Observations and Modeling the Advection of Carbon from an Inland Lake Surrounded By a Forest

Abstract:

Lake Douglas is a small inland lake in Northern Michigan, surrounded by the forest of the University of Michigan Biological Station (UMBS). Over a decade of eddy-covariance measurements of carbon fluxes at the UMBS provide us good knowledge of the rates and dynamics of fluxes between the forest and the atmosphere. However, there is very little knowledge about the flux rates from the near-by lake and how they relate to the conditions in the surrounding forest. We have conducted eddy-covariance measurements of CO₂ fluxes from the lake over two summers. We found the microclimate predictably different over the lake. Carbon uptake rates by the lake ecosystem were high during the daytime, but lower than the adjacent forest, which led to higher atmospheric CO₂ concentrations over the lake surface than over the forest. Using concurrent wind and CO₂ concentration data from the lake and two flux towers in the forest we estimated the rate of lateral advection of CO₂ from the lake to the forest. Though the lake is almost never within the forest flux tower footprint, this advection term may bias forest carbon budget estimates.

To further study this advection, we generated a virtual experiment using a high resolution canopy-resolving large eddy simulation model (RAFLES). We assumed a circular lake surrounded by a homogeneous forest and prescribed typical conditions representing different times of day during the summer growing season. The observed latent and sensible heat flux rates at the forest and lake tower where prescribed to the simulated forest and lake patches in the simulation, respectively. The resolved wind field includes the effects of the different surface heat fluxes, as well as the roughness transition caused by the trees at the water edge. We used this wind field in a series of simulations using Hi-VACC, a scalar diffusion-advection model. In these Hi-VACC simulations we assumed the observed carbon flux rates as surface sinks for CO₂ at the two patch types. The model resolved the apparent flux rates that would have been measured at any virtual location along the lake-forest continuum. We show that the distance at which the lake effect is negligible is highly dependent of wind speed and sensible heat flux.