Hectometric urban climate modelling over a tropical city

Abstract Hectometric (100 m) numerical weather prediction (NWP) models have the potential to improve our understanding of urban‐scale processes. Such high‐resolution models may provide valuable information for sectors such as energy, transportation, and emergency management, which rely on accurate, localized data for decision‐making. This study explores the development and evaluation of a 100‐m version of urban ‘SINGV’, a kilometre‐resolution model currently used by the Meteorological Service of Singapore (MSS) for operational NWP. The model incorporates updated datasets for land use/land cover, urban morphology, and anthropogenic heat. To assess model performance, we conducted 12 clear‐sky day simulations and compared the results with data from 25 observation locations, sourced from MSS's automatic weather stations, local HOBO stations, and an urban flux tower. We compared the 100‐m model results with those from a 300‐ m resolution model, using performance metrics such as Kling–Gupta efficiency, root mean‐squared error (RMSE), and mean absolute error (MAE) for variables including screen‐level temperature, specific humidity, 10‐m wind speed, radiation components, and turbulent fluxes. Since the sensor footprint is larger than the model grid at 100‐m resolution, we used wind direction‐sensitive evaluation for turbulent fluxes. The results indicate that the 100‐m resolution model performs well, with similar or improved accuracy, for example, showing an 8.75% and 9.32% reduction in RMSE and MAE, respectively for screen‐level air temperature, and a 10% RMSE reduction for incoming short‐wave radiation, compared to the 300‐m model. A major benefit of 100‐m simulations is their ability to provide nine times more spatial information with similar overall performance compared to 300‐m simulations. The inclusion of a wind direction‐sensitive adjustment for turbulent fluxes indicates the need for such adjustments when the sensor footprint exceeds model resolution. This study demonstrates the potential of hectometric NWP models and highlights the importance of sensor footprint when evaluating turbulent fluxes. Concurrent work investigates boundary layer dynamics and convective precipitation.