MUltiscale Flood Forecasting with INnovating Solutions – MUFFINS

Floods are the first weather related disaster worldwide, expected to increase with global change, and effective flood warning systems are crucially needed to mitigate their devastating effects. Nevertheless, performing accurate flood forecasts in terms of location, magnitude and timing of runoff and flooding, and identifying areas "at risk", i.e.prone to trigger fatalities and economic losses, remains a key challenge. Forecasting lead-time to manage the crisis can range from minutes to days, depending if assets at risk are located downstream a small watershed (<1km²) or a large catchment (thousands of km²). The development of suitable flood forecasting systems, in France and worldwide, is still faced with serious challenges due to:
(1) the multi-scale complex processes generating floods and damages,
(2) the high modeling uncertainties especially on ungauged (often most devastating) watercourses and
(3) an insufficient integration of assets vulnerability, risk-damages assessments and real end-users’ needs.
Therefore the overarching objective of the MUFFINS project is to develop new accurate and computationally efficient flood forecasting approaches, enabling transferring information between modelings (meteo-hydrology-hydraulic-damage) and scales (from local runoff generation over areas lesser than 1km² to flood propagation on catchments of thousands km²), and taking advantage of innovative data (in situ, earth observation, opportunistic) to reduce forecasts uncertainties.
A multidisciplinary consortium, gathering end-users, hydrométéorologists, hydroinformaticians and applied mathematicians, will address the key elements required to build such effective flood forecasting chains:
• multi-scale forecasting capabilities, from small scales (<1km²) to catchment scales (thousands km²) with lead times ranging from 1h or less to ~1d, obtained by coupling short range precipitation forecasts, spatially distributed scalable and regionalized hydrological models and high resolution effective hydraulic models including hydrological processes;
• innovative methods performed to integrate information from multi-source data and to reach fast computation times, which include model reduction and offline-online strategies combined with statistical, data assimilation, machine learning methods;
• impact modeling and expert specifications-evaluations for the forecasting chains, achieved thanks to multi-scale (flash) flood cases, rich datasets, damage models and expertise provided by end users.
The case studies will be located in the French Mediterranean region, especially in the Var, Bouches du Rhône and Alpes-Maritime departments where several recent flood events offer a particularly favourable context in terms of data availability (including damage data) and of applicability of the panel of multi-scale approaches developed. These events include the June 2010 and October-November 2018 floods in the Argens watershed, the January 2014 floods in the Gapeau watersheds, the October 2015 and October 2020 floods in the Alpes-Maritimes region. Rich datasets composed of physiographic and hydrometric datasets, post-flood field surveys and new data (earth observation, video camera/drone, opportunistic) will enable improving models accuracy.
The main expected benefits of the project should be:
(1) New methodologies and tools for the next generation of spatially distributed flood forecasting methods with multi-sourced data integration,
(2) Operational demonstrators of hydrological-hydraulic flood forecasting methods affordable in real time and transferable to end users at national,regional and local scales,
(3) Upgrade of the national ’Vigicrues Flash’ operational flood forecasting chain for ungauged catchments, in the perspective of the future infra-departmental flood vigilance,
(4) Refined risk-cost modeling and improved capability of insurance or reinsurance (CCR) companies to take effective action after each event.