This multidisciplinary thesis will be carried out in two ZRR laboratories, LMFA and LIRIS, at the Centrale Lyon site, under the supervision of Pietro Salizzoni (LMFA), Emmanuel Dellandréa (LIRIS), and Lionel Soulhac (LMFA).

Accelerated urbanization and the urban heat island effect are increasing heat stress on populations, a phenomenon exacerbated by climate change. In Southern Europe, where an aging population increases vulnerability, urban adaptation is becoming a priority. By 2070, cities in southern France could experience more than 30 days of heat waves per year, with temperatures reaching 50°C.

Urban climate modeling is a fundamental tool for anticipating these impacts and designing effective mitigation strategies. However, existing approaches, based on the numerical resolution of equations modeling the dynamics and thermodynamics of atmospheric flows, have limitations: while mesoscale models allow for the simulation of large-scale urban processes, they oversimplify urban morphology and the effect of vegetation. Conversely, more accurate microscale models remain limited by their high computational cost, making their large-scale operational application difficult. In this context, this project is part of recent research efforts aimed at developing operational models that combine accuracy and speed of calculation through the integration of machine learning methods.

 The objective of this thesis project is therefore to strengthen urban climate simulation capabilities at the street level and apply these digital tools to the evaluation of greening scenarios, using Lyon as a case study. Local authorities have a crucial need for insight into the cooling potential of adaptation strategies based on greening. The project addresses scientific questions related to the adaptation of territories to climate change and modeling tools for climate risk management in urban areas. It will aim to develop an optimized operational model, integrating machine learning algorithms and field data, in order to accurately simulate thermal exposure at the metropolitan level and evaluate the most effective mitigation levers.