Soil thermal conductivity is one of the most important parameters to determine the heat transfer performance of the soil layer, leading to the ground temperature distribution, soil environment, and crop growth. The composition of organic matter is directly related to the thermal conductivity of high organic soil. However, the current model of soil thermal conductivity cannot consider the organic matter content and decomposition degree. This study aims to analyze the influence of the undisturbed turfy soil on the thermal conductivity in the different layers. Additionally, more than 10 improved models of soil thermal conductivity were proposed and then compared on the turfy soil. The results indicate that: 1) The thermal conductivity in each layer of unfrozen turfy soil was similar (0.51~0.66 W/(m·K)). There was a significant difference in the thermal conductivity among the layers (1.00~1.62 W/(m·K)) after freezing, indicating that the freezing altered the composition of the soil. The higher proportion of components was found with the low heat transfer performance, due to more organic matter components and pores in turfy soil. The thermal conductivity of unfrozen turfy soil was lower than that of other organic soil with higher dry density. Most water in the soil was turned into ice after freezing, indicating the greatly improved thermal conductivity. The high content of water greatly contributed to the thermal conductivity of frozen turfy soil. Furthermore, a correlation analysis was carried out between the fundamental physical properties of turfy soil and the thermal conductivity. The soil particle size distribution, organic matter content, and decomposition degree depended mainly on the thermal conductivity of unfrozen turfy soil. 2) Most prediction models of soil thermal conductivity (Campbell, Johansen, and their derived models) failed to directly consider the proportion of organic matter components in the turfy soil, leading to overestimation of the thermal conductivity of organic matter in the solid phase. Alternatively, the soil thermal conductivity model was used to consider the dry density (Nikoosokhan model) and component weight (Tian model), indicating the excellent applicability to predict turfy soil. It indirectly quantified the low performance of heat transfer in the organic matter components and pores, according to the density differences after the calculation of soil thermal conductivity. The high level of accuracy was still difficult to achieve (RMSE>0.07 W/(m·K) for unfrozen soil; RMSE>0.28 W/(m·K) for frozen soil). 3) According to the soil properties, the parameters were introduced to characterize the turfy soil, including organic matter content (O_c) and decomposition degree (D_d), in order to improve the model of thermal conductivity. The improved model was obtained to comprehensively consider the low dry density, high water content, and high organic matter of turfy soil. The parameters were modified to reduce the overestimation of the thermal conductivity of organic matter components. Furthermore, a better prediction (R² > 0.75) of the soil thermal conductivity model was achieved for both unfrozen and frozen turfy soil with high organic matter, in terms of applicability and accuracy. The research findings can provide a strong theoretical reference for the thermophysical properties of seasonal frozen turfy soil with high organic matter in agricultural cultivation and engineering construction.