The University of Arizona's Air Pollution Sensor Network seeks to deploy low-cost air pollution sensors across campus in order to construct machine learning-based models that forecast and depict real-time air pollution levels across campus. The Campus Sustainability Fund initially supported the confirmation of potential sensor locations across campus, the calibration of 15 indoor and outdoor sensors known as PurpleAirs with Pima DEQ, and the use of 7 portable mobile sensors known as AirBeams for mobile sampling with the help of 6 students testing the PM2.5 levels across campus. To create concentration levels in places without monitors, machine learning algorithms (e.g., random forest) will be used in conjunction with geographic information system (GIS) variables and local meteorology as predictors. To identify and display pollution hotspots and trends, real-time mapping and visualization tools will be developed. Once processed and built into a data interactive portal, the data and visualizations will be published online so that community people may view and better understand air quality patterns and personal exposures.
The UA Air Pollution Sensor Network installed 15 outdoor sensors and 5 indoor sensors across campus to create a dashboard for air quality monitoring. They also recruited 6 students to carry the sensors around campus for 5 days to gather data on daily pollutant exposure for an average student. These results varied from 1 ug/m3 to 2200 ug/m3 of Particulate Matter 2.5 exposure per day.
