We developed a floating outdoor laboratory for continuous measurements of mass and energy fluxes and limnic and atmospheric variables on two reservoirs of different trophic state. The installation as well as operation of such a complex observation system was not a standard issue but included a series of serious technical challenges, which had been successfully overcome. We could successfully test our hypotheses and three valuable data sets have been produced documenting the mass and energy exchange occurred at air-water interface and the temporal change of the atmospheric and limnic conditions in half-hourly temporal resolution over three years. A series of novel processes and unknown feedbacks between atmospheric and limnic variables have been discovered. The diurnal diurnal patterns of the sensible heat flux (H) are characterised by contrary behaviour com-pared to land surfaces. This behaviour is directly coupled with day-night-characteristic of the gradient between water surface temperature (Ts) and air temperature (Ta). At night heat is carried from the water surface to the atmosphere. This flux direction is supported by the additional lift due to the reduced density of moist air. The observed evaporation was unexpectedly low for a site where water is infinitely available and where the evaporation process is only limited by the availability of energy. This observation is in conflict with the standard assumption the evaporation of a water surface can be sized by the potential evaporation. Thus, we were surprised by the low measured values, which cannot be explained by measurement uncertainties and the underestimations caused by the energy balance closure problem. In contrary, our data clearly demonstrate that ET of a water surface is significantly smaller than estimates of potential evaporation. We hypothesize that large spatial differences and a gradient from the shore to the centre exist in the evaporation rate, which are controlled by interferences with sounding terrestrial sites, fetch, wind speed and atmospheric stability. Furthermore, we expect complex interactions between the evaporation rates and the spatial patterns of air and water surface temperature, air humidity and the heat storage in the water body. We plan to intensify these investigations in our recently submitted project proposal. The special temporal characteristic of energy fluxes is closely linked to temporal patterns of atmospheric stability in surface air. An unstable stratification was typical for night and a neutral stratification for day. In general, periods of stable stratification were extremely rare and occurred more frequently during the day. Thus, water surfaces are characterised by a significantly different temporal pattern of atmospheric stability than terrestrial sites. We were able to demonstrate and to quantify the effects of the atmospheric stability on momentum flux as well as on the flux rates of H and LE. This is a novel cognition, as the atmospheric stability is typically neglected when fluxes from inland waters are modelled. Thus, our research is an important contribution to improve models and our capabilities to simulate the mass and energy exchange from inland waters. Clear seasonal patterns were observed for the methane flux (FCH4) at the eutrophic Bautzen reservoir, which is coupled with the seasonal change of the oxygen concentration in the hypolimnion and largely driven by ebullition. However, we also observed for the first time short-term variations in both measured FCH4 and concentrations of CH4 in surface water. The carbon dioxid fluxes (FCO2) were characterised by distinct diurnal and seasonal patterns, which are caused by the temporal characteristic of photosynthesis in epilimnion and respiration in both epilimnion and hypolimnion. We could show that GHG emissions from exposed sediments in the drawdown area significantly affect the GHG budget of reservoirs and other dry surface waters. These emissions are dominated by CO2 and are regulated by a complex interaction of temperature, moisture, and organic matter content of the sediment. From the analysis of a remote sensing dataset we conclude that on the global scale 13% of all reservoir surfaces are dry. Consideration of reservoir drawdown increases total CO2 emissions from reservoirs on the global scale by 67%.