The prediction of snowmelt in mountainous forests strongly depends on the accurate description of sensible and latent heat turbulent fluxes. Uncertainty about the withincanopy wind conditions especially poses a challenge, with relatively few studies examining both above-and below-canopy turbulent fluxes. In this study, turbulent flux predictions from a state-of-the-art watershed model GEOtop were verified against eddy covariance data from one above-canopy tower and two below-canopy towers in a snow-dominated coniferous forest in south-eastern Wyoming. The model was applied in one-dimensional vertical mode using field-observed vegetation parameters and laboratory-measured soil water retention data. The model was calibrated by identifying optimum values for the canopy fraction and the within-canopy eddy decay coefficient using the brute-force method. Above-canopy sensible heat flux at the Glacier Lakes Ecosystem Experiments Site was predicted reasonably well (r 2 = .851). The prediction of above-canopy latent heat flux was weaker (r 2 = .426). For latent heat flux, errors in 30-min values offset each other when fluxes were aggregated over time, resulting in realistic mean diurnal trends. Below-canopy turbulent flux at two sites in the Libby Creek Experimental Watershed were predicted with variable success with r 2 = .031-.146 for sensible heat flux and r 2 = .445-.581 for latent heat flux. Modelled below-canopy sensible heat flux was too low due to the underestimation of daytime ground surface temperature, because of not enough solar radiation reaching the soil surface. This study suggests that future work on GEOtop and related models should include better parameterizations of the ground surface energy balance to more reliably predict snowmelt and streamflow from mountainous forests. KEYWORDS forest, modelling, snow cover, surface energy balance 1 | INTRODUCTION Streamflow from snow-dominated mountainous ecosystems is an important source of water in many parts of the world, including the Western United States (Bales et al., 2006). The timing of snowmelt is strongly influenced by the canopy and ground surface energy balances. These energy balances show high spatio-temporal variability with differences in elevation, slope, aspect, and vegetation cover, and diurnal and seasonal fluctuations in solar radiation, temperature, and precipitation, all playing a role. As a result, watershed computer