Global warming is expected to greatly impact the thermal conditions of inland water bodies. Publicly available weather forecasts sufficiently resolve weather patterns above large lakes to satisfactorily force a lake hydrodynamics model, we suspect this is not the case for smaller lakes. We address this issue using weather outputs from a numerical weather prediction model at three different resolutions (3 km, 1 and 0.33 km) to force a hydrodynamics model of Lake George, NY for the month of June 2017. We find that increasing the resolution of the weather model improves the accuracy of the simulation, in part through a significant increase in total heat content of the lake. This warming is linked to an increase in downward heat transfer from the atmosphere in the higher resolution experiment, especially in the northern part of the lake where the lake is less than 2 km wide and the simulated surface winds are stronger. Interrogation of the weather model shows the stronger winds originate from a better representation of the land-lake interface (and thereby surface roughness length) at higher resolution. At the end of the one month simulation, we find the hypolimnetic volume of Lake George has reduced by 15% in the 0.33 km experiment compared to the 3 km experiment. If this discrepancy were systemic it would have obvious consequences for predicting the ecoysystem function over an extended period of small-to-medium sized lakes.