Cerfacs Enter the world of high performance ...

Assessing and reducing uncertainty in large-eddy simulation for microscale atmospheric dispersion

Air quality is severely degraded during events such as industrial accidents. Harmful gases and particles are released into the atmosphere and carried by the wind. In built environments, these pollutants can lead to local pollution peaks due to buildings blocking the flow, resulting in short-term health and environmental risks. Locating these peaks requires the use of models solving the fundamental equations of fluid dynamics and their interactions with the built environment. Despite their complexity, these models are subject to uncertainties that are partly linked to atmospheric conditions. The aim of this thesis is to build and validate a modeling system able of estimating these uncertainties and identifying possible dispersion scenarios. This is achieved by using tools derived from statistical learning and by informing the model with in-situ observations.