What's That Smell?: Understanding Urban Smellscapes Using Crowd-sourced Data, Monitoring, and Modeling

Odor exposure is associated with adverse and inequitable impacts on health and quality of life. However, it is challenging to study the effects of odors due to variability in individual sensitivity and perception, weather and chemistry, and emissions of mixtures of odorous contaminants. The Smell Vancouver project combines multiple data sources and approaches to better understand urban odor pollution in Vancouver, Canada. We conduct quantitative and qualitative analyses of a 12-month data set from a web application collecting crowd-sourced odor reports (including time, location, and description), demographics, and descriptive self-reported impacts.

We find significant spatiotemporal patterns of odor counts characterized by hotspots as well as spatiotemporal outliers. These results highlight the influence of persistent odor sources (e.g., waste management, chemical industries) as well as transient events (e.g., accidents, wildfires). Multiple linear regression models with air quality and meteorological data suggest that meteorological ventilation and air pollutants account for 47-83% of the variability in daily odor counts across most months. Users report diverse odors, symptoms, actions taken, and potential causes (OSAC) with strong seasonality (Fig. 1) and spatial variability. Underlying patterns in the structure of the odor experience (e.g., chains of causes associated with odors, and those odors being further related to symptoms and actions) point to the potential utility of OSAC data for source-specific impacts characterization (Fig. 1). We also observe concurrent neurological, respiratory, and emotional disturbance symptoms in multiple months, highlighting the complexity of odor-related well-being impacts. The collection of descriptive text provides new insights into the odor experience of app users. For example, odors can trigger maladaptive behavior (e.g., "close windows on summer evening"), putting individuals in situations where they must choose amongst environmental stressors (cumulative impacts of heat and indoor air pollutants versus odor and outdoor air pollutants).

Results from this project are policy-relevant beyond Vancouver, and provide evidence that human-centered approaches could help mitigate negative impacts of odorous emissions on health and well-being.