The Field Evaluation of Innovative PM2.5 Monitoring Using Air Sensor Co-location with Regulatory Equipment

Abstract

This study aims to evaluate the performance of PM_2.5 air sensors using the light scattering technique in comparison to the Federal Equivalent Method (FEM) operated by the Pollution Control Department, during an extended period of operation, from June 2020 to December 2021. Pearson’s correlation reveals good agreement among the sensor units (0.70 – 0.98) and data completeness above 90%. However, a notable issue arises with the tendency to report extremely high concentrations (445 – 564 µg/m³) compared to FEM (235 µg/m³). Regression analysis shows substantial biases in the sensor measurements with the slope values ranging 0.8 - 1.3, and low to moderate fit (R2 0.18 - 0.58) for the hourly concentration dataset. The regression performs significantly better with the rolling 24-hour average values (R2 0.47 - 0.91). The trend lines of the offset values showed general agreement among the sensor units for most of the study period, except for the dry season where the deviation increased from each other and from the FEM, suggesting possible influence from the weather. This paper shows that the sensors are suitable for moderate accuracy requirements. However, for public warning purposes, there are potential areas of improvement, the first priority being the report of extreme values, which may trigger false alarms and cause unnecessary panic among the public.