Global and regional calibration of low-cost PM 2.5 sensors (PurpleAir): assessing the transferability of U.S.-based corrections worldwide

Low-cost optical particle sensors (LCS) can improve the spatio-temporal coverage of particle matter (PM) air pollution monitoring. These sensors require calibration to ensure accuracy across diverse aerosol and environmental conditions. Using a global dataset (January 2023–March 2025) of 1,013 PurpleAir sensors co-located (≤500 m) with reference grade monitors on six continents, we evaluated three widely used U.S. correction equations and developed global and regional calibration equations. Performance was assessed for 24-h and 1-h averages using R2, RMSE, nRMSE, precision (CV) and Air Quality Index (AQI) category correctness (24-h and NowCast EPA AQI). Raw PurpleAir readings overestimate PM2.5 in most regions and do not meet the EPA sensor performance targets. Previously suggested calibrations based on sensors located in the USA improve the accuracy but exhibit limited transferability outside North America. A data-driven global quadratic model improves results in some regions, such as Australia and New Zealand, but underperforms in other places. However, regional calibrations consistently reach the best agreement, meeting EPA’s 24-h performance thresholds in multiple regions and increasing EPA’s 24-h AQI correctness to ≥77%. These findings indicate that, given the strong RH sensitivity and aerosol composition variability, region-specific corrections should be preferred where co-locations exist. We provide the derived equations and policy-relevant suggestions to support more robust use of LCS in air-quality research, management, and public communication.