Related Experiment Video
Updated: Jun 26, 2025
10:29
Calibrated Passive Sampling - Multi-plot Field Measurements of NH3 Emissions with a Combination of Dynamic Tube Method and Passive Samplers
Published on: March 21, 2016
12.3K
Leveraging Temporal Information to Improve Machine Learning-Based Calibration Techniques for Low-Cost Air Quality
Sharafat Ali1, Fakhrul Alam2, Johan Potgieter3
1Department of Mechanical and Electrical Engineering, Massey University, Auckland 0632, New Zealand.
Sensors (Basel, Switzerland)
|May 11, 2024
Summary
Temporal information improves low-cost air quality sensor calibration. Utilizing deployment duration and time of day enhances accuracy for carbon monoxide (CO) and nitrogen dioxide (NO2) monitoring.
Area of Science:
- Environmental Science
- Sensor Technology
- Data Science
Background:
- Low-cost ambient sensors offer high spatio-temporal air pollution monitoring.
- These sensors require calibration due to lower accuracy compared to reference instruments.
Purpose of the Study:
- To improve the calibration of low-cost air pollutant sensors.
- To investigate the utility of temporal information (deployment duration, time of day) for sensor calibration.
Main Methods:
- Utilized three global datasets from co-located low-cost and reference sensors for CO and NO2.
- Applied machine learning models (Random Forest, LSTM) incorporating temporal features.
Main Results:
- Temporal information significantly enhances the accuracy of machine learning-based calibration.
- Demonstrated improved reliability for low-cost sensor data.
Conclusions:
- Temporal data is a valuable, underutilized covariate for low-cost sensor calibration.
- Enhanced calibration improves the potential of low-cost sensors for air quality monitoring.
