Publications·October 31, 2025
This manual provides comprehensive technical and operational guidance for implementing Hybrid Air Quality Monitoring Networks (HAQMN) across the EANET (Acid Deposition Monitoring Network in East Asia) region. It responds to the growing interest in low‑cost sensors (LCSs) as complementary tools to reference-grade monitors (RGMs) and aims to standardize their use for broader, more cost-effective air quality monitoring.
The report begins by highlighting the challenges associated with traditional monitoring systems, such as high infrastructure costs, technical complexity, and limited spatial coverage. Low-cost sensors—compact, energy-efficient devices—address these issues by enabling flexible deployment in diverse environments, including areas with unstable electricity supply or limited technical capacity. However, because LCSs have lower accuracy and durability compared to RGMs, the manual provides a rigorous system of quality assurance/quality control (QA/QC), calibration methods, and maintenance procedures to ensure data reliability.
A major contribution of the manual is its tiered classification system, adapted from the U.S. EPA and European CEN frameworks. The system defines three Tiers (I–III) based on intended monitoring purpose and required accuracy. Tier I sensors may be used for public awareness or detecting spatial trends, while Tier III sensors are suitable for research and regulatory support. The report details performance benchmarks such as Mean Normalized Bias (MNB), Coefficient of Determination (R²), RMSE, and slope/intercept requirements for high‑quality monitoring.
The manual then presents detailed guidelines on network design, including LCS assembly, sensor characteristics, limits of detection, and technical considerations such as data storage (local, cloud, blockchain), communication protocols (Wi‑Fi, cellular, Bluetooth), and sampling frequency. Examples from India, Greece, and Siberia illustrate how LCS networks have been deployed globally for hotspot detection, satellite validation, and tracking pollution movement.
A significant portion of the document is dedicated to site selection and installation practices, emphasizing environmental representativeness, airflow conditions, distance from emission sources, and elevation of sensor inlets. The manual stresses the importance of avoiding obstructed settings, ensuring secure installation (e.g., locking enclosures, fencing), and protecting sensors from heat, rain, lightning, and animals. Illustrations demonstrate placement on rooftops, poles, and building walls.
The operational section foregrounds the parallel monitoring test (co‑location) process, which is essential for calibrating LCSs by comparing them against RGMs for at least one week. The manual provides mathematical formulas for slope, intercept, R², MNB, and RMSE to evaluate sensor performance. It outlines conditions that trigger recalibration or replacement, such as declining R², sensor drift, or spike anomalies. These procedures help ensure that LCS data aligns with reference-grade measurements.
Extensive guidance is also provided for maintenance, including routine physical checks (housing integrity, air inlet cleanliness, fan operation), data inspection (missing timestamps, spikes, flatlines), and troubleshooting of communication failures, hardware issues, environmental interference, and data corruption. Photos and graphs illustrate common LCS malfunctions and data anomalies.
A structured data screening framework is presented, combining physical range filters, statistical anomaly detection (via median absolute deviation, rolling windows), flatline detection, rate-of-change checks, and cross-sensor consistency tests. These methods help differentiate between true pollution events and erroneous data caused by sensor malfunction or environmental artifacts.
The manual also explains data correction techniques, introducing simple linear regression (SLR), multivariate linear regression (MLR), machine learning models, and transfer learning methods. It presents recommended procedures for building correction models based on co-location results and validating corrected data through time-series and scatterplot comparisons. Visual examples illustrate substantial improvements in accuracy and precision after correction.
The final chapters highlight data visualization tools and strategies, including maps, heatmaps, time-series charts, wind roses, and pollution source direction analyses. Recommended software includes Excel, R, Python, QGIS, and Power BI. Visualization is shown as essential for communicating results to policymakers, technicians, and the public.
The report concludes by emphasizing the need for SOPs, training programs, capacity building, and continued regional collaboration to implement HAQMN successfully. While acknowledging differences in technical capacity across East Asia, it stresses the importance of consistent QA/QC, calibration, and transparency to ensure meaningful, comparable air quality data. The manual is intended as an adaptable but rigorous reference for EANET members seeking to expand air quality monitoring through cost‑effective, scientifically robust hybrid networks.
Keywords
Low-cost sensors, reference-grade monitors, HAQMN, calibration, co-location, QA/QC, parallel monitoring test, data correction, sensor networks, air quality monitoring, EANET, PM2.5, ozone, NO₂, machine learning, installation guidelines, maintenance, data screening.