Braud et al. (2014) Multi-scale hydrometeorological observation and modelling for flash flood understanding

Identification

• Journal: Hydrology and earth system sciences
• Year: 2014
• Date: 2014-09-26
• Authors: Isabelle Braud, Pierre-Alain Ayral, Christophe Bouvier, Flora Branger, Guy Delrieu, Jérôme Le Coz, Guillaume Nord, J.P. Vandervaere, Sandrine Anquetin, Marko Adamovic, Julien Andrieu, C. Batiot, Brice Boudevillain, Pascal Brunet, Julie Carreau, Audrey Confoland, Jean‐François Didon‐Lescot, Jean‐Marc Domergue, Johnny Douvinet, Guillaume Dramais, Rémi Freydier, Simon Gérard, Jessica Huza, Étienne Leblois, Olivier Bourgeois, R. Le Boursicaud, Pierre Marchand, Philippe Martin, Laurent Nottale, Nicolas Patris, Benjamin Renard, J.L. Seidel, J.D. Taupin, O. Vannier, Béatrice Vincendon, Annette Wijbrans
• DOI: 10.5194/hess-18-3733-2014

Research Groups

• Irstea, UR HHLY, Hydrology-Hydraulics, Villeurbanne, France
• Ecole des Mines d’Alès, Alès, France
• ESPACE, UMR 7300 CNRS, “Antenne Cevenole”, Université de Nice-Sophia-Antipolis, Université d’Avignon et des Pays de Vaucluse, France
• Hydrosciences, UMR 5569 CNRS, IRD, Université de Montpellier II, Montpellier, France
• LTHE, UMR 5564 CNRS, IRD, Université de Grenoble, Grenoble, France
• ESPACE, UMR 7300 CNRS, Université de Nice, Nice, France
• ESPACE, UMR 7300 CNRS, Université d’Avignon et du Pays de Vaucluse, Avignon, France
• Centre for Water and Climate, Wageningen University, Wageningen, the Netherlands
• Observatoire de Paris-Meudon, CNRS LUTH et Université de Paris Diderot, Meudon, France
• CNRM-GAME, UMR 3589, Météo-France et CNRS, Toulouse, France

Short Summary

This paper presents a multi-scale observation and modeling strategy to enhance the understanding of flash flood processes in the Mediterranean region. The study demonstrates consistent hydrological behavior across hillslope, small catchment, and regional scales, where progressive wetting leads to saturated conditions and subsequently quicker, larger discharge responses.

Objective

• To document the variability of active hydrological processes from the hillslope scale to the regional scale, both between and during flash floods.
• To describe and simulate the corresponding hydrological processes across these various scales.

Study Configuration

• Spatial Scale: Nested catchments ranging from hillslope scale to regional scale.
  • Hillslope scale: Processes influencing runoff generation and concentration.
  • Small to medium catchment scale (1–100 km²): Impact of network structure and spatial variability of rainfall, landscape, and initial soil moisture.
  • Larger scale (100–1000 km²): River routing and flooding processes.
  • Regional scale: Two large French catchments, Gard (2062 km²) and Ardèche (2388 km²), with nested sub-catchments (e.g., Valescure 3.9 km², Tourgueille 10 km², Gazel 3.4 km², Claduègne 43 km², Auzon 116 km², Avène 60 km²).
• Temporal Scale:
  • Enhanced Observation Period (EOP) of the HyMeX program (4 years, 2012–2015).
  • Continuous measurements over 4 years to capture normal and extreme catchment behavior.
  • Opportunistic measurements during high-intensity rainfall events (Special Observation Periods, SOPs).
  • Long-term time series from operational networks.
  • Rainfall data at 5-minute, hourly, and daily resolutions.
  • Continuous soil moisture monitoring for 1 year at selected hillslopes.

Methodology and Data

• Models used:
  • Hydrological Models: CVN model (developed within LIQUID framework), RuiCells model (cellular automaton), J2000 model (JAMS platform).
  • Hydrodynamic Models: 1-D hydrodynamic model for river routing (to be coupled).
  • Reanalysis/Estimation Systems: SAFRAN–ISBA–MODCOU hydrometeorological chain, TRADHy (radar data processing system).
  • Uncertainty Quantification: BaRatin (Bayesian Rating curve) for stage-discharge relationships, geostatistical space-time rainfall generator, generalized additive models for location, scale and shape (GAMLSS).
  • Data Analysis Methods: Kirchner (2009) method for recession analysis, statistical approaches (e.g., flow duration curves, Bayesian networks, fractal analysis).
• Data sources:
  • Meteorological Observations: Operational rain gauge networks (Météo-France, SPC Grand Delta, EDF; 252 hourly, 160 daily gauges), research rain gauge networks (HPiconet), disdrometers (23 units), operational weather radars (Météo-France at Nimes, Bollène, Sembadel, St Nizier), research X-band Doppler-polarimetric radars (2 units), non-coherent fast-scanning X-band radars (2 units).
  • Hydrological Observations: Operational discharge gauging stations, research discharge gauging stations (LS-PIV – Large-Scale Particle Image Velocimetry, SVR – Surface Velocimetry Radars), limnimeter networks (dense networks, thermo-buttons), salt dilution, current meter methods, hydroacoustic profilers.
  • Soil and Sub-surface Observations: Continuous soil moisture sensors (capacitive, ThetaProbe) at various depths (10, 20, 25, 30, 40, 50 cm), piezometers, tensiometers, electrical resistivity tomography (ERT), lysimeters, infiltrometers, Beerkan tests, perforation method for soil depth.
  • Geochemical Sampling: Opportunistic collection of rainfall, stream water, soil water, and groundwater samples for stable isotopes (δ18O, δD), physico-chemical parameters (pH, electrical conductivity, temperature, Na, Ca, K, Mg, NH4, F, Ci, NO3, SO4, alkalinity), total and dissolved organic carbon (TOC, DOC), and trace elements.
  • Remote Sensing and Geospatial Data: 1 m resolution Lidar Digital Elevation Models (DEMs), 0.5 m resolution satellite images (Quickbird, Pléiades), Landsat images, detailed land cover maps, pedology maps, geology maps, Geographical Information System (GIS) layers.
  • Reanalysis Data: Rainfall reanalysis products (5 min, 1 km² resolution radar rainfall fields; hourly, 1 km² resolution or 5–300 km² hydrological mesh KED rainfall amounts).

Main Results

• Consistent Multi-scale Hydrological Response: Observations across hillslope, small catchment, and regional scales show a consistent pattern: catchments progressively wet, reaching saturated conditions, after which discharge responses become quicker and larger, even for moderate rainfall amounts.
• Flash Flood Process Understanding (9–11 November 2012 event):
  • Geochemical analysis (electrical conductivity, δ18O, Ca, Al, TOC) in the Valescure catchment indicates that "new water" (rainfall) contributed 36–56% to peak runoff, highlighting the importance of sub-surface flow and initial soil moisture.
  • Limnimeter networks and field surveys reveal a significant extension of the active hydrographic network before, during, and after the event, correlating with high soil moisture levels.
  • Catchment responses are differentiated by lithology and land use, with some areas (e.g., badlands, basaltic scoria) showing very rapid reactions to rainfall.
  • Specific peak discharge varies across scales and catchments, influenced by cumulative rainfall and potentially intensity.
• Uncertainty Reduction in Rainfall and Discharge:
  • Radar-rain-gauge merging (KED) significantly reduces the standard deviation of rainfall estimation errors compared to rain gauge-only kriging, particularly for small space-time scales.
  • Combining standard gaugings with non-contact LS-PIV and SVR measurements substantially improves the accuracy and reduces the uncertainty of stage-discharge relationships (e.g., uncertainty at 1.5 m for Volane River reduced from 49% to 29%).
• Enhanced Regional-Scale Process Understanding through Modeling:
  • Integrating catchment storage capacity and saturated hydraulic conductivity of weathered bedrock (derived from recession analysis and linked to geology) into the CVN model significantly improves both long-term and event-based discharge simulations, especially for granite geology.
  • Both bottom-up (CVN) and top-down (Kirchner method) modeling approaches provide good results during wet periods, demonstrating complementary strengths in simulating hydrograph timing and recession dynamics.

Contributions

• Presents a novel, comprehensive multi-scale observation and modeling strategy specifically designed for understanding flash flood processes in the complex Mediterranean environment.
• Establishes a long-term (4-year) experimental setup with nested catchments and diverse instrumentation, including innovative remote sensing and in-situ techniques, to capture both standard and extreme hydrological behaviors.
• Develops and applies an iterative "hypothesis testing" modeling framework that directly links observations to process understanding and model improvement, combining both bottom-up and top-down approaches.
• Advances methods for quantifying and reducing uncertainties in both rainfall fields and discharge measurements, which is critical for accurate flash flood analysis and modeling.
• Provides new insights into the consistent hydrological response across different spatial scales and highlights the crucial roles of initial soil moisture, lithology, and sub-surface flow in flash flood generation.

Funding

• French National Research Agency (ANR) under contract no. ANR 2011 BS56 027 (FloodScale project).
• MISTRALS/HyMeX program.
• Institut National des Sciences de l’Univers/Surface et Interfaces Continentales (OHM-CV).
• Region Rhône-Alpes (Ph.D. theses of M. Adamovic and A. Wijbrans).
• CNES (Pléiades images and data processing through a TOSCA program project).
• EPFL, LaMP, NSSL/NOAA, NOA (additional radars deployed during SOPs 2012 and 2013).

Citation

@article{Braud2014Multiscale,
  author = {Braud, Isabelle and Ayral, Pierre-Alain and Bouvier, Christophe and Branger, Flora and Delrieu, Guy and Coz, Jérôme Le and Nord, Guillaume and Vandervaere, J.P. and Anquetin, Sandrine and Adamovic, Marko and Andrieu, Julien and Batiot, C. and Boudevillain, Brice and Brunet, Pascal and Carreau, Julie and Confoland, Audrey and Didon‐Lescot, Jean‐François and Domergue, Jean‐Marc and Douvinet, Johnny and Dramais, Guillaume and Freydier, Rémi and Gérard, Simon and Huza, Jessica and Leblois, Étienne and Bourgeois, Olivier and Boursicaud, R. Le and Marchand, Pierre and Martin, Philippe and Nottale, Laurent and Patris, Nicolas and Renard, Benjamin and Seidel, J.L. and Taupin, J.D. and Vannier, O. and Vincendon, Béatrice and Wijbrans, Annette},
  title = {Multi-scale hydrometeorological observation and modelling for flash flood understanding},
  journal = {Hydrology and earth system sciences},
  year = {2014},
  doi = {10.5194/hess-18-3733-2014},
  url = {https://doi.org/10.5194/hess-18-3733-2014}
}