Improving Seasonal Diffusive Flux Estimates through Inverse Calibration of Lake Surface Methane Concentration
Abstract:Diffusive flux of methane to the atmosphere from freshwater lakes is thought to be better understood than the more stochastic ebullitive (bubble) flux. Simple models describing the flux are largely driven by wind speed and surface water concentrations. Rooted in 3 seasons of direct open water chamber measurements at Inre Harrsjön, a small and shallow lake located in the subarctic region of Sweden, we present a comparison between measured exchanges and those calculated from several piston velocity models ran at varying temporal resolutions, as well as a method to increase precision in modeled results.
On a majority of days the calculated piston velocities, per day, exceeded lake average depth. Under such conditions it is unlikely that the surface concentration can remain fixed. The result is that high piston velocities will effectively alter the gas content of the shallow lake, unless the sediment-to-lake flux matches the lake-to-atmosphere flux. Initial model runs using fixed surface concentrations confirmed this with seasonal averages of up to four times the up-scaled chamber measurements.
In an attempt to compensate for the lake’s inability to keep up with predicted fluxes, the fixed lake surface concentration was abandoned for variable wind-dependent lake surface concentrations. By inverting the gas exchange equation a required, or ideal, lake surface concentration - the concentration required by the models to match the measured fluxes at a given wind speed - could be obtained for dates where both wind and chamber measurements were available. These ideal concentrations were plotted as function of wind.
Linear and exponential functions describing the wind-dependent surface concentrations were used in subsequent runs, completely replacing the static concentration approach. We found that incorporation of variable wind-driven lake surface concentration into the models greatly improved the accuracy of seasonal diffusive flux estimates, reducing the apparent error to as low as 1%. We suggest that this method may be useful in other similar lakes to obtain a better estimate of seasonal diffusive methane fluxes.