<p>Recently, global warming has increased the occurrence of intense precipitation and strong winds, leading to frequent mountain disasters such as wildfires and landslides. Land–atmosphere energy exchange in mountainous regions strongly influences local meteorological conditions. In winter, snow cover, through its high albedo and insulating effects, exerts a strong feedback on the surface energy balance; however, the accuracy of current atmospheric models in simulating these processes is limited owing to uncertainties in key snow-related variables, particularly snow depth, snow cover fraction, and snow water equivalent. The objective of this study was to improve the snow-related representation in land surface parameterization schemes suitable for simulating surface energy balance processes in mountainous areas and to propose an optimal method for realistic model initialization. For this, the Weather Research and Forecasting model (version 4.3) was used for experiments adopting three surface exchange heat coefficient parameterization schemes (N11, Z98, and C97). Flux tower observations from Hongcheon, Samcheok, and Pyeongchang, Gangwon province, Republic of Korea, were used for model validation. Additionally, sensitivity experiments of the melt factor for snow depletion curve (MFSNO) and the improvement of snow-related initial input data based on a snow-cover correction algorithm were performed over the target domain. Model evaluation showed that the Z98 scheme showed the best performance for simulating surface energy balance processes, while all three schemes tended to overestimate net radiation and sensible heat flux during snow-covered periods because of underestimated snow cover. In MFSNO sensitivity experiments based on Z98, MFSNO2.0 most realistically reproduced changes in albedo and snow cover. In addition, the proposed model initialization algorithm and MFSNO adjustment effectively enhanced the model’s performance in representing snow cover and albedo in mountainous areas.</p>