood duration, identified significantly larger high-risk areas compared to traditional indicators. Specifically, in the "23⋅7" rainstorm scenario, the Level 4 high-risk area identified by HRS&T was 14.0% (40 hectares) larger than that identified by the traditional Hazard Rating (HR) indicator.
• Including flood duration in HRS&T increased the Level 4 area by 38.7% (91 hectares) compared to HRS (which includes land-surface type and slope but not duration), highlighting the critical impact of temporal variations.
• Field surveys confirmed that all observed flood damage sites corresponded to high-risk zones delineated by the HRS&T indicator.

Contributions

• Developed a novel semi-quantitative comprehensive flood risk indicator (HRS&T) that integrates flood inundation depth, flow velocity, duration, land-surface sensitivity (exposure and vulnerability), and terrain slope (potential for secondary hazards like debris flows and landslides).
• Demonstrated the effective use of high-resolution GF-7 satellite imagery to extract precise DEM and land cover data, addressing data scarcity in mountainous regions and enhancing the accuracy of hydrological-hydrodynamic modeling inputs.
• Quantitatively showed that incorporating land-surface types, terrain slope, and flood duration significantly improves the accuracy and comprehensiveness of flood risk assessment, particularly in identifying areas susceptible to secondary disasters and refining risk delineation in complex terrain.
• Provided a more realistic and transferable framework for flood risk assessment in mountainous environments, offering methodological and practical advancements over traditional flood mapping techniques.

Funding

• National Natural Science Foundation of China (U2243224, 52579009, 52179014)
• National Key Research and Development Program of China (2022YFC3002702–4)
• Belt and Road Special Foundation of the National Key Laboratory of Water Disaster Prevention, Nan-jing Hydraulic Research Institute (2024nkzd01)
• Fundamental Research Funds for the Central Universities (2025MS075)

Citation

@article{Li2025Comprehensive,
  author = {Li, Jiqing and Huang, Ke and Wu, Liang and Ma, Qiumei and Xie, Yutao and Cao, Yuqing and Zheng, Wei and Wang, Yi and Gao, Junbin},
  title = {Comprehensive flood risk identification and assessment in small mountainous watersheds using GF-7 satellite imagery and hydrological-hydrodynamic modeling},
  journal = {Journal of Hydrology Regional Studies},
  year = {2025},
  doi = {10.1016/j.ejrh.2025.103037},
  url = {https://doi.org/10.1016/j.ejrh.2025.103037}
}