[1] Significant amounts of terrestrial carbon are processed in lakes and emitted into the atmosphere as CO₂. However, due to lack of appropriate measurements the absolute role of lakes in the landscape as sinks or sources of CO₂ is still uncertain. We conducted the first long‐term, ecosystem‐level CO₂ flux measurements with eddy covariance technique in a boreal lake within a natural‐state catchment covering 5 years. The aim was to reveal the natural level of CO₂ flux between a lake and the atmosphere and its role in regional carbon cycling. On average, the lake emitted ca 10% of the terrestrial net ecosystem production of the surrounding old‐growth forest and the main immediate drivers behind the fluxes were physical rather than biological. Our results suggest that lakes are an integral part of terrestrial carbon cycling.

Interaction of under‐ice physical, chemical, and biological processes with lake ice/snow cover is examined to better understand how changing winter climate may affect lake ecosystems. We derived under‐ice dissolved oxygen (DO) dynamics from high‐frequency observations and modified a widely used lake metabolism model by including the effect of freezing and thawing on DO concentration. Estimates were produced for the production and respiration in a shallow lake on the Mongolian Plateau in three winters. Diel, synoptic, and seasonal variations in DO concentration were detected as responses to solar radiation, episodic snowfall events, and occasional convective mixing. Based on the observations and a radiative transfer model, incident solar radiation was partitioned into reflectance, absorbance, and transmittance by the snow and ice cover. For bare ice, the contributions of these three parts were 35%, 39%, and 26%, respectively, while under a new 4.5 cm thick snow cover, the corresponding values were 79%, 17%, and 3%. This points out the critical role of snow and ice on under‐ice light conditions, which is the primary forcing for the temperature and the rate of photosynthesis under ice. The results showed three principal factors, which influenced under‐ice DO and metabolism: (1) thickness and optical properties of ice and snow, which affected the light transfer and depth of the euphotic zone, (2) mediated radiation and ice‐water heat transfer which controlled water temperature, and (3) DO exclusion during freezing and dilution by melt water. This study highlights the ecosystem characteristics in shallow ice‐covered lakes in arid temperate regions and promotes our understanding of the response of the cold aquatic environment to climate change.

To improve the understanding of the seasonal evolution of the mass and heat budget of ice-covered lakes in the cold and arid climate zone, in-situ observations were collected during two winters (2016–2017 and 2017–2018) in Lake Wuliangsuhai, Inner Mongolia, China. The mean snow thickness was 5.2 and 1.6 cm in these winters, due to low winter precipitation. The mean ice thickness was 50.9 and 36.1 cm, and the ice growth rate was 3.6 and 2.1 mm day⁻¹ at the lower boundary of ice. Analyses of mass and heat balance data from two winters revealed that the surface heat budget was governed by solar radiation and terrestrial radiation. The net heat flux loss of the ice was 9–22 W m⁻², affected by the snow and ice thickness. Compared to boreal lakes, Lake Wuliangsuhai received more solar radiation and heat flux from the water. The ice temperature had a strong diurnal variation, which was produced by the diurnal cycles of solar radiation, and air and water temperatures. These results expand our knowledge of the evolution of mass and heat balance in temperate lakes of mid-latitude arid areas.

[1] The largest uncertainties in accurately resolving the role of rivers and streams in carbon cycling stem from difficulties in determining gas exchange between water and the atmosphere. So far, estimates for river‐atmosphere gas exchange have lacked direct ecosystem‐scale flux measurements not disturbing gas exchange across the air‐water interface. We conducted the first direct riverine gas exchange measurements with eddy covariance in tandem with continuous surface water CO₂ measurements in a large boreal river for 30 days. Our measured gas transfer velocity was, on average, 20.8 cm h⁻¹, which is clearly higher than the model estimates based on river channel morphology and water velocity, whereas our floating chambers gave comparable values at 17.3 cm h⁻¹. These results demonstrate that present estimates for riverine CO₂ emissions are very likely too low. This result is also relevant to any other gases emitted, as their diffusive exchange rates are similarly proportional to gas transfer velocity.