The thermal properties of soil play important roles in biogeochemical cycles. The soil thermal diffusivity can accurately reflect the transient process of soil heat conduction. In this study, we use observation data from the 5, 10, 20, 40, and 80 cm layers in Golmud from October 2012 to July 2013 and comprehensively compare the solution of soil thermal diffusivity thereafter. A new model is established using the thermal conduction-convection equation under Fourier boundary conditions. The results show that (1) the amplitude method and the phase method are based on a single temperature sine wave, which is used to describe the general soil, although the accuracy is not high enough; the logarithmic method and the arctangent method are performed four times a day, the accuracy of the obtained result is also low; moreover, the Laplace method does not have a clear soil temperature boundary function and thus can better address extreme weather effects or nonperiodic changes in soil temperature. (2) When solving the thermal conduction equation by a numerical method, format 2 (Crank-Nichalson-Sch format) is unconditionally stable, the data utilization is higher; in addition, the obtained soil thermal diffusivity is less discrete, and the result is more accurate. (3) When the soil temperature is simulated by the Fourier series, as the order n becomes larger, the result becomes more accurate. The Fourier series performs well in simulating the soil thermal properties. This study provides a useful tool for calculating soil thermal diffusivity, which may help to further characterize biogeochemical cycles.