Diurnal evolution and spatial extent of urban park cooling from large-eddy simulations in Villa Ada (Rome)

Approximately 57% of the global population lives in urban areas, and this proportion is expected to increase in the coming decades. Consequently, cities are where a large and growing share of the population experiences the impacts of climate change. Urban environments can also intensify local warming because of their morphology and physical characteristics, including building density, anthropogenic heat emissions, impervious surfaces, and the thermal inertia of urban materials. Urban green spaces, such as parks, can partly offset these effects through shading, improving human thermal comfort. However, the thermal effect of urban parks is commonly assessed using satellite-derived surface temperature, which provides only instantaneous observations and does not capture its temporal evolution.

In this study, we propose an alternative method to evaluate the thermal effect of urban parks using large-eddy simulations (LES) with the PALM model system. The analysis focuses on Villa Ada, a 160 ha urban park in Rome, during a heatwave event in October 2023. Two nested simulation domains, with isotropic grid resolutions of 20 m and 10 m, are forced by boundary conditions from the MOLOCH mesoscale model. The simulations are used to investigate the diurnal variability of surface temperature and 2 m air temperature. The park thermal effect is quantified using a buffer-based approach with concentric rings extending up to 600 m from the park boundary, enabling the analysis of horizontal temperature gradients and their temporal evolution throughout the day. The framework also enables comparison among three land-cover scenarios: the current park configuration (baseline), a fully tree-covered configuration, and a short-grass configuration.

Results reveal a marked diurnal cycle in the park thermal effect, with lower magnitudes in the morning, a distinct maximum in the early afternoon, and a gradual decrease toward the evening. The dense-tree scenario shows the strongest cooling effect, with values reaching up to 8.5 K for surface temperature and 6.1 K for 2 m air temperature, whereas the baseline scenario also produces cooling, though of slightly lower intensity. By contrast, the short-grass scenario leads to higher surface temperatures across all buffers, indicating a warming effect.

The spatial extent of the cooling signal remains relatively stable in the baseline and dense-tree scenarios, with values typically between 330 m and 380 m. In contrast, the short-grass scenario does not show a comparable cooling signal. The results also reveal a bimodal distribution of surface temperature in the park surroundings, reflecting the contrast between vegetated and built surfaces. These findings highlight that the park cooling effect is a dynamic process, strongly controlled by diurnal evolution and surface heterogeneity, and cannot be fully assessed by instantaneous satellite observations.