This study aims to study the risk of simultaneous flooding in river networks. The objective is to develop a graphical model that uses climate variables as input to estimate the extreme river flow values that may occur under specific climatic conditions and then know which rivers are more likely to experience simultaneous extreme river flows.The challenge is to develop a parcimonious/non-complex model.

The research investigates the spatial relationships between extremes of precipitation and rivers flows within the Garonne watershed using the Husler-Reiss graphical model for multivariate extremes. This graphical model for extremes captures the conditional dependencies between variables. This includes dependencies between river flow stations, between precipitation at different locations and between precipitation and river flows.