Data driven analysis of urban odour perception : geospatial insights from Hong Kong

Abstract

Urban odours play a crucial role in shaping the development and liveability of urban environments. While existing research has focused on classifying and evaluating urban odours, there is a relative dearth of studies that examine odour factors from human perception. This study seeks to fill this gap by investigating the spatial distribution of urban odour perception reported by people and evaluating the correlation between these perception and various urban environment factors, including social, natural, and built factors. Using social media and multi-source spatial data, our research innovatively utilizes olfactory-related sentiment analysis to map odour perceptions, integrates a multi-dimensional environmental framework underexplored in odour research, and constructs a data-driven pipeline employing a three-step approach: firstly, keyword extraction and text sentiment analysis are utilized to map the spatial distribution of odour perception; secondly, global and local regression analysis is employed to explore the relationships between odour perception scores (OPS) and multidimensional factors of the urban environment. Thirdly, utilizing hierarchical clustering to group regions that exhibit consistency in odour perception and its influencing factors. The findings reveal distinct spatial patterns in odour perception, with significant associations to factors like population density, air pollutant concentration, and building density. Spatial clustering further identifies regions with consistent odour perception profiles, highlighting key environmental drivers that shape these patterns. These results provide a foundation for understanding the spatial distribution of odour perception and offer actionable insights for urban planning and governance. This study not only enhances the understanding of urban odours but also provides a data-driven framework for addressing perception-related urban challenges in future research.