The AirSensEUR system demonstrated robust performance in maintaining a 40°C temperature-controlled chamber for CO₂ sensors (K96 and D300) during extended testing (June 2024–April 2025), overcoming seasonal challenges. A two-week calibration (2023-09-17–2023-10-01) yielded high R² values between reference and predicted CO₂ for both sensor types, though drift emerged in the third campaign (May 2023–April 2025). While K96 sensors exhibited positive drift, D300 sensors showed negative drift, in both cases likely due to detector sensitivity loss or contamination. A time-dependent calibration model incorporating drift as a covariate improved K96 accuracy, meeting the 10 ppm Data Quality Objective (DQO), but most D300 sensors failed to meet DQO even with correction. The study highlights the need for real-world drift estimation strategies, proposing a post-hoc method using long-term (>1 year) regression of daily minimum sensor data to infer drift rates without reference CO₂. This approach, though requiring extended data collection, addresses the critical challenge of maintaining sensor accuracy in unmonitored environments. Future work focuses on optimizing PID feedback loops for temperature stability and refining drift correction protocols for operational scalability.

YATKIN Sinan;  GERBOLES Michel;  MANCA Giovanni;  SIGNORINI Marco; 

2025-07-09

Publications Office of the European Union

JRC142994

978-92-68-29262-4 (online),   

1831-9424 (online),   

EUR 40377,    OP KJ-01-25-367-EN-N (online),   

https://publications.jrc.ec.europa.eu/repository/handle/JRC142994,   

10.2760/1114139 (online),