A new calibration system for low-cost Sensor Network in air pollution monitoring

In recent years, low-cost air sensors designed to measure ambient particulate matters and trace gases have been changing the world of air quality monitoring. Although low-cost sensors have the capability of gathering high temporally and spatially-resolved air quality information, they tend to have more severe data quality issues compared to the conventional monitoring instruments. To enhance the data quality of low-cost sensors, different calibration methods have been developed. In this study, based on the laboratory and field data collected in mainland China, the performances of PM2.5, PM10, CO, NO2, SO2, and O-3 sensors were estimated. Then, the sensors were calibrated by different methods and their output data were compared to the reference data to determine the validity and performance of these calibration methods. The results indicated that 1) field calibration with supervised learning technique is an effective technique to decrease the error, and 2) mobile air quality monitoring stations can calibrate the sensors at where standard air quality monitoring stations are not available. In conclusion, a complete laboratory and field calibration system (four-stage calibration) covering the whole life cycle of the sensor is proposed in this paper.