A 30-month Field Evaluation of Low‐Cost CO₂ Sensors Using a Reference Instrument

Abstract. CO₂ monitoring networks with low-cost and medium-precision sensors (LCSs) have become an exploratory direction for CO₂ observation under complex emission conditions in cities. Yet the performance of such LCS after deployment in the field faces significant challenges due to environmental impacts (e.g., temperature and humidity) and long-term drifts due to sensor degradation (e.g., the light source). Here, we conducted 30 months of co-located observations using LCS instruments (named SENSE-IAP) with a reference instrument (Picarro) to study the long-term performance of the LCSs under field conditions, which is essential for the correction and validation of mid-low cost CO₂ observation networks. The environmental correction system we developed effectively corrected the impact of daily environmental changes, which reduced the root mean square errors (RMSE) from 5.9±1.2 ppm to 1.6±0.5 ppm for SENSE-IAP. The corrections remained robust against seasonal environmental variations, and the daily RMSE was generally 1–3 ppm over the 30 months of observation. Long-term drifts, commonly occurring in LCS, resulted in biases reaching up to 27.9 ppm over two years. Furthermore, the seasonal drift cycle contributed an RMSE of up to 25 ppm after six months of the deployment. While the environmental correction system could not correct these errors, a linear interpolation method effectively corrected the long-term drift. The long-term drift calibration significantly decreased the RMSE to 2.4 ± 0.2 ppm over the 30-month observation. To improve the accuracy of high-density CO₂ networks utilizing LCSs, we recommend that the calibration frequency be no less than three months and not exceed six months, with optimal calibration performed during winter and summer to maintain daily accuracy within 5 ppm. These findings suggest that SENSE-IAP instruments can be deployed for a long period without the need for taking back to re-calibrate in the laboratory or frequent standard gas calibration in the field, thereby significantly reducing time, labor, and financial costs.