Performance validation and calibration conditions for novel dynamic baseline tracking air sensors in long-term field monitoring

Abstract. The rapid expansion of low-cost sensor networks for air quality monitoring necessitates rigorous calibration to ensure data accuracy. Despite numerous published field calibration studies, a universal and comprehensive assessment of factors affecting sensor calibration remains elusive, leading to potential discrepancies in data quality across different networks. This study deployed eight sensor-based monitors in strategically chosen locations continuously for 2 years in Hong Kong, Macau and Shanghai. These locations covered a wide range of climatic conditions: Hong Kong's subtropical climate, Macau's similar yet distinct urban environment and Shanghai's more variable climate. Each monitor employed a dynamic baseline tracking method for the gas sensors, which isolates the concentration signals from temperature and humidity effects, enhancing the sensors' accuracy and reliability. This strategic deployment ensured that the sensors' performance and calibration processes were tested across diverse atmospheric conditions. The tests, which involved evaluating the validation performance by analysing randomly selected calibration sample subsets ranging from 1–15 d, indicated that the length of the calibration period, pollutant concentration range and time-averaging period are pivotal for sensor calibration quality. We determined that a 5–7 d calibration period minimizes calibration coefficient errors, and a wider concentration range improves the validation R2 values for all sensors, suggesting the necessity of setting specific concentration range thresholds. A time-averaging period of at least 5 min for data with 1 min resolution was recommended to enable optimal calibration in field operation. This study emphasizes the need for a comprehensive calibration assessment and the importance of considering environmental variability in the sensor calibration condition. These findings offer methodological guidance for the calibration of other sensor types, providing a reference for future research in the field of sensor calibration.