Challenges in Defining Alarm Thresholds to Improve Crisis Management Procedures: A Case Study on the French Riviera

Abstract

When dealing with crisis management in the context of river floods, clear and straightforward procedures must be established. SMIAGE (Syndicat Mixte Inondation, Aménagement et Gestion de l’Eau maralpin) is a public structure dealing with flood, river and water management in the AlpesMaritimes department in the South-East of France. This structure was created to gather the department operational forces to consider water management from the river basin’s point of view, rather than from the administrative limits point of view. Indeed, SMIAGE was created after the deadly event that occurred on the October 3^rd, 2015, when the needs of a wider consideration of flood events were demonstrated on the French Riviera. Moreover, in the Mediterranean context, flash floods must be considered. These events are widespread and poorly understood. One of the main missions of this entity is to assist municipalities by operating a flood warning system, especially for the Siagne catchment.

Everywhere else on the SMIAGE’s territory, the objective is to maintain dikes and other hydraulic structures and to be able to help municipalities to get accurate and clear information while a flood event is happening and afterwards.

While crisis procedures are implemented already, warning thresholds are mainly defined by historical experience and local knowledge. Water depths generating road flooding are often well-known by municipalities. However, intense urbanization has changed catchments’ response and vulnerable areas’ location. Meanwhile, hydrometric and rain gauges together with radar data are available. A reflection is undertaken about a large-scale modelling system. Data of different types are examined with several modelling philosophies. First results show that large-scale modelling can help defining warning water depths at selected river sections. Event management procedures can be refined and improved thus reducing false and missed alarms.