This work presents a numerical study of water and energy transfers within a permafrost dominated experimental watershed of Central Siberia, the Kulingdakan catchment (e.g.: Viers et al., 2015). This watershed has been studied for years in order to characterize and quantify the elementary transfers from the slopes to the stream along the seasonal cycles, and in the context of climate change. The water fluxes, strongly coupled with the thermal fluxes due to the presence of permafrost, are the main vectors of these matter transfers. In this study we aim to build a mechanistic model of the water and energy fluxes that may produce rigorous thermal and hydrological background for modellings the geochemical transfers from the slopes to the stream. The use of a mechanistic approach consists in resolving numerically the governing equations of the considered physical phenomena, established in the framework of continuum mechanics. The tool we used to produce a mechanistic model of the thermo-hydrological transfers in soils is permaFoam (Orgogozo et al., 2015), a solver for the coupled equations that describe unsaturated water transfers (with evapotranspiration) and thermal transfers (with freeze/thaw), implemented in the framework of OpenFOAM®, a well-known open source tool box for computational fluid dynamics. The main interest of using OpenFOAM® lays in its good performances in massively parallel computing (e.g.: Orgogozo et al., 2014). Indeed, due to the strong couplings and the strong non-linearities that are encountered in such physical problems, the use of high performance computing methods is needed to deal with the fine spatial and temporal discretizations required for the numerical resolutions. This is especially true when large scales are involved as is the case for the experimental watershed scale (here, Kulingdakan watershed, about 41 km² of surface). The Kulingdakan watershed is dominated by continuous permafrost, with larch forests growing on soils produced from weathering of basaltic rocks. The main variability of the landscape is related to the aspect of the slope translated to the amount of solar radiation received, which in turn controls the thermal status of the soil. Specifically, the aspect of the slopes may be either south aspected (higher solar energy input) or north aspected (lower solar energy input). With permaFoam two 2D vertical sections are build up to represent both south and north aspected slopes, with heterogeneous soils (taking into account the presence of a moss layer and of an organic (mor) layer overlaying the mineral soil). The topography and the spatial scales of these 2D models are derived from a simple Gravelius rectangle approach, which lead to modelling domains of 2.5 km of width, with slopes of about 20 %. We produce a quantitative modelling of the seasonal cycle of the active layer dynamics, based on monthly soil temperatures and meteorological forcings obtained by multiannual averaging of data acquired in the watershed (observations done between 2006 and 2012). A representative behavior of the active layer dynamics in present conditions is modelled in this way. The numerical results and the field data are in reasonable agreement. The model is capable of reproducing the variability of the thermo-hydrological dynamics between slopes of north and south aspects, as well as the variability of thermo-hydrological conditions along the slopes. Finally, we discuss numerous perspectives related to this numerical development for the modelling of present and future geochemical fluxes under climate change scenarios.