Addressing Effects of Environment on Eddy-Covariance Flux Estimates at a Temperate Sedge-Grass Marsh

Abstract

We examined the net-ecosystem-exchange (NEE)-based annual carbon-balance estimates obtained from eddy-covariance (EC) measurements at an unmanaged sedge-grass marsh ecosystem (Třeboň, Czech Republic, 49°1′ N, 14°46′ E), seeking methods to improve the EC measurements in inhomogeneous environment. The data filtering procedure was developed using three thresholds: (a) a stationarity test; (b) a stability {u}_{*}-threshold; and (c) a high relative humidity RH-threshold. This procedure was tested in 2014, a year without significant floods and drought events led to a stable water table, reducing the effect of soil respiration on the EC measurements. Estimates of annual carbon-balance were reduced from 182 to 234 ± 12 gC m−2 year−1 for the initial data to 39–44 ± 8 gC m−2 year−1 after the RH ≤ 95% filtering and to 24–26 ± 7 gC m−2 year−1 after the further {u}_{*} ≥ 0.1 m s−1 filtering. Applying the precipitation/fog threshold reduced this balance to 10–12 ± 7 gC m−2 year−1, closer to carbon neutrality. Up to 9.5% of this identified shift occurred during apparent nocturnal downslope katabatic drainage flows or plume descent coming from the nearby town of Třeboň. High-RH conditions account for up to 27% of this shift. Moreover, both conditions together account for an additional 67% of the identified carbon-balance change. Removing these non-ecosystem-related processes brings EC measurements closer to values of an unmanaged-ecosystem productivity, providing a better NEE-based estimate for the net ecosystem production. The presented procedure is applicable to EC measurements conducted at different wetlands or terrestrial ecosystems with similar conditions.