GenDA: Generative Data Assimilation on Complex Urban Areas via Classifier-Free Diffusion Guidance

Urban wind flow reconstruction is essential for assessing air quality, heat dispersion, and pedestrian comfort, yet remains challenging when only sparse sensor data are available. We propose GenDA, a generative data assimilation framework that reconstructs high-resolution wind fields on unstructured meshes from limited observations. The model employs a multiscale graph-based diffusion architecture trained on computational fluid dynamics (CFD) simulations and interprets classifier-free guidance as a learned posterior reconstruction mechanism: the unconditional branch learns a geometry-aware flow prior, while the sensor-conditioned branch injects observational constraints during sampling. This formulation enables obstacle-aware reconstruction and generalization to held-out mesh geometries, wind directions, and sensor configurations within the studied urban-flow setting, without retraining. We consider both sparse fixed sensors and trajectory-based observations using the same reconstruction procedure. When evaluated against supervised graph neural network (GNN) baselines and classical reduced-order data assimilation methods, GenDA reduces the relative root-mean-square error (RRMSE) by 25-57% and increases the structural similarity index (SSIM) by 23-33% across the tested meshes. Experiments are conducted on Reynolds-averaged Navier-Stokes (RANS) simulations of a real urban neighborhood in Bristol, United Kingdom, at a characteristic Reynolds number of $\mathrm{Re}\approx2\times10^{7}$, featuring complex building geometry and irregular terrain. The proposed framework provides a scalable path toward generative, geometry-aware data assimilation for environmental monitoring in complex domains.

Urban wind patterns affect air quality, heat exposure, pollutant dispersion, and pedestrian comfort, but measuring the full wind field across a city is difficult because sensors are sparse and detailed simulations are expensive. This paper introduces GenDA, a machine-learning method that reconstructs detailed urban wind fields from limited sensor measurements on complex city geometries. GenDA uses a generative diffusion model on unstructured meshes, allowing it to produce physically plausible wind reconstructions that are consistent with the available observations. In experiments on simulated wind flow over a real urban neighborhood, GenDA outperforms graph neural networks, reduced-order data assimilation methods, and other diffusion-based baselines, especially when very few sensors are available. This approach could support future environmental monitoring and urban-planning tools by making high-resolution wind reconstruction more practical from limited measurements.