Toxic air pollutants such as PM2.5 and NO 2 have been linked to various health implications, including respiratory diseases, cardiovascular diseases, and cancers. With recent updates from the World Health Organisation introducing even stricter daily and annual exposure limits, the expectations placed upon governing bodies to better manage urban air quality have increased. It is suggested in this research that air quality may be better managed through the use of smart city infrastructure and data-driven techniques to introduce dynamic policy making. The primary goal of this study is to evaluate how clustering methods can be used to identify spatiotemporal patterns in air quality data that support more effective, location-specific interventions. In this research, it is proposed that agglomerative hierarchical clustering with dynamic time warping can provide a solution to identifying areas of similar spatiotemporal patterns, leading to targeted policies for each cluster. Three geographic areas are explored, which are mainland Taiwan, Beijing and Newcastle upon Tyne. These areas drastically differ in spatial scale, offering insight to the effectiveness of the method at different areas. Five pollutants are considered, which are CO, NO 2 , O 3 , PM2.5 and PM10. The results show that clustering is well-achieved in large spatial areas, however, the effectiveness degrades as the spatial scale decreases, suggesting that targeted policies at the city level may be most appropriate. • Increasing spatial scale leads to more well-defined clusters. • Clustering at the city scale can produce overly high resolution, introducing noise. • Spatially distributed sensors produce more effective clusters than concentrated deployments.
Booth et al. (2026) studied this question.