Technical note: Time lag correction of aquatic eddy covariance data measured in the presence of waves

Abstract

Abstract. Extracting benthic oxygen fluxes from eddy covariance time series measured in the presence of surface gravity waves requires careful consideration of the temporal alignment of the vertical velocity and the oxygen concentration. Using a model based on linear wave theory and measured eddy covariance data, we show that a substantial error in flux can arise if these two variables are not aligned correctly in time. We refer to this error in flux as the time lag bias. In one example, produced with the wave model, we found that an offset of 0.25 s between the oxygen and the velocity data produced a 2-fold overestimation of the flux. In another example, relying on nighttime data measured over a seagrass meadow, a similar offset reversed the flux from an uptake of −50 mmol m −2 d −1 to a release of 40 mmol m −2 d −1 . The bias is most acute for data measured at shallow-water sites with short-period waves and low current velocities. At moderate or higher current velocities (> 5-10 cm s −1 ), the bias is usually insignificant. The widely used traditional time shift correction for data measured in unidirectional flows, where the maximum numerical flux is sought, should not be applied in the presence of waves because it tends to maximize the time lag bias or give unrealistic flux estimates. Based on wave model predictions and measured data, we propose a new time lag correction that minimizes the time lag bias. The correction requires that the time series of both vertical velocity and oxygen concentration contain a clear periodic wave signal. Because wave motions are often evident in eddy covariance data measured at shallow-water sites, we encourage more work on identifying new time lag corrections.

Dynamics of benthic metabolism, O2, and pCO2 in a temperate seagrass meadow

“…In situations with well-defined current, flux estimates can often be improved by applying a time shift correction of the O 2 data relative to the velocity data (Fan et al 1990;McGinnis et al 2008;Lorrai et al 2010). Conversely, in the presence of waves, this correction can introduce substantial biases in the flux estimate (Berg et al 2015). Combined with a generally insignificant flux bias due to temporal sensor signal misalignment found in cospectral analysis of the vertical velocity and the O 2 concentration for selected deployments , this prompted us not to apply any time shift correction.…”

Reach‐scale river metabolism across contrasting sub‐catchment geologies: Effect of light and hydrology

“…While the technique has previously been successfully applied to investigate benthic oxygen exchange in rivers (Chipman et al 2012;Berg et al 2013;Murniati et al 2015) and impounded riverine systems (McGinnis et al 2008;Lorke et al 2012), it has only very recently been applied to assess seasonal stream metabolism (Koopmans and Berg 2015). General considerations on AEC applications have been extensively evaluated in recent publications (e.g., Lorrai et al 2010;Berg et al 2013;Holtappels et al 2013;Berg et al 2015;Donis et al 2015; see also Methods in this study). However, there are aspects that are of particular relevance for deployments in headwaters and lower order streams that need further consideration.…”

Benthic Carbon Mineralization and Nutrient Turnover in a Scottish Sea Loch: An Integrative In Situ Study

“…Provided the sensor characteristics and measuring conditions the maximum effect was assessed by the procedure described in Berg et al (2015) and amounted to 8 % of the measured flux. However, as the applied sensors will not express maximum stirring sensitivity from all current directions ) the average effect for the respective deployments must have been considerably less.…”