The interaction between large inland water bodies and the atmosphere impacts the evolution of regional weather and climate, which in turn affects the lake dynamics, thermodynamics, ice-formation, and, therefore, ecosystems. Over the last decades, various approaches have been used to model lake thermodynamics and dynamics in standalone mode or coupled to numerical atmospheric models. We assess a turbulence-closure $$k-\epsilon$$ k - ϵ multi-column lake model in standalone mode as a computationally-efficient alternative to a full three-dimensional hydrodynamic model in the case of Lake Geneva. While it struggles to reproduce some short-term features, the multi-column model reasonably reproduces the seasonal mean of the thermal horizontal and vertical structures governing heat and mass exchanges between the lake surface and the lower atmosphere (stratified period, thermocline depth, stability of the water column). As it requires typically two orders of magnitude less computational ressources, it may allow a two-way coupling with a RCM on timescales or spatial resolutions where full 3D lake models are too demanding.

The interaction between large inland water bodies and the atmosphere impacts the evolution of regional weather and climate, which in turn affects the lake dynamics, thermodynamics, ice-formation, and, therefore, ecosystems. Over the last decades, various approaches have been used to model lake thermodynamics and dynamics in standalone mode or coupled to numerical atmospheric models. We assess a turbulence-closure k − ε multi-column lake model in standalone mode as a computationally-efficient alternative to a full three-dimensional hydrodynamic model in the case of Lake Geneva. While it struggles to reproduce some short-term features, the multi-column model reasonably reproduces the seasonal mean of the thermal horizontal and vertical structures governing heat and mass exchanges between the lake surface and the lower atmosphere (stratified period, thermocline depth, stability of the water column). It may therefore allow a two-way coupling with a RCM on timescales or spatial resolutions where full 3D lake models are too demanding in terms of computational resources.