Land-atmosphere interaction is one of the important causes of climate change. The accurate characterization of physical mechanism of the soil layer can help us better understand the land-atmosphere interaction. As one of the important thermal properties of soil, soil thermal diffusivity is an important parameter reflecting soil thermal conductivity and a necessary condition for simulating soil thermodynamic environment. Based on the comparison of several methods for calculating soil thermal diffusivity (amplitude method, phase method, logarithmic method, arctangent method, Laplace method and numerical methods), this paper develops a new soil thermal diffusivity model (Thermal conduction-convection method based on Fourier series boundary condition for calculating soil thermal diffusivity, TCCMF) and compares the results of the new model with other methods. The results show that: (1) The accuracy of the results obtained by the amplitude method, phase method, logarithmic method, and arctangent method is not high enough; the Laplace method can better solve the effects of extreme weather or non-periodic changes in soil temperature. (2) When the numerical methods are used to solve the thermal conduction equation, the Crank-Nichalson-Sch format is unconditionally stable, the data utilization rate is higher, the obtained soil thermal diffusivity is less discrete, and the result is more accurate. (3) The new model (TCCMF) uses the Fourier series method, which is easy to calculate, and has more complete physical process and more flexible precision control. The simulation results of soil temperature using this method show that this method (TCCMF) has better simulation accuracy, indicating that it has potential value in simulating soil thermodynamic characteristics and land-atmosphere interaction.