October 31, 2019
Applications of Satellite and Low-cost Sensor Data in Estimating PM2.5 Concentrations
Prof. Yang Liu (Emory University)
2019년 11월 5일 (화) 16:00
Low-cost air quality sensors are promising supplements to regulatory monitors for PM2.5 exposure assessment. However, little has been done to develop a framework to incorporate low-cost sensor measurements in large-scale PM2.5 exposure modeling. Using the PurpleAir sensor network in California as an example, we developed a spatially varying calibration method for low-cost sensor measurements and a down-weighting strategy based on residual measurement errors to optimize the use of low-cost sensor data with regulatory monitoring data in PM2.5 modeling. Spatially varying calibration of PurpleAir against US EPA monitors was performed at the hourly level with Geographically Weighted Regression (GWR). The calibrated PurpleAir measurements were given lower weights according to their residual errors and then combined with AQS measurements into a Random Forest (RF) prediction model as a dependent variable to generate 1-km daily PM2.5 exposure estimates. The calibration reduced the systematic bias in PurpleAir measurements to ~0 μg/m³ and decreased their residual errors by 36%. The RF model with both AQS and down-weighted PurpleAir data outperformed the RF model based solely on AQS with an improved random CV R2 of 0.86, an improved spatial CV R2 of 0.81, and a lower prediction error. The inclusion of PurpleAir data allowed the prediction model to show more PM2.5 spatial details and better detect pollution hotspots. The proposed framework can be generalized to regions worldwide for PM2.5 exposure assessment and health effects research.