Dynamic Cross-Sensor Calibration for Low-Cost Particulate Matter Sensors
Abstract. Low-cost particulate matter (PM) sensors are increasingly used for air-quality and occupational exposure monitoring, but their mass-concentration response can vary with aerosol size distribution and combustion source. This variation limits measurement accuracy in environments where particle properties change over time and space, including wildfire smoke settings. We developed a real-time dynamic calibration method for a dual-sensor platform combining an Analog Devices ADPD188BI blue-light photometric sensor with a Sensirion SPS30 low-cost optical particle counter to improve PM1.0 mass-concentration estimates. Nebulized monodisperse polystyrene latex spheres with diameters of 0.49, 0.60, 0.72, 0.83, and 0.88 µm were measured with a Lasair 1003 optical particle counter to characterize the particle-size-dependent responses of both sensors. The SPS30 PM1.0 output and ADPD188BI blue-LED signal exhibited distinct size-dependent responses, providing the basis for adjustment using paired sensor measurements. We developed a ratio-based framework in which the instantaneous ADPD188BI blue-LED-to-SPS30 response ratio was used to estimate a PM1.0 mass-conversion coefficient. The method was evaluated against gravimetric PM1.0 measurements from separate combustion experiments involving pine and oak under both smoldering and flaming conditions, and bamboo and candles under smoldering conditions only. Both sensor outputs were strongly linearly related to gravimetric PM1.0 within individual combustion conditions (SPS30, R2 = 0.992–0.999; ADPD188BI blue-LED, R2 = 0.990–0.998), although regression slopes varied across aerosol sources and conditions. Applying the dynamic adjustment improved agreement with gravimetric PM1.0 across the tested combustion aerosols, reducing the pooled root mean square error (RMSE) by 84.5 %. These findings support the use of paired low-cost optical sensors to adjust PM1.0 mass response in changing smoke environments. Further evaluation is needed in wearable and field-deployed systems for wildland firefighter exposure monitoring, where routine use of reference instruments is often impractical.