The proportion of the global population living in cities is rapidly increasing and is expected to reach 80\% by 2050. This trend highlights the urgent need for efficient and reliable tools to model the urban microclimate—crucial for helping urban planners and policymakers design more comfortable and sustainable cities. Since pollutant dispersion in urban environments depends heavily on local weather conditions, high-fidelity computational fluid dynamics (CFD) simulations are often required. These simulations involve fine spatial and temporal resolution and must be repeatedly evaluated, leading to high computational costs and necessitating high-performance computing (HPC) resources. To overcome these challenges, Reduced Order Models (ROMs) and scientific machine learning provide a promising path by enabling real-time predictions with significantly reduced computational requirements and minimal loss of accuracy. In this talk, I will present recent advances in real-time simulation tools for urban and environmental flows. Starting from classical POD-Galerkin model reduction methods, I will introduce novel techniques that go beyond the limitations of standard linear approaches. These include nonlinear approximation strategies using autoencoders and graph neural networks, combined with nonlinear manifold reduction [1] to preserve key physical principles. Computational efficiency is maintained through hyper-reduction techniques [2]. I will also discuss recent efforts in downscaling meteorological data to urban-scale simulations using physics-informed machine learning methods. REFERENCES [1] Francesco Romor, Giovanni Stabile, and Gianluigi Rozza. Non-linear manifold ROM with Convolutional Autoencoders and Reduced Over-Collocation method. Journal of Scientific Computing, 94(3), 2023. [2] Francesco Romor, Giovanni Stabile, and Gianluigi Rozza. Explicable hyper-reduced order models on nonlinearly approximated solution manifolds of compressible and incompressible navier-stokes equations. Journal of Computational Physics, 524:113729, March 2025.