Evaluation of calculation models for the thermal conductivity of soils

In the study, a model matrix, based on the combinations of 1 thermal conductivity model for saturated soils (λsat), 4 thermal conductivity models for dried soils (λdry) and 4 Kersten number models for soils (Ke), was proposed to calculate the thermal conductivity of soils. 16 calculation models were included in the matrix. The matrix could be used to investigate the effects of the λdry models and Ke models on the calculational results of the 16 calculation models for the thermal conductivity of soils. The matrix was evaluated by 40 Canadian soils, and it is found that the He et al. model for λdry performs best among the 4 λdry models, the Côté and Konrad model for Ke performs best among the 4 Ke models. It also shows that the results of the calculation models in the matrix are different significantly, but the JCM (combination of the geometric mean model for λsat, the Johansen model for λdry, and the Côté and Konrad model for Ke) performs best among the calculation models for the thermal conductivity of soils. The effects of the λdry models and Ke models on the calculational results of calculation models for the thermal conductivity were evaluated by the modified difference index (MDI), showing that the Ke models have much larger effect on the calculational results than the λdry models. Besides, the calculation models containing the Côté and Konrad model for Ke were recommended to calculate the thermal conductivity of soils for its high accuracy.