Improving Low-Cost Air Quality Monitoring: Developing User-Friendly Visualization Tools for Factor and Anomaly Identification, and Implementing Calibration Algorithms

Abstract

Air pollution is a significant global issue with serious toxicological impacts on human health, causing millions of deaths annually according to the World Health Organization. Public awareness of the adverse health effects of air pollution is crucial. Citizen science projects on air quality have the potential to increase public awareness and facilitate the assessment of air pollution. However, the cost and logistics of traditional monitoring approaches have significantly constrained indoor and outdoor air quality monitoring efforts. Developing low-cost air quality monitoring devices using inexpensive sensors that deliver reliable data can help engage citizens in raising awareness about air quality.

Leveraging new miniaturized, low-cost air quality sensors and computational methods, this research introduces Pearson Correlation Coefficient heatmaps as an effective tool for visualizing and identifying factors influencing low-cost air quality sensor performance. These factors include environmental conditions (temperature and humidity), cross-sensitive pollutants, and sensor aging. Insights from these heatmaps led to the development of new compensation algorithms based on multiple linear regression, effectively mitigating the impact of environmental factors, degradation, and pollutant interference on sensor readings.

The compensation algorithms significantly enhance the accuracy of interpreting raw sensor data, aligning it more closely with reference measurements. This work advances the field of low-cost air quality monitoring, improving the usefulness of such devices for research, local community monitoring, and citizen science projects.

Additionally, Experiments were designed to calibrate low-cost air quality sensors against governmental references in both hot and cold seasons. Calibration during the cold season showed excellent agreement for NO2 sensors with references (correlation coefficient R around 0.80). Good agreement (R around 0.70) between NO2 and PM2.5 with reference sensors was achieved in the warm season, making the sensors reliable for awareness-raising citizen science projects.

This project engaged around 30 people in five workshops, three device-making work- shops, and two data engagement workshops. Local community members built air quality monitoring devices through two workshops and deployed them in eight lo- cations in the Easton area of Bristol. These devices operated continuously from February to March 2022, collecting real-time data for NO2, O3, and PM, which owners could access via a Google spreadsheet. Data from three devices were used in additional workshops to enhance air quality literacy among residents. The work- shops significantly increased participants’ knowledge of pollution sources and their willingness to discuss air pollution risks with others.

Date of Award	23 Jul 2025
Original language	English
Awarding Institution	University of Bath
Supervisor	Julian Padget (Supervisor) & Nick McCullen (Supervisor)