Climate sensitive designs for policy makers: How LES model resolution affects accuracy in capturing urban micro-scale weather during heatwaves

Climate sensitive designs have been implemented recently in science to fill the niche of developing scientific tools to help mitigating urban heat island effects. A model capable of identifying hot and cool spots and testing adaptation and mitigation strategies to form recommendations for policy makers is in high demand. We present a novel two-step validation approach using 1) absolute comparison and 2) space-time evaluation of model performance across resolutions and against observations ensemble-averaged for representative urban microclimate types. Two Large Eddy Simulation (LES) models with 5 m and 20 m resolution are evaluated against a 14-point measurement station network during an extreme heatwave in Germany. To show space time behavior relationship between model and measurements, multiresolution decomposition (MRD) was used to investigate air temperature, specific humidity, and wind speed across time scales ranging from 11.25 to 180 min. The MRD analysis revealed strong correlations between stations with similar microclimatic conditions. Increasing model resolution did not uniformly improve accuracy across all climate elements; wind speed showed the greatest benefit from higher resolution. Absolute comparisons between simulations and observations demonstrated well-represented diurnal cycles for 5 m wind speed, whereas scalar variables remained sensitive to land surface parameterization. The simulated specific humidity exhibited deviations from measurements and nighttime air temperature tended to be overestimated. Space-time behavior instead was generally well captured except for nighttime specific humidity. This study contributes to correctly identifying and quantifying urban heat effects during extremes and supports urban planners and decision-makers in evaluating modeling tools for spatial planning. • Observations from a multi-station measurement network yielding 7 climate elements are analyzed • Real-time LES simulation using 5 m and 20 m grid resolution are analyzed and compared against observations • Validation of absolute differences showed discrepancies for modeled nighttime air temperature and specific humidity • Spatiotemporal validation of modeled results using stochastic mrd shows good agreement with observations • Wind speed at high resolution ( ≤ 5 m), daytime humidity and temperature for 20 m and 5 m resolution were well represented.