Impact of different eddy covariance sensors, site set-up, and maintenance on the annual balance of CO2 and CH4 in the harsh Arctic environment

Improving year-round data coverage for CO2 and CH4 fluxes in the Arctic is critical for refining the global C budget but continuous measurements are very sparse due to the remote location limiting instrument maintenance, to low power availability, and to extreme weather conditions. The need for tailoring instrumentation, site set up, and maintenance at different sites can add uncertainty to estimates of annual C budgets from different ecosystems. In this study, we investigated the influence of different sensor combinations on fluxes of sensible heat, CO2, latent heat (LE), and CH4, and assessed the differences in annual CO2 and CH4 fluxes estimated with different instrumentation at the same sites. Using data from four sites across the North Slope of Alaska, we found that annual CO2 fluxes estimated with heated (7.5 ± 1.4 gC m−2 yr−1) and non-heated (7.9 ± 1.3 gC m−2 yr−1) anemometers were within uncertainty bounds. Similarly, despite elevated noise in 30-min flux data, we found that summer CO2 fluxes from open (−17.0 ± 1.1 gC m−2 yr−1) and close-path (−14.2 ± 1.7 gC m−2 yr−1) gas analyzers were not significantly different. Annual CH4 fluxes were also within uncertainty bounds when comparing both open (4.5 ± 0.31 gC m−2 yr−1) and closed-path (4.9 ± 0.27 gC m−2 yr−1) gas analyzers as well as heated (3.7 ± 0.26 gC m−2 yr−1) and non-heated (3.7 ± 0.28 gC m−2 yr−1) anemometers. A continuously heated anemometer increased data coverage (64%) relative to non-heated anemometers (47–52%). However, sensible heat fluxes were over-estimated by 12%, on average, with the heated anemometer, contributing to the overestimation of CO2, CH4, and LE fluxes (mean biases of −0.03 μmol m−2 s−1, −0.05 mgC m−2 h−1, and −3.77 W m−2, respectively). To circumvent this potential bias and reduce power consumption, we implemented an intermittent heating strategy whereby activation only occurred when ice or snow blockage of the transducers was detected. This resulted in comparable coverage (50%) during winter to the continuously heated anemometer (46%), while avoiding flux over-estimation. Closed and open-path analyzers showed good agreement, but data coverage was generally greater when using closed-path, especially during winter. Winter data coverage of 26–32% was obtained with closed-path devices, vs 10–14% for the open-path devices with unheated anemometers or up to 46% and 35% using closed and open-path analyzers, respectively with heated anemometers. Accurate estimation of LE remains difficult in the Arctic due to strong attenuation in closed-path systems, even when intake tubes are heated, and due to poor data coverage from open-path sensors in such a harsh environment