Predictive Modelling in Urban Environments

Urban problems (e.g. turbulent flows, chemical reactions, radiation, heat exchange and interaction between the physical processes) typically have important dynamics that operate over a range of length scales. In this chapter, we first introduce multiscale adaptive mesh numerical methods for the dispersion of urban pollutants at cityCity, building and street scales down to personal scales. The use of adaptive unstructured meshes in urban environmentUrban environments modellingModelling can provide accurate results since the mesh is dynamically adapted according to the evolving physical features. We also present new numerical techniques such as machine learningMachine learning (ML), reduced-order modelling (ROM) and data assimilationData assimilation (DA), for real-time operational modelling and uncertainty analysis. The combination of ROM, ML and DA enables a rapid and accurate modellingModelling response in emergencies. The multiscale adaptive mesh and rapid response modelling capability will be demonstrated in realistic cases—London (UK) and Denmark.