Mapping local urban climate and ventilation corridors using clustering approach

The urban microclimate model, urbanMicroclimateFoam developed by the authors, is used to simulate thermal comfort in an urban park and adjacent neighborhood in tropical Singapore during a hot humid period. The thermal comfort is analyzed using the Universal Thermal Climate Index (UTCI) which depends on air temperature, relative humidity, wind speed and mean radiant temperature.

To better understand the spatial variability of urban microclimate, we apply clustering approach to classify different local climate zones using scaled values of comfort variables as input. In the case study, six clusters are identified, each representing areas with similar local climate characteristics. One key cluster corresponds to the zone shadowed by trees in the park, where UTCI is significantly lower due to tree coverage. However, unshaded zone in between the park trees is also represented in a distinct cluster, where UTCI are higher since they experience lower wind speed and higher relative humidity, due to the wind blocking and transpiration by trees. This effect, in contrast to the local shading by trees, is referred to as nonlocal heating effect by trees. Interestingly, the proposed approach identifies different clusters with different thermal conditions in between the building blocks, with wind speed emerging as the primary differentiating factor. This highlights the importance of wind on thermal comfort in hot-humid context.

Using a clustering approach with six variables as input, we were able to detect different wind corridors with a wind speed higher than average. Some of these clusters indicate hot and dry air, or cool and wet air ventilation corridors, which are not easily distinguishable using conventional methods and also unfavorable for thermal comfort improvement. One cluster indicates cool and dry air ventilation, which favors thermal comfort improvement.

In conclusion, clustering approach allows to map different urban microclimate patterns and analyze the underlying reasons for the observed thermal comfort.