Frozen soil parameterization in SiB2 and its validation with GAME-Tibet observations

This paper presents a new frozen soil parameterization in the land surface scheme (LSS). A modified approximation Stefan solution is incorporated in the framework of the land surface model-SiB2 to calculate the frost/thaw depth over time and to estimate the soil moisture and temperature profiles during the freeze/thaw cycle. The structure of the soil model in Simple Biosphere Model 2 (SiB2) is kept, but the governing equations of water balance and surface heat balance are modified to account for soil freezing/thawing. The model is calibrated and validated using the GEVVEX Asian Monsoon Experiment (GAME)-Tibet observations at MS3608 site, Tibetan Plateau. First, the original SiB2 without frozen soil parameterization is calibrated using the observations in summer; an optimized parameter set for the short grassland in the Tibetan Plateau was obtained. Then, the modified SiB2 with frozen soil parameterization is validated using the soil moisture and temperature observations in winter. The results show that the new model estimates the frost depth precisely with an error less than 9% and predicts the soil temperature reasonably and phase transition time realistically. It also improves the soil moisture estimation in the surface layer and the root zone significantly. The mean absolute errors of simulated soil moisture using the new model are 0.020 and 0.013 in the surface layer and root zone, respectively, far below than those without using the frozen soil parameterization. (C) 2003 Published by Elsevier Science B.V.