Snow plays an important role in land surface models (LSM) for climate and model applied over Fran studies, but its current treatment as a single layer of constant density and thermal conductivity in ORCHIDEE (Organizing Carbon and Hydrology in Dynamic Ecosystems) induces significant deficiencies. The intermediate complexity snow scheme ISBA‐ES (Interaction between Soil, Biosphere and Atmosphere‐Explicit Snow) that includes key snow processes has been adapted and implemented into ORCHIDEE, referred to here as ORCHIDEE‐ES. In this study, the adapted scheme is evaluated against the observations from the alpine site Col de Porte (CDP) with a continuous 18 year data set and from sites distributed in northern Eurasia. At CDP, the comparisons of snow depth, snow water equivalent, surface temperature, snow albedo, and snowmelt runoff reveal that the improved scheme in ORCHIDEE is capable of simulating the internal snow processes better than the original one. Preliminary sensitivity tests indicate that snow albedo parameterization is the main cause for the large difference in snow‐related variables but not for soil temperature simulated by the two models. The ability of the ORCHIDEE‐ES to better simulate snow thermal conductivity mainly results in differences in soil temperatures. These are confirmed by performing sensitivity analysis of ORCHIDEE‐ES parameters using the Morris method. These features can enable us to more realistically investigate interactions between snow and soil thermal regimes (and related soil carbon decomposition). When the two models are compared over sites located in northern Eurasia from 1979 to 1993, snow‐related variables and 20 cm soil temperature are better reproduced by ORCHIDEE‐ES than ORCHIDEE, revealing a more accurate representation of spatio‐temporal variability.