• Funding : Artois, Université Mohammed V de Rabat
  • Start year :
  • 2026

This thesis focuses on flood risk mapping through advanced approaches of information fusion from heterogeneous data sources, including satellite imagery, geospatial datasets, and environmental observations. It is framed within Geographic Information Systems (GIS) and leverages artificial intelligence methods, including machine learning, statistical approaches, and spatio-temporal modeling.

The main objective is to exploit data fusion techniques to improve the understanding and representation of flood risk by identifying exposed areas and inferring potentially vulnerable zones, even in the absence of direct observations. A particular emphasis is placed on the handling of uncertainty, which arises from data heterogeneity, sensor limitations, and the inherent complexity of hydrological processes. Accounting for uncertainty enables the production of more robust and reliable flood risk maps.

Overall, this approach aims to provide a decision-support tool for flood prevention and management by enhancing the accuracy, consistency, and reliability of spatial and temporal risk analysis.