Urban Air Quality\nManagement at Low Cost Using Micro\nAir Sensors: A Case Study from Accra, Ghana

Urban air quality management is dependent on the availability\nof\nlocal air pollution data. In many major urban centers of Africa, there\nis limited to nonexistent information on air quality. This is gradually\nchanging in part due to the increasing use of micro air sensors, which\nhave the potential to enable the generation of ground-based air quality\ndata at fine scales for understanding local emission trends. Regional\nliterature on the application of high-resolution data for emission\nsource identification in this region is limited. In this study a micro\nair sensor was colocated at the Physics Department, University of\nGhana, with a reference grade instrument to evaluate its performance\nfor estimating PM<sub>2.5</sub> pollution accurately at fine scales\nand the value of these data in identification of local sources and\ntheir behavior over time. For this study, 15 weeks of data at hourly\nresolution with approximately 2500 data pairs were generated and analyzed\n(June 1, 2023, to September 15, 2023). For this time period a coefficient\nof determination (<i>r</i><sup>2</sup>) of 0.83 was generated\nwith a mean absolute error (MAE) of 5.44 μg m<sup>–3</sup> between the pre local calibration micro air sensor (i.e., out of\nthe box) and the reference-grade instrument. Following currently accepted\nbest practice methods (see, e.g., PAS4023) a domain specific (i.e.,\nlocal) calibration factor was generated using a multilinear regression\nmodel, and when this factor is applied to the micro air sensor data,\na reduction, i.e. improvement, in MAE to 1.43 μg m<sup>–3</sup> was found. Daily variation was calculated, a receptor model was\napplied, and time series plots as a function of wind direction were\ngenerated, including PM<sub>2.5</sub>/PM<sub>10</sub> ratio scatter\nand count plots, to explore the utility of this observational approach\nfor local source identification. The 3 data sets were compared (out\nof the box, domain calibrated, and reference-grade) and it was found\nthat although there were variations in the data reported, source areas\nhighlighted based on these data were similar, with input from local\nsources such as traffic emissions and biomass burning. As the temporal\nresolution of observational data associated with these micro air sensors\nis higher than for reference grade instruments (primarily due to costs\nand logistics limitations), they have the potential to provide insight\ninto the complex, often hyperlocalized sources associated with urban\nareas, such as those found in major African cities.