Télédétection hypertemporelle de la strate arborée urbaine pour l’aménagement urbain

Cities in the 21st century are facing major climate challenges. Urban planning policies are increasingly emphasising sustainability, notably through measures aimed at enhancing urban vegetation. Urban trees are highly valued for their multiple benefits, including improved well-being, better air quality, and enhanced local cooling.In this context, accurate and regular monitoring of their dynamics is essential to fully understand and optimise these ecosystem services. Recent advances in remote sensing, combined with innovative image processing approaches (deep learning), now enable the extraction of detailed information on the structure and temporal behaviour of urban trees. This knowledge can be integrated into spatial indicators, providing actionable data to support urban management decisions.This thesis investigates the potential of Satellite Image Time Series for monitoring and characterising urban trees, with the objective of producing reproducible and scalable information to support sustainable urban planning and climate-adapted management strategies.