Deployment and evaluation of a low-cost sensor network based on spatial representativeness analysis

Air quality monitoring networks are essential for population exposure estimation. Official monitoring stations or (to a lessen extent) some low – cost sensors (LCS) are expensive, and generally only a few of them can be deployed by public administrations. The Spatial Representativeness Area (SRA) of a station defines the area for which the station is representative and is characterized by the area in which the annual mean of a specific pollutant is within a range of ±15% of the location concentration value. Maximizing the SRA would minimize the number of sensors needed to characterize air quality, and hence would contribute to reducing the associated cost while optimising the network. In this work, we present the deployment of 40 air quality LCS in Bilbao, a medium-sized city located in the north of Spain. We used the high resolution GRAL dispersion model results to locate and highlight the areas where NOx values maximize the SRA in each neighbourhood. These areas are then used to find the most appropriate locations for each sensor. Additionally, to represent potential hotspots we also characterised the areas where NOx concentration exceed the neighbourhood 90^th percentile. Lastly, model results are further used as a starting point for spatial representativeness discussions considering sub annual variations and specific meteorological situations.