Improvement and validation of a daily soil temperature submodel for freezing/thawing periods

The soil temperature submodel of the daily version of the CENTURY ecosystem model, DAYCENT, has been improved for freezing/thawing conditions in soils. The simulation of the soil heat propagation in the model is based on the Fourier heat now equation and approximated by use of the explicit finite difference scheme. In our study, we propose a tipping bucket algorithm to incorporate the discontinuous freezing/thawing process in the model by means of a calculated fusion energy pool. Daily average, maximum and minimum soil temperatures for each soil layer are estimated. The soil thermal diffusivity for each soil layer is simulated by the method of De Vries using upper and lower limits as well as a correction factor. A correction factor was optimized for the tipping bucket approximation of the freezing/thawing process. The model was tested and validated at three different sites with different climates and soils and showed good agreement with measured soil temperatures during winter periods. The mean deviation for maximum soil temperatures at 10 cm depth during the winter period was less than 1 degrees C for all locations; for the minimum temperatures the mean deviation was slightly higher. At a depth of 20 cm, mean deviations were higher but can nevertheless be seen as showing good results for the given approximations.