Calibration Methods for Low-Cost Particulate Matter Sensors Considering Seasonal Variability.

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Sensor Technology

Background:

• Low-cost sensors are crucial for high-resolution monitoring of particulate matter (PM2.5 and PM10) for public health management.
• Existing calibration methods for these sensors struggle to accurately account for seasonal variations in PM concentration.

Purpose of the Study:

• To develop an improved calibration methodology for low-cost particulate matter sensors.
• To enhance the accuracy of PM sensor readings by incorporating meridian altitude to better represent seasonal variability.

Main Methods:

• Utilized meridian altitude as a novel variable for calibrating seasonal variations in PM concentration.
• Applied feedforward neural networks, support vector machines, generalized additive models, and stepwise linear regression for model validation.
• Treated calibrated PM2.5 as a subset of PM10 for PM10 calibration.

Main Results:

• Inclusion of meridian altitude significantly improved the accuracy and explanatory power of PM calibration models.
• For PM2.5, a combination of relative humidity, temperature, and meridian altitude achieved an R² of 0.93 and RMSE of 5.6 µg/m³.
• For PM10, calibration with meridian altitude reduced mean absolute percentage error from 27.41% to 18.55%, further dropping to 15.35% with calibrated PM2.5 inclusion.

Conclusions:

• Meridian altitude is an effective variable for improving seasonal calibration of low-cost PM sensors.
• The proposed methodology enhances the reliability of air quality data derived from low-cost sensor networks.
• Accurate PM monitoring through improved sensor calibration supports better public health strategies.