Abstract
<jats:p>Urban weather/climate extremes arise from interactions between large-scale atmospheric circulation and heterogeneous urban land-atmosphere processes, yet existing models rarely represent both within a unified framework. Regional models provide detailed urban physics but rely on lateral boundary conditions, whereas global models generally lack explicit urban canopy representations and sufficient spatial resolution for city-scale applications. Here we develop MPAS&ndash;Urban, the first urban-capable implementation of the Model for Prediction Across Scales (MPAS), by coupling a single-layer urban canopy model with the Noah-MP land model on a variable-resolution global mesh from 30 km globally to 500 m over Hong Kong. This framework incorporates a global 100-m Local Climate Zone dataset to represent detailed spatially-varying urban classifications. The system is evaluated for record-breaking 2022 Hong Kong heatwave and compared with the default bulk urban representation. MPAS-Urban reproduces the large-scale atmospheric circulation associated with the event, including the Northwest Pacific Subtropical High, and captures regional distributions of near-surface temperature, humidity, and wind. At the local scale, the explicit urban canopy treatment enhances urban heat storage and aerodynamic drag, producing more realistic near-surface air temperature and substantially improved urban wind speed especially in high-density urban areas. Surface flux observations show that these improvements arise from more realistic partitioning of energy and momentum exchange within the urban canopy, although daytime humidity remains underestimated. MPAS&ndash;Urban enables seamless cross-scale modeling of interactions between synoptic forcing and local urban processes, establishing a foundation for next-generation urban weather prediction and climate assessment.</jats:p>