How Critical Are Urban Canopy Parameters in a State-of-the-Art Coupled Bulk Urban Canopy Model?

In a warming world, accurately modeling urban climate is essential for sustainable urban planning. Bulk urban canopy models, such as TERRA_URB coupled to the icosahedral non-hydrostatic weather and climate model (ICON), rely on urban canopy parameters (UCPs) to capture the surface–atmosphere interaction in cities. Consequently, model performance strongly depends on the quality of these UCP inputs to realistically represent urban boundary layer processes - such as boundary layer height, wind drag, heat island effects, and greenhouse gas concentrations. UCPs can be prescribed uniformly across all urban grid cells or vary according to local climate zone (LCZ) classification or real-world datasets. While LCZ-based approaches provide a globally consistent categorization of urban environments, they may overlook the fine-scale heterogeneity within individual cities. In this study, we examine whether using high-resolution urban characteristics, derived from 3D building geometries, satellite data, and other local information, improves urban climate simulations in the state-of-the-art ICON TERRA_URB model compared to LCZ-based UCPs. Focusing on the Swiss cities of Zurich and Basel, we refine UCPs by incorporating local datasets that capture actual building heights, densities, and solar-thermal properties, then compare simulations from the summer of 2023 against observational data. Our results highlight that activating TERRA_URB significantly enhances model accuracy for nighttime temperatures in regions characterized by high artificial surface fractions (ASF). Among the tested configurations, employing realistic UCP data provides a slight advantage over uniform and LCZ-based alternatives. However, these benefits observed in temperature do not extend to wind speed, where no single scenario demonstrates a clear overall advantage. Moreover, neither LCZ-based TERRA_URB nor realistic TERRA_URB proves sufficient in less densely urbanized contexts (low ASF), suggesting that disabling TERRA_URB - or opting for simpler model approaches - may be more suitable where urban influences are small.