Addressing a sensible heat bias in open-path eddy covariance carbon dioxide flux measurements

Open-path (OP) infrared gas analyzers (IRGA) are widely used for CO₂ eddy covariance flux measurements in diverse ecosystems, including in arid desert environments. These high sensible heat and low CO₂ flux conditions can lead to a systematic bias in the estimation of the carbon exchange. Numerous studies using both open- and closed-path IRGAs report large overestimates of CO₂ uptake in the OP measurements, which persists for all seasons and is not driven by biological activity, but rather by instrumentation artefacts. Despite the attempts to address these biases, their origin and the appropriate correction approaches remain unresolved. Sensor-path heat exchange has been considered as a potential source of the bias. Consequently, later models OP gas analyzers have eliminated the self-heating effects, yet they still exhibit apparent CO₂ uptake. In this study we consider the influence of ambient air temperature on the absorption in the CO₂ spectral band typically used in non-dispersive broadband IRGAs as the source of the bias. We show the results from simulations of infrared transmission in the CO₂ spectral band using high resolution molecular transmission (HITRAN) database.  We evaluated the temperature sensitivity of an IRGA by simulating integrated absorption spectra for a typical interference optical filter with a 100 nm passband where the CO₂ density was kept constant, and the gas mixture temperature was varied between 244 and 385 K. The data show that if the absorption is not corrected for temperature of the air in the optical sensing path a bias is introduced. The bias causes underestimation of CO₂ density at warmer temperatures and overestimation of CO₂ density at low temperatures. We conclude that OP gas analyzer measurements need to be corrected for the effects of changes in air temperature in the sensing path. We demonstrate that the correction is not universal, but rather instrument specific and depends on the actual pass band of the specific interference filter used.