Measuring air quality with low-cost sensors in citizen science applications

Dissertation submitted in partial fulfilment of the requirements for the degree of Master of Science in Geospatial TechnologiesAir pollution is unquestionably a public health emergency, and the rates of pollution continue to rise at an alarming rate in cities all over the world. Nevertheless, the traditional monitoring equipment is very expensive, and the available measurements are not sufficient to precisely classify air quality in several locations in a city. Recent advancements in air quality measuring technology provide a potential opportunity to increase the air quality data, and to raise public awareness of health issues arising from air pollution. This study focuses on the development and evaluation of a new prototype for the monitoring of fine particulate matter (PM2.5). It describes the design approach and the evaluation methods, in which a series of field experiments were conducted to evaluate the performance of the prototype and of a commercial low-cost device in comparison with a reference monitor. The results showed that the prototype presented a good performance in environments with a high variation of particle concentrations (variations above 100μg/m³), such as cooking-environments and exposure to cigarette smoke, for most of the experiments (R² = 0.55-0.85). However, their agreement was very poor in environments without high variability of particle concentrations. The performance comparison between identical sensors purchased in the same year revealed a very high agreement (R² = 0.92), but prototypes which utilized sensors acquired in different years presented a very weak correlation in most of the experiments. The analysis of the commercial low-cost device’s performance revealed a moderate to strong linear correlation with the reference monitor in all the experiments (R= 0.51-0.93); this study also demonstrates that the maximum limit of detection of the device was much lower than the value given by the manufacturer (approximately 180μg/m³, in contrast to the value of 400μg/m³). For applications of real-time measurements, the prototype developed in this research may be especially utilized as indicative of PM2.5 hotspots and trends in ambient conditions, primarily in residences, monitoring the frequency and duration of high exposure events, such as cooking, smoking, and biomass burning. Nevertheless, this research demonstrates the necessity for individual sensor performance testing prior to field use, and that presumptions about the representativeness of measurements of PM2.5 carried out by low-cost sensors should be made with caution