The CO2 exchange of a small mountain lake as affected by the local thermo-topographically driven flow regime

Abstract

<p>Lakes receive large amounts of carbon (C) from the surrounding catchment and, together with the connecting streams, play an important and active role in the global C cycle. The received C can either be lost through the outflow and eventually transported to the ocean, or transformed and stored in sediments or outgassed to the atmosphere. Globally, lakes are estimated to emit 0.3 – 0.64 Pg C m-2 in form of CO<sub>2</sub> annually.  Although subalpine and alpine lakes were observed to be supersaturated with CO<sub>2</sub>, long-term measurements of lake-atmosphere CO<sub>2</sub> exchange are sparse. Several methods to quantify water-atmosphere gas exchange exist, like chambers, eddy covariance (EC), mass-balance or gradient based methods including boundary layer models (BLM), each having its own advantages and disadvantages. However, quantifying CO<sub>2</sub> exchange in aquatic ecosystems has often proved to be challenging. Here, both the BLM and the EC methods were used to estimate the air-water CO<sub>2</sub> exchange of Lake Lunz, a small lake situated in complex mountainous topography of the Austrian Alps. The results indicated that the lake was a small source of CO<sub>2</sub>. Fluxes were affected by the thermo-topographic flow regime of the field site and its surroundings which drove the local wind pattern but also determined the local atmospheric CO<sub>2</sub> concentration.  During most nights, a significant increase in atmospheric CO<sub>2</sub> was observed which decreased the differential CO2 concentration at the air-water interface and therefore led to decreased nocturnal CO<sub>2</sub> efflux. This diurnal pattern, however, was obscured in the EC measurements, because the method itself highly depends on the local wind regime. Because lakes are an integral part of mountain ranges which are characterized by catchments with complex topography, our findings are most likely of broader impact.</p>