Thermokarst lakes play a key role in Arctic landscapes. Even if global available freshwater on the surface makes up less then 1%, permafrost areas in circumarctic regions show a lake cover of up to 50%. Effects on energy and water balances as well as biogeochemical cycles are still discussed in permafrost research. Many remote sensing studies investigated water balances of Arctic lakes, mainly in Northern America. In-field data is still very rare but gives deeper insight into water balance dynamics. Field measurements can also be used as ground-truth data for future remote sensing studies. In this thesis, a water balance model was set up for the thermokarst lake "Lucky Lake" on Kurungnakh Island in the western part of the Lena River Delta, Northeastern Siberia. This study aims to investigate main drivers of the water balance as well as possibly missing in- and output sources. Surface discharge and water level change was measured directly at the lake, whereas additional meteorological data was derived from a climatological site at Samoylov Island (Boike2019), 10 km distance to the study lake. Snow-water-equivalent was estimated from snow properties. Evaporation was calculated using three different methods. The aerodynamic approach models evaporation the best in terms of absolute values and dynamics as a comparison of calculated and measured evaporation rates in summer 2014 shows (by an eddy flux covariance system at a floating raft on Lucky Lake, Franz2018). Mean evaporation rate is 1.2 mm/d. Results are used in the water balance model. The Penman equation underestimates actual values but calculates short term dynamics well. The Priestley-Taylor model is only suitable for a rough full-summer estimate. Due to lacking data, water balances can only be assessed in 2015 and 2017. In both years, overall water level change was measured to be positive, which confirms remote sensing observations of increasing lake surface areas in Russian continuous permafrost (e.g. Smith2005). Contrary, water level was modelled to be negative in both years. Under complete data availability, the model represents negative water balances right after the beginning of the snow free period well. Snow melt input is overestimated by on third compared to actual rise in water level. The model fails to calculate positive or stable water balances appropriate. One reason can be a missing input source as two small inflow channels connecting a more northern thermokarst lake to Lucky Lake are not considered in this study. Overall, the water balance of Lucky Lake is snowmelt-influenced in the beginning of the open water season. The effect of melt water declines rapidly, so that rainfall and discharge dominate water level changes for the rest of the summer. To improve the model and input sources quantification, three suggestions are made: i) measurements of full-summer discharge, ii) local measurements of snow properties, iii) measurement of the additional input through two small channels in the north of the lake. However, data used in this study can be used for further investigation on carbon release due to thermokarst lakes or as validation data for remote sensing studies.