Impact of tree representation on urban climate modelling: Insights from ENVI-met simulations in Augsburg, Germany

The representation of urban trees in microclimate models such as ENVI-met has advanced significantly in recent years. While older tree models remain available, the newest approach uses Lindenmayer systems, enabling species-specific tree modelling, realistic heights, trunk structures, and leaf area density profiles (L-trees). However, the sensitivity of simulated air temperature, mean radiant temperature (MRT), and the Universal Thermal Climate Index (UTCI) to different tree representations and model configurations remains unclear. This study developed a high-resolution ENVI-met model of an urban park and its adjacent residential district in Augsburg, Germany, utilising a database of 4264 trees, including species, height, and trunk diameter. Sixteen scenarios were simulated to assess the influence of vegetation geometry (L-trees, grid-based 3D plants, simple plants), forcing method, terrain inclusion, spatial resolution, and season. Air temperature differed only slightly across scenarios, whereas MRT and UTCI were highly sensitive to vegetation representation. UTCI in L-tree scenarios was sensitive to species composition, tree height, forcing approach, and inclusion of terrain. Tree height and tree species were key drivers of thermal comfort, while spatial resolution had a minor influence. Grid-based trees sometimes yielded lower UTCI values, whereas simple plants failed to reproduce the cooling effects in the park. Across all scenarios, air temperature was consistently overestimated compared with measurements at nine sites. These findings underscore the importance of accurate data on tree locations, species, heights, and leaf area density profiles for reliable microclimate modelling, providing a sound basis for urban climate planning. • Sixteen scenarios modelled urban climate with 4000+ accurately mapped trees. • L-trees outperformed grid-based and simple plant models. • Tree species, height, and leaf-area density strongly influenced thermal comfort. • Models overestimated air temperature and underestimated urban heat gradients. • Spatial resolution had minor effects compared to tree representations.