Error characterization of methane fluxes and budgets derived from a long-term comparison of open- and closed-path eddy covariance systems

Authors: DEVENTER, JULIAN - University Of Minnesota; GRIFFIS, TIMOTHY - University Of Minnesota; ROMAN, TYLER - University Of Minnesota; KOLKA, RANDALL - University Of Minnesota; WOOD, JEFFREY - University Of Minnesota; ERICKSON, MATTHEW - University Of Minnesota; Baker, John; MILLET, DYLAN - University Of Minnesota

Submitted to: Agricultural and Forest Meteorology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/23/2019
Publication Date: 12/15/2019
Citation: Deventer, J.M., Griffis, T.J., Roman, T.J., Kolka, R.B., Wood, J.D., Erickson, M.D., Baker, J.M., Millet, D.B. 2019. Error characterization of methane fluxes and budgets derived from a long-term comparison of open- and closed-path eddy covariance systems. Agricultural and Forest Meteorology. 278:107638. https://doi.org/10.1016/j.agrformet.2019.107638.
DOI: https://doi.org/10.1016/j.agrformet.2019.107638

Interpretive Summary: Methane(CH4) is one of the major anthropogenic greenhouse gases, with a global warming potential nearly 30 times that of CO2 on a molecule per molecule basis. There is much uncertainty about the relative contributions of different methane sources, so there has been much effort devoted in recent years to the measurement of methane emissions by eddy covariance, the primary method of choice. However, efforts to interpret reported results have been hampered by poor understanding of the errors and uncertainty inherent in the measurement process and by a lack of standardized methods for mathematical processing of flux data. We have addressed this issue using a 3.5 year data set of methane flux measurements made with two different types of gas analyzers in a wetland in northern MN.The results showed that after applying appropriate corrections for frequency losses, systematic errrors were small in comparison to other sources of uncertainty. In fact, random errors in the half-hourly fluxes can be quite large, and must be considered when comparing eddy covariance measurements to chamber-based data. Another challenge with measuring methane emissions is that there are inevitably gaps in the data associated with precipitation events, insufficient turbulence, and equipemtn malfunctions. Thus, in order to develop annual emission estimates it is necessary to employ gap-filling techniques. We evaluated four methods that have been proposed in the literature, and found that annual emission estimates were only slightly affected by the choice of method. Calculated annual emission totals agreed within 7% across all tested techniques, and averaged just less than 18 g-CH4 m-2yr-1. Total uncertainty was about 17%, comparable in magnitude to the interannual variability. These results will be useful in evaluating the significance of reported differences in emissions from different ecosystems, and in assessing the effectiveness of mitigation strategies.

Technical Abstract: Wetlands represent the dominant natural source of methane (CH4) to the atmosphere. Thus, substantial effort has been spent examining the CH4 budgets of global wetlands via continuous ecosystem-scale measurements using the eddy covariance (EC) technique. Robust error characterization for such measurements, however, remains a major challenge and no standardized processing or data QA/QC protocols have yet been established. Here, we quantify systematic, random and gap-filling errors and the resulting uncertainty in CH4 fluxes using a 3.5 year time series of simultaneous open- and closed path CH4 flux measurements over a sub-boreal wetland. We found that after correcting for high- and low frequency flux attenuation, the magnitude of systematic errors is negligible relative to other uncertainties. Based on an ensemble of estimates from three different approaches, we found that the 90% confidence interval of the relative random flux errors corresponded to 50% of the measured fluxes. Thus, errors on individual halfhourly CH4 fluxes can be large, and need to be appropriately characterized when fluxes are compared to chamber-derived or modeled CH4 emissions. Integrated annual fluxes were only moderately sensitive to gap-filling, based on an evaluation of 4 different methods. Calculated budgets agreed on average to within 7% (=1.5 g-CH4 m-2 yr-1). Marginal distribution sampling using open source code was among the best-performing of all the evaluated gap-filling approaches and it is therefore recommended given its transparency and reproducibility. Overall, estimates of annual CH4 emissions for both EC systems were in excellent agreement (within 0.6 g-CH4 m-2 yr-1) and averaged 17.7 g-CH4 m-2 yr-1. Total uncertainties on the annual fluxes were estimated to be within 17%, which was comparable to the observed interannual variability in emissions. Flux uncertainties thus need to be taken into account when comparing CH4 budgets between years or sites.