Assessing urban microclimate modeling variability: A comparative analysis of ENVI-met and single-layer urban canopy model across multiple local climate zones in Hong Kong

As cities worldwide experience rapid growth, their diverse urban forms generate complex microclimatic patterns affecting thermal comfort, energy use, and human health, necessitating computational modeling for climate adaptation. This study compares the widely used ENVI-met and the Single-Layer Urban Canopy Model (SLUCM) across Local Climate Zones (LCZs) in subtropical Hong Kong. Both models show reasonable accuracy in simulating pedestrian-level air temperature (SLUCM mean MAE: 1.02 °C vs. ENVI-met 1.3 °C) and relative humidity (SLUCM 8.72% vs. ENVI-met 10.12%), with performance varying by urban morphology. Largest errors occur in dense high-rise (LCZ 1) and sparsely built (LCZ 9) areas for temperature, and vegetated zones (LCZ 4, 5, 6, 9) for humidity. ENVI-met produces more uniform vertical profiles than SLUCM, with summer convection enhancing mixing across LCZs, overriding site-specific features. Seasonal analysis reveals a substantial increase in temperature differences in winter (ΔT₂: 1.00 °C December vs. 0.11 °C July), driven by solar angle effects favoring ENVI-met's 3D resolution. Thermal comfort simulations (UTCI, PET) show ENVI-met outperforming SLUCM, with midday discrepancies of 2.4–2.7 °C, reflecting its strength in capturing radiative dynamics. We recommend SLUCM for high-density LCZs (1,2) and summer conditions for computational efficiency; ENVI-met for medium-density (LCZ 4, 5, 6) and low-density (LCZ 9) areas with heterogeneous vegetation or complex geometries, and winter conditions for shadowing; and parameter-specific guidance favoring ENVI-met for thermal comfort and SLUCM for average temperature and humidity patterns. • SLUCM and ENVI-met compared across diverse LCZs in Hong Kong. • Models show comparable accuracy with temperature errors peaking at density extremes and humidity errors in vegetated zones. • Solar radiation and turbulence drive simulation differences • Model selection guidelines provided based on LCZ type, season, vegetation, and meteorological parameters.