An earlier version of the hydro-thermodynamic soil vegetation scheme (HTSVS) was further developed to numerically investigate the long-term evolution of water budget elements (water supply to the atmosphere, ground water recharge, change in storage) in climate studies. In doing so, parameterizations of root effects, infiltration, soil frost, and snow insulation were included into HTSVS to predict these water budget elements for a period of 2050 days continuously covered by routine data of a lysimeter and a climate station. The results of simulations without and with inclusion of the new parameterizations as well as various sensitivity studies indicate that the insulating effect of a snow-pack, soil water freezing as well as the Dufour- and Ludwig-Soret-effects play a notable role on the long-term water budget and soil temperature evolution. Including frost effects yields improved soil temperature predictions. The results show that the kind of vertical root distribution can be important for predicting water budgets. Based on these results we conclude that land surface models considered for long-term integration purposes require parameterizations of soil frost, snow and water uptake by roots to appropriately catch the broad cycle of soil water budget elements.