Multi-scale validation of a new soil freezing scheme for a land-surface model with physically-based hydrology

Soil freezing is a major feature of boreal regions with substantial impact on climate. The present paper describes the implementation of the thermal and hydrological effects of soil freezing in the land surface model ORCHIDEE, which includes a physical description of continental hydrology. The new soil freezing scheme is evaluated against analytical solutions and in-situ observations at a variety of scales in order to test its numerical robustness, explore its sensitivity to parameterization choices and confront its performance to field measurements at typical application scales. Our soil freezing model exhibits a low sensitivity to the vertical discretization for spatial steps in the range of a few millimetres to a few centimetres. It is however sensitive to the temperature interval around the freezing point where phase change occurs, which should be 1 C to 2 C wide. Furthermore, linear and thermodynamical parameterizations of the liquid water content lead to similar results in terms of water redistribution within the soil and thermal evolution under freezing. Our approach does not allow firm discrimination of the performance of one approach over the other. The new soil freezing scheme considerably improves the representation of runoff and river discharge in regions underlain by permafrost or subject to seasonal freezing. A thermodynamical parameterization of the liquid water content appears more appropriate for an integrated description of the hydrological processes at the scale of the vast Siberian basins. The use of a subgrid variability approach and the representation of wetlands could help capture the features of the Arctic hydrological regime with more accuracy. The modeling of the soil thermal regime is generally improved by the representation of soil freezing processes. In particular, the dynamics of the active layer is captured with more accuracy, which is of crucial importance in the prospect of simulations involving the response of frozen carbon stocks to future warming. A realistic simulation of the snow cover and its thermal properties, as well as the representation of an organic horizon with specific thermal and hydrological characteristics, are confirmed to be a pre-requisite for a realistic modeling of the soil thermal dynamics in the Arctic.