Field Intercomparison of Mid-Cost CO2 Sensors for Urban Atmospheric Monitoring

Urban environments represent major hotspots of atmospheric CO₂ emissions; however, the availability of measurements resolving fine spatial and temporal variability remains limited. This limitation is largely due to the high cost and sparse deployment of reference-grade monitoring systems in complex urban settings. In recent years, sensor-based approaches have shown that dense networks of low- to mid-cost CO₂ sensors can substantially enhance spatial coverage in complex urban settings. Urban sensor networks, such as the Zurich CO₂ Sensor (ZiCOS) network (Grange et al., 2025) and the BErkeley Atmospheric CO₂ Observation Network (BEACO₂N) in California (Shusterman et al., 2016), have shown that such deployments can improve the characterization of the urban CO₂ spatial patterns and temporal dynamics. These developments highlight how sensor-based networks can enable denser spatial coverage and provide an effective pathway toward a high-resolved  description of urban CO₂ variability. 

The current study outlines a recently initiated project on urban CO₂ budgeting in the Po Valley, a densely populated area in southern Europe, combining atmospheric monitoring by reference equipment and mid-cost sensors, with atmospheric modelling by urban-scale Lagrangian particle dispersion modelling. We assessed the field comparability of two identical mid-cost CO₂ sensors GMP343 (Vaisala Oy). The instruments were deployed between May–November 2025 at the rooftop of the Geophysical Observatory of the University of Modena and Reggio Emilia, a 40 m high tower located in the city centre of Modena, Italy. The sensors were laboratory calibrated prior to deployment, and measured CO₂ concentrations were corrected for temperature and pressure using the built-in firmware algorithm, followed by the application of sensor-specific calibration offsets.

To assess the inter-sensor agreement and operational stability, we processed the six months of continuous measurement data. Sensor performance was evaluated using correlation analysis, error statistics, and Deming regression, demonstrating strong agreement between the sensors and good stability. The two mid-cost sensors exhibited a high linear correlation (Pearson’s r = 0.965) and a mean bias of 2.47 ppm during the intercomparison. The results achieved so far showed their suitability for high-resolution urban monitoring and for an integration with reference-grade eddy covariance (EC) observations in urban CO₂ assessment studies. 

Since January 2026, a 2 m pole on the rooftop of the Observatory has been equipped with a reference-grade EC measurement setup consisting of a LI-COR 7200RS gas analyzer (LI-COR Biosciences) and a 3D Gill WindMaster anemometer (Gill Instruments), and provides continuous measurements of urban CO₂ fluxes. Concurrently since December 2025 the mid-cost sensors were moved at two urban air quality regulatory monitoring stations under urban background and urban traffic conditions, where several regulatory atmospheric pollution monitors are already in place. Study outlooks include the maintenance of the CO₂ network for at least 12 months and the setup of an urban scale CO₂ dispersion model combining both biogenic and anthropogenic fluxes within a lagrangian particle dispersion model.