Low-cost air quality sensors (LCS) provide high-resolution monitoring with affordability and real-time data but require performance evaluation and calibration due to probable lower accuracy compared to reference instruments and sensitivity to meteorological factors such as air relative humidity and temperature. Although most studies on low-cost sensors have been conducted in Europe and America, we did not find any study in regions like Iran, where air pollution levels and meteorological conditions differ significantly. In this study, performance evaluation and calibration of a low-cost particulate matter sensor were performed based on Federal Reference Method (FRM) and Federal Equivalent Method (FEM) in Tehran, from May 27, 2023, to August 19, 2024. Simple Linear Regression (SLR) and Multivariable Linear Regression (MLR) models were applied, validated via leave-one-out cross-validation (LOOCV). Results showed that LCS underestimated PM2.5 and PM10 concentrations in both warm and cold periods, with more pronounced and variable underestimation during warm months and more systematic, smaller underestimation in cold months. Underestimation was greater for PM10 than PM2.5 in both periods. Despite this, LCS measurements tracked reference instrument trends and responded well to concentration variations. Seasonal calibration was performed in addition to calibration for the entire sampling period due to varying LCS performance under different meteorological conditions. After SLR calibration model was applied the performance criteria were met only during the cold period for both PM2.5 and PM10. Applying the MLR calibration model improved accuracy, achieving acceptable performance for PM2.5 throughout the entire sampling period and for PM10 across the entire period and warm months, with R² values of 0.62–0.93 and normalized root mean square error (NRMSE) of 10.7–24.36%. This improvement resulted from incorporating air relative humidity, temperature, PM1, and PM2.5. Nevertheless, the performance of the sensor for PM2.5 in the warm season proved satisfactory after calibration based only on the FEM. Overall, this study revealed notable variations in sensor performance under different meteorological conditions, emphasizing the need for seasonal calibration. The results show that, when LCS are properly calibrated using site-specific and multivariable regression models, they can serve as a reliable tool for high-resolution air quality monitoring. These findings support the broader adoption of LCS in national and local monitoring networks as well as in individual exposure assessment for epidemiological studies.

Long-term performance evaluation and seasonal field calibration of a low-cost particulate matter sensor

期刊论文
发表日期：

2026年7月17日

查看全文

Farzaneh Gharibzadeh,
Mohammad Sadegh Hassanvand,
Ramin Nabizadeh,
Kazem Naddafi,
Masud Yunesian

摘要

Low-cost air quality sensors (LCS) provide high-resolution monitoring with affordability and real-time data but require performance evaluation and calibration due to probable lower accuracy compared to reference instruments and sensitivity to meteorological factors such as air relative humidity and temperature. Although most studies on low-cost sensors have been conducted in Europe and America, we did not find any study in regions like Iran, where air pollution levels and meteorological conditions differ significantly. In this study, performance evaluation and calibration of a low-cost particulate matter sensor were performed based on Federal Reference Method (FRM) and Federal Equivalent Method (FEM) in Tehran, from May 27, 2023, to August 19, 2024. Simple Linear Regression (SLR) and Multivariable Linear Regression (MLR) models were applied, validated via leave-one-out cross-validation (LOOCV). Results showed that LCS underestimated PM_2.5 and PM₁₀ concentrations in both warm and cold periods, with more pronounced and variable underestimation during warm months and more systematic, smaller underestimation in cold months. Underestimation was greater for PM₁₀ than PM_2.5 in both periods. Despite this, LCS measurements tracked reference instrument trends and responded well to concentration variations. Seasonal calibration was performed in addition to calibration for the entire sampling period due to varying LCS performance under different meteorological conditions. After SLR calibration model was applied the performance criteria were met only during the cold period for both PM_2.5 and PM₁₀. Applying the MLR calibration model improved accuracy, achieving acceptable performance for PM_2.5 throughout the entire sampling period and for PM₁₀ across the entire period and warm months, with R² values of 0.62–0.93 and normalized root mean square error (NRMSE) of 10.7–24.36%. This improvement resulted from incorporating air relative humidity, temperature, PM₁, and PM_2.5. Nevertheless, the performance of the sensor for PM_2.5 in the warm season proved satisfactory after calibration based only on the FEM. Overall, this study revealed notable variations in sensor performance under different meteorological conditions, emphasizing the need for seasonal calibration. The results show that, when LCS are properly calibrated using site-specific and multivariable regression models, they can serve as a reliable tool for high-resolution air quality monitoring. These findings support the broader adoption of LCS in national and local monitoring networks as well as in individual exposure assessment for epidemiological studies.