Vojtek et al. (2025) Fluvial Flood Inundation Modeling: a Comparative Assessment of 1D and 2D Hydraulic Approach Using MIKE+

Identification

• Journal: Water Resources Management
• Year: 2025
• Date: 2025-12-26
• Authors: Matej Vojtek, Jana Vojteková
• DOI: 10.1007/s11269-025-04423-z

Research Groups

• Department of Geography, Geoinformatics and Regional Development, Faculty of Natural Sciences and Informatics, Constantine the Philosopher University in Nitra, Slovakia
• Institute of Geography, Slovak Academy of Sciences, Slovakia

Short Summary

This study compares 1D and 2D hydraulic modeling approaches using MIKE+ for fluvial flood inundation mapping across three flood scenarios (Q10, Q100, Q1000) in a narrow river valley in western Slovakia. The findings indicate that the 1D model significantly underestimates flood extents and flow depths compared to the 2D model and official 2D benchmark maps, concluding its unsuitability for such geomorphological settings.

Objective

• To compare the performance of 1D and 2D hydraulic modeling approaches for fluvial flood inundation mapping using MIKE+ software and Geographic Information Systems (GIS) for three flood scenarios (Q10, Q100, Q1000).
• To determine if the MIKE+ 1D hydraulic model is suitable for fluvial flood inundation mapping in similar narrow valley/upstream topographical settings and if its results are comparable to the MIKE+ 2D hydraulic model and 2D benchmark (official) flood maps.

Study Configuration

• Spatial Scale: A 3.1 km upstream reach of the Gidra River in western Slovakia, within a hydraulic domain of 0.92 km². The river reach has a bed slope of 0.019 m/m and is located primarily in a narrow valley.
• Temporal Scale: Steady-state flow conditions were simulated for three design flood scenarios with return periods of 10 years (Q10), 100 years (Q100), and 1000 years (Q1000). The 2D model used a simulation time step of 10 s.

Methodology and Data

• Models used: MIKE+ 1D hydraulic model (based on Saint–Venant equations) and MIKE+ 2D hydraulic model (based on 2D shallow water equations/depth-averaged Navier–Stokes equations). Models were run in uncalibrated conditions with identical input parameters.
• Data sources:
  • Digital Elevation Model (DEM): Airborne laser scanned LiDAR DEM (DMR 5.0) with 1 m spatial resolution.
  • Land Use/Land Cover (LULC): Vector layer from Basic Data Base for the Geographic Information System (ZBGIS) 2023, complemented by 2023 orthophotos (15 cm resolution), used to assign Manning’s roughness coefficients.
  • Hydrologic Input: Officially determined design discharges for Q10 (11.6 m³/s), Q100 (20.0 m³/s), and Q1000 (44.5 m³/s) from the Slovak Hydrometeorological Institute (SHMI).
  • Benchmark Data: Official 2D flood hazard maps (created with MIKE+ 2D approach) for Q10, Q100, and Q1000.
  • Evaluation Metrics: Critical Success Index (CSI), Bias, Recall, Precision, F1-score for pixel-to-pixel flood extent comparison; Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Error (ME), and Standard Deviation (SD) for overlapping flow depth pixels.

Main Results

• Flood Extent Comparison (1D vs. 2D Benchmark): The 1D model significantly underestimated flood extents compared to official 2D benchmark maps: Q1000 by 5.59%, Q100 by 49.63%, and Q10 by 56.23%.
• Flood Extent Comparison (2D vs. 2D Benchmark): The 2D model showed better agreement, overestimating Q1000 by 22.31%, slightly underestimating Q100 by 0.71%, and underestimating Q10 by 33.58%.
• Flood Extent Comparison (1D vs. 2D Model): The 1D model's flood extent was consistently smaller than the 2D model's: Q1000 by 26.65%, Q100 by 49.27%, and Q10 by 34.10%.
• Flow Depth Differences (1D vs. 2D Model): For overlapping areas, the mean flow depth differences were -0.11 m (Q1000), -0.09 m (Q100), and -0.10 m (Q10), indicating the 1D model generally produced lower flow depths. Over 61% of overlapping pixels showed 1D flow depths lower by 0.50 m to 1.00 m compared to 2D.
• Cross-Evaluation Metrics (Bias):
  • 1D vs. 2D model: Bias values (Q1000: 0.733, Q100: 0.507, Q10: 0.659) confirmed 1D underestimation.
  • 1D vs. Official: Bias values (Q1000: 0.944, Q100: 0.504, Q10: 0.438) showed significant 1D underestimation, especially for lower return periods.
  • 2D vs. Official: Bias values (Q1000: 1.287, Q100: 0.993, Q10: 0.664) indicated slight overestimation for Q1000, near-perfect match for Q100, and underestimation for Q10.
• Conclusion: The 1D MIKE+ model is not suitable for fluvial flood inundation mapping in similar narrow valley/upstream topographical settings due to its inability to provide reliable and comparable flood extents and flow depths.

Contributions

• First comparative assessment of MIKE+ 1D and 2D hydraulic models for fluvial flood inundation mapping.
• Novel application of pixel-to-pixel cross-evaluation metrics (CSI, Bias, Recall, Precision, F1-score) for comparing MIKE+ 1D and 2D model outputs against each other and against official 2D benchmark flood maps.
• Demonstrated the significant limitations of 1D hydraulic models in narrow valley geomorphological settings, even when using identical input parameters and uncalibrated conditions, highlighting the necessity of 2D or coupled approaches for practical flood mapping.

Funding

• EU NextGenerationEU through the Recovery and Resilience Plan for Slovakia under the project No. 09I03-03-V03-00085.

Citation

@article{Vojtek2025Fluvial,
  author = {Vojtek, Matej and Vojteková, Jana},
  title = {Fluvial Flood Inundation Modeling: a Comparative Assessment of 1D and 2D Hydraulic Approach Using MIKE+},
  journal = {Water Resources Management},
  year = {2025},
  doi = {10.1007/s11269-025-04423-z},
  url = {https://doi.org/10.1007/s11269-025-04423-z}
}