Simulating an extreme heat event in a mid-sized city in Europe: validating and analyzing the relevance of spatial resolution in the urban LES model PALM4U with an observation network

Urban areas are known to be exposed to higher temperatures than rural areas making urban citizens particularly vulnerable to extreme heat events. Large eddy simulation (LES) models can simulate micrometeorological heat transport and mixing processes by directly resolving large-scale turbulence. These models can be used to simulate urban development strategies aiming at mitigating the adverse effects of heat waves in cities by analyzing their influence on urban microclimate. Despite their use in formulating recommendations for city planning, these models are often not validated with observed meteorological data. We here present results from conducting a model-observation comparison for a mid-size city in Germany. Model simulations were computed with the LES model PALM4U run at two different resolutions (Δx,y,z = 5 and 20 m) and evaluated against observations from a network of microweather stations for a heat wave in 2019 reaching maximum near-surface air temperatures of 37 °C.

During daytime, differences between observed and modeled near-surface air temperatures were small (-3.8 to 1.1 K, mean = 0.9 K), but much larger during night-time and the early morning transition. The latter findings can be explained by an overestimated modeled ground heat flux resupplying too much energy offsetting the radiative cooling leading to overestimated modeled air temperatures by up to +9 K (mean = 5.3 K). Further, results showed that in areas where the actual urban structure is reproduced well by the model resolution, differences between observed and modeled wind speeds were lower. Our findings indicate that a spatial resolution smaller than the mean building height produce more accurate model results for wind speeds. Many differences in model-observation intercomparison are explained by an overestimated modeled turbulent kinetic energy (TKE) causing inflated turbulent mixing in the air, which leads to distorted model output particularly for the urban nocturnal boundary layer.