Isolating urban form impacts on spatiotemporal distribution of surface meteorology in coastal cities during extreme heat events

• Coastal cities show distinct temperature and humidity gradients, altering UHI. • Gridded UCPs enable control experiments isolating climatic impacts of urban form. • The independent climatic impact of diverse urban form metrics is quantified. • Empirical urban form impacts aid rapid prediction of future thermal environments. Coastal cities in tropical and temperate zones are prone to heat waves, and the interplay between moist air and urban heat islands exacerbates the risk of heatstroke. However, the complexities introduced by land–sea temperature and humidity gradients pose challenges in quantifying the influence of urban forms on thermal environments. Thus, we conducted simulations using the Weather Research and Forecasting model for 49 high-temperature days across 2018–2021 in Tokyo—a developed coastal city. Control experiments were designed to eliminate the land–sea gradients and isolate the independent effects of urban form on the spatial fluctuation of pedestrian-level air temperature (ΔTa_u) and specific humidity (ΔSH_u). The findings reveal that the correlation between urban form and ΔTa_u is stronger than that with ΔSH_u. A positive linear relationship exists with building density at midday ( R 2 = 0.72) and building surface area at night ( R 2 = 0.78). In contrast, ΔSH_u exhibits a negative linear relationship at midday ( R 2 = 0.60) and a weak correlation with urban form at night ( R 2 < 0.15). Notably, ΔTa_u and ΔSH_u displayed clear diurnal variations, with the most significant spatial dispersion observed at midday (−1.3 to 2.2 K and −0.8 to 0.4 g/kg, respectively). Conversely, as the spatial patterns of ΔTa_u and ΔSH_u exhibited minimal fluctuations across varying high-temperature days, their historical averages effectively represented their general distribution. Furthermore, a preliminary method is proposed, utilizing a simple meteorological model along with empirical ΔTa_u and ΔSH_u for the rapid prediction of thermal environments during future heat events.