Integrated screening techniques reveal insight into hyperlocal non-traffic emission sources

Distributed air quality monitoring networks using low-cost sensors (LCS) promise to empower policymakers and citizens to enhance air quality management by developing tailored interventions. However, relatively few studies have investigated how to extract information on emission sources from citywide, fixed-site LCS networks, especially in terms of non-traffic sources. Commonly measured NO₂ concentrations are mostly indicative of traffic sources and PM_2.5 concentrations usually exhibit low diurnal variability. In this paper, we present an innovative, scalable screening method to acquire hyperlocal insight into non-traffic emission sources. This method integrates network analysis and peak analysis. Network analysis leverages the statistical power of the sensor network to compare data at a monitoring location to its peers within the network to identify hotspots driven by local sources rather than regional or meteorology-driven events. Peak analysis resorts to clustering on concentration spikes above the background values, reducing the influence of high background concentrations and low diurnal variability and emphasizing the impact of local emission sources. We demonstrate the capability of the proposed integrated screening techniques in identifying the influence of construction and nighttime cooking activities in the Greater London area without prior site-specific knowledge.