Improving the long-term accuracy of low-cost sensors through a novel Automatic Drift Correction

Low‑cost air quality sensors are increasingly deployed to complement reference monitoring stations due to their low cost and ease of installation. However, these sensors are susceptible to environmental conditions and long‑term nonlinear drift, which can substantially degrade data accuracy over time. Existing calibration strategies, particularly those based on machine learning and periodic co‑location with reference instruments, improve performance but they often involve considerable maintenance effort and costs, especially when managing large-scale networks.

Kunak has developed an innovative Automatic Drift Correction (ADC) algorithm that autonomously correct the baseline and sensitivity drift in electrochemical gas sensors. This new method works alongside the Kunak algorithm that compensates for environmental factors such as temperature and humidity across the full operating range. Together, they allow for accurate measurements without the need for reference data or frequent manual recalibrations.

The ADC algorithm enables continuous and sensor-specific baseline and sensitivity adjustments independently from the location and the ambient conditions, ensuring consistent data quality over time, and without using Machine Learning or Artificial Intelligence models. This significantly reduces the operational complexity and costs associated with maintaining air quality sensor networks, especially in large deployments.

We evaluated the proposed method on a NO2 sensors co‑located with a regulatory air quality monitoring station (AQMS) and compared the performance against a conventional manual calibration procedure. Results demonstrate that the ADC algorithm maintains data integrity over time with performance comparable to the periodic manual calibration, even under variable environmental conditions.

The method offers a scalable and reliable alternative to traditional approaches and supports the recommendations outlined by the WMO, which highlight the need for automated, low-effort maintenance solutions.

This work presents a practical and efficient tool for sustaining the long-term reliability of air quality data, making it especially suitable for distributed air quality monitoring networks.