In this paper, firstly, new version of frozen soil model is presented, in which the enthalpy and total volumetric water equivalent content in soil is used as predictive variables to replace the predictive variables of soil temperature and volumetric liquid water content in the current models respectively. It makes the required estimation of liquid-ice phase change rate in the current models unnecessary, and therefore avoids the introduction of the error in estimation of the phase change rate into the new model and in turn will improve the new version model in both speeding convergence of numerical iterative process and enhancing the model performance. At the same time, a numerical scheme is also developed for the diagnostic equations in the new version, which is very efficient for finding the true solution with a few iteration times. It's proved the numerical simulation results of the new version model are in good agreement with two field observed data. Besides, based on analyzing three schemes for soil freezing-thaw from theoretical view and comparing the numerical simulation results, it is found that the parameterization scheme, which takes the relation between moisture potential and temperature in soil derived from the equilibrium thermodynamics theory into consideration, can describe the freezing-thaw processes well. Numerical results have shown that different freezing-thaw parameterizations have greater influence on simulated soil temperature, water content, and latent and sensible heat fluxes at soil surface.