Addressing the influence of instrument surface heat exchange on the measurements of CO2 flux from open-path gas analyzers


George G. Burba, tel. +1 402 467 0608, fax +1 402 467 2819, e-mail:


There is a growing concern in the flux community that using the eddy covariance method with open-path CO2 analyzers often leads to measurements of an apparent ecosystem CO2 uptake during off-season periods, especially in cold climates. Such uptake has not been observed when measurements were made with closed-path analyzers, chambers, or profile methods, suggesting it is an artifact due in some way to the use of open-path analyzers. In this study, a series of laboratory tests and field experiments were conducted to determine the magnitude of the instrument surface heat exchange in the open path and its relationship with the measured CO2 flux. Results showed that (1) the surface of an open-path instrument became substantially warmer than ambient due to electronics and radiation load during daytime, while at night, radiative cooling moderated temperature increases in the path; (2) high-frequency temperature measurements inside the path were correlated with vertical wind speed producing sensible heat flux inside the instrument path exceeding the ambient heat flux by up to 14%; (3) enclosing the open-path instrument eliminated the sensible heat flux in the path, and caused measured CO2 flux to match a closed-path reference; (4) using sensible heat flux measured directly inside the open path in the WPL term instead of the ambient sensible heat flux also led to a match in CO2 flux between open-path instrument and closed-path reference; and (5) correcting previously collected open-path CO2 flux data was possible by estimating the instrument heating effect with a semi-empirical model using standard weather variables. Results showed that all proposed techniques led to a significant reduction in apparent CO2 uptake during off-season periods and to a reduction of the underestimation of CO2 release in other periods. Close agreement between the open-path measurements and closed-path references was achieved in all cases.