Source attribution of air pollution by spatial scale separation using high spatial density networks of low cost air quality sensors

To carry out detailed source attribution for air quality assessment it is necessary to distinguish pollutant contributions that arise from local emissions from those attributable to non-local or regional emission sources. Frequently this requires the use of complex models and inversion methods, prior knowledge or assumptions regarding the pollution environment. In this paper we demonstrate how high spatial density and fast response measurements from low-cost sensor networks may facilitate this separation. A purely measurement-based approach to extract underlying pollution levels (baselines) from the measurements is presented exploiting the different relative frequencies of local and background pollution variations. This paper shows that if high spatial and temporal coverage of air quality measurements are available, the different contributions to the total pollution levels, namely the regional signal as well as near and far field local sources, can be quantified. The advantage of using high spatial resolution observations, as can be provided by low-cost sensor networks, lies in the fact that no prior assumptions about pollution levels at individual deployment sites are required. The methodology we present here, utilising measurements of carbon monoxide (CO), has wide applicability, including additional gas phase species and measurements obtained using reference networks. While similar studies have been performed, this is the first study using networks at this density, or using low cost sensor networks.