Roux et al. (2011) A physically-based parsimonious hydrological model for flash floods in Mediterranean catchments

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2011
• Date: 2011-09-27
• Authors: Hélène Roux, David Labat, Pierre‐André Garambois, Marie‐Madeleine Maubourguet, Jacques Chorda, Denis Dartus
• DOI: 10.5194/nhess-11-2567-2011

Research Groups

• Institut de Mécanique des Fluides de Toulouse (IMFT), Université de Toulouse (INPT, UPS)
• Centre National de la Recherche Scientifique (CNRS), IMFT
• Laboratoire des Mécanismes et Transferts en Géologie (LMTG), Université de Toulouse (CNRS, IRD, OMP)

Short Summary

This study develops and validates MARINE, a physically-based, parsimonious, spatially distributed hydrological model for flash flood simulation in Mediterranean catchments. The model demonstrates good predictive capability, highlighting the importance of subsurface flow and the spatial variability of rainfall and soil properties for improved flash flood forecasting, especially in ungauged basins.

Objective

• To develop and validate a physically-based, parsimonious, spatially distributed hydrological model (MARINE) for flash flood simulation and forecasting in Mediterranean catchments.
• To exploit the potential of distributed models using physically meaningful parameters while maintaining a simple parameterization.
• To clarify the model's data and calibration requirements and its flood forecasting capabilities, particularly for ungauged basins.

Study Configuration

• Spatial Scale: Gardon d’Anduze catchment (545 km²), located in southern France. Spatial discretization uses a 50 m grid resolution from a Digital Elevation Model (DEM). Radar rainfall data has a 1 km by 1 km spatial resolution.
• Temporal Scale: Individual flash flood events (1994, 1995, 2000, 2002) were simulated. Radar rainfall measurements were available at 360 s (6 min) time steps (1994-2001) and 300 s (5 min) time steps (since 2002).

Methodology and Data

• Models used:
  • MARINE (Modélisation de l’Anticipation du Ruissellement et des Inondations pour des Événements Extrêmes): A spatially distributed hydrological model based on physical process representation (infiltration, overland flow, channel routing).
  • Green and Ampt model: Used for local-scale infiltration.
  • Darcy's law: Used for subsurface flow.
  • 1-D kinematic wave approximation of Saint-Venant equations: Used for surface runoff (overland flow and drainage network flow).
  • Manning friction law: Used to relate flow depth and velocity.
  • Monte-Carlo simulations: Employed for sensitivity analysis of model parameters.
  • LNP criterion: A linear combination of the Nash efficiency coefficient and the error of peak time and runoff, used for model calibration and performance evaluation.
• Data sources:
  • Digital Elevation Models (DEM): 50 m grid scale from the National Geographic Institute (IGN – BD TOPO®) for topography, flow pathways, and drainage network characteristics.
  • Soil surveys: From INRA and BRGM (BDSol-LR) for soil texture and thickness. Rawls and Brakensiek (1983) soil classes were used to determine hydraulic properties (saturated hydraulic conductivity, saturated water content, soil suction).
  • Vegetation and land-use map: 2000 Corine Land Cover (Service de l’Observation et des Statistiques (SOeS)) for deriving distributed surface roughness.
  • Hydrometeorological data: Radar rainfall measurements combined with rain gauge data (6 min or 5 min time steps, 1 km by 1 km resolution).
  • Observed discharge data: From gauging stations at Anduze, Saumane, and Mialet.

Main Results

• The MARINE model, calibrated with four parameters (correction coefficients for hydraulic conductivity (CK) and soil thickness (CZ), overbank roughness (nD2), and initial soil saturation (θi)), showed good agreement with observed hydrographs at the Anduze station for validation events (LNP values of 0.94 for the 2000 flood and 0.91 for the 2002 flood).
• Sensitivity analysis indicated that the model is highly sensitive to the overbank roughness coefficient (nD2) and parameters controlling infiltration (CK, CZ, θi).
• Initial soil moisture content (θi) was identified as a critical parameter, requiring event-specific adjustment for accurate simulations, especially for events occurring in different seasons (e.g., the October 1995 event).
• Simulations at upstream gauging stations (Mialet, Saumane) were less accurate than at the main outlet, particularly at Mialet due to unmodeled karst processes, suggesting the need for regionalization methods.
• Soil saturation dynamics were significantly influenced by the spatial distribution of rainfall and soil properties variability.
• Including a subsurface flow component improved hydrograph agreement during lower rainfall rates and recession periods (LNP increased from 0.69 to 0.92 for 1994, and 0.82 to 0.95 for 1995), and profoundly impacted soil saturation dynamics by highlighting the role of the drainage network in exfiltration.
• The correction coefficients CK and CZ were interpreted as accounting for vertical macropores (accelerating wetting front) and bedrock percolation (loss term), respectively.

Contributions

• Development of MARINE, a novel physically-based, parsimonious, spatially distributed hydrological model tailored for flash flood prediction and analysis in Mediterranean catchments, designed to be applicable in ungauged basins.
• Comprehensive validation of the model using multiple flood events in a complex Mediterranean catchment, demonstrating its predictive capability for extreme events.
• Identification of key model parameters and their sensitivity, providing insights into the dominant hydrological processes during flash floods.
• Emphasis on the critical role of initial soil moisture conditions and the spatial variability of rainfall and soil properties for accurate flash flood modeling.
• Demonstration of the significant impact of incorporating subsurface flow on hydrograph simulation and soil saturation dynamics, suggesting a crucial improvement for model structure, especially for recession periods.
• Offers a physically interpretable framework for calibration parameters, enhancing understanding of underlying hydrological processes.

Funding

Not explicitly stated in the paper.

Citation

@article{Roux2011physicallybased,
  author = {Roux, Hélène and Labat, David and Garambois, Pierre‐André and Maubourguet, Marie‐Madeleine and Chorda, Jacques and Dartus, Denis},
  title = {A physically-based parsimonious hydrological model for flash floods in Mediterranean catchments},
  journal = {Natural hazards and earth system sciences},
  year = {2011},
  doi = {10.5194/nhess-11-2567-2011},
  url = {https://doi.org/10.5194/nhess-11-2567-2011}
}