Mélodie DELAMARE, Laboratoire IDEES - Université de Rouen, France
Delphine GRANCHER, Laboratoire de Géographie Physique - Université Paris 1 Panthéon Sorbonne, France
Eric DAUDÉ, Laboratoire IDEES - Université de Rouen, France
On 09/26/2019, a fire starts in Lubrizol and Normandie Logistique factories, in the industrial area of Rouen (Normandie). This accident had a national stir. First, the factory burned is registered in Seveso directive, with toxic effects risks. Then, heavy smoke dispersed throughout the day and beyond Métropole Rouen Normandie, until Hauts-de-France. Last, that crisis exposed alert difficulties again, also communication issues to the population.
To decipher how the population reacted, two laboratories, IDEES in Normandie-Université and LGP in Meudon, collaborated using different data sources: behaviours survey, mobile phone data and social network data. More than 1 600 people took part in the survey; hundreds of thousands twitter messages have been collected and Orange Flux-Vision data have been acquired in Métropole Rouen Normandie. All in all have been used to build a typology of behaviours timelines. Data show that the population reacted in different ways during the event (escape, lockdown, indifference…) and that these reactions evolved during the day. This variety observed in the behaviors must be linked both to individual dispositions (economic category, place of residence, socio-demographic status) and to a more massive presence of social networks on the information scene than the crisis management authority. Indeed, with a fire that started at 2:40am, first authorities’ informations to the population came at 6:00am during a press conference and through an alert system at 7:50am. In spite of an official message stipulating that day to avoid unnecessary travel, and a general safety instruction which establishes the principle of confinement in case the alert system is triggered, on Thursday and following days, between 10% and 20% of people escaped their home. We present these results in the form of chronogram to shows that population make decisions throughout the day, and following days, as a decision-making process rather one-time decision.
Mots clés : Industrial incident|alert|population behaviours|risk management
A104200MD