Modeling Eddy Correlation Biases Created by Velocity Sensitivities of Clark-type Oxygen Microelectrodes Under Waves

Graduation Date: 2015Field deployments, flume experiments, and a 2D wave model have been used to identify the characteristics of Clark-type oxygen microelectrodes and measurement parameters that could possibly bias eddy correlation (EC) flux measurements made under waves. Eddy correlation is a technique adapted from atmospheric sciences that couples concurrent and co-located velocity and oxygen measurements in order to calculate how much oxygen is being taken up by the sediment in aquatic environments. Field deployments made with two co-located microelectrodes suggest that individual sensor sensitivities to changes in the velocity field under wave conditions are biasing the EC calculation. Experiments performed in two different flumes illustrate the non-linear response of microelectrodes to changes in velocity and the individual nature of each microelectrode’s performance based on the shape and placement of the sensing cathode. Long, thin cathodes that are offset slightly from the membrane seem to have the necessary response time for EC measurements and also minimize the velocity effect. \ud Furthermore, a 2D wave model was used to test the sensitivity of the EC flux determination to various parameters specific to the wave conditions and to the microelectrode response. The model demonstrates that most parameters do not directly affect the flux, as long as both the velocity and oxygen time series are in phase temporally with the exception of a non-zero mean vertical velocity. The mean horizontal velocity can also greatly affect the flux if there is a delay in the microelectrode response or a spatial separation between the sampling volumes for the ADV and microelectrodes that results in a time lag between the velocity and oxygen time series. The results from the model suggest that correctly aligning the oxygen and velocity time series is extremely important and that non-zero mean vertical velocities may be biasing the flux calculation in some instances