WRF-to-OpenFOAM Downscaling for Urban Wind Fields

Kilometre-scale weather models represent cities as roughness, not as individual buildings. For low-altitude UAV corridors that omission matters: street-scale shear and building wakes sit inside the same hour of mesoscale weather, and a single grid-cell wind cannot describe them.

This project builds an offline, one-way WRF–OpenFOAM pipeline over downtown Guangzhou. A kilometre-scale nested WRF run supplies hourly inflow; a building-resolving RANS solver (steady k–ε, one snapshot per hour) downscales the canopy-layer wind. The evaluation window is September 2025, covering about two weeks that include a pre-typhoon regular period and one typhoon-affected case. Winds are compared with multi-station urban Doppler LiDAR profiles at several morphologically different urban sites.

Findings

The added value of building-resolving downscaling is height-dependent. Below about 300 m, OpenFOAM reshapes the mesoscale wind toward the LiDAR profiles; above that layer, WRF and the downscaled field converge. How much a given site improves still depends on local urban form. Along representative low-altitude paths, an open corridor and a dense-built corridor that look similar in WRF show different wind exposure at typical UAV flight heights.

Limits and next steps

Because the nesting is one-way, microscale skill remains bounded by the driving WRF field. The physics-based archive is also the basis for faster, data-driven surrogates aimed at operational urban wind products.