Sufficient water storage in root zone during winter time not only benefits crop growth in next spring, but also potentially increases the yield. Water movement in root zone was often simulated by using the coupled water and heat transport (CWHT) model during soil freezing-thawing process and influenced by multiple factors such as boundary conditions, parameters determination method, and so on. An improved method based on in-situ measured data was proposed to optimally estimate the parameters in the CWHT model. The applicability of this method was validated as well. The hydraulic parameters in the model were estimated insitu by fitting soil freezing-thawing characteristic curve (SFC), which showed the relationship between the unfrozen soil water content (USWC) and soil temperature (ST). Then, the effect of surface evaporation (Es) on prediction accuracy in winter was evaluated. The field experiment was conducted over two winters (2011—2012 and 2012—2013) in an experimental farm at Changping County, Beijing, China. The USWC, ST and Es were monitored with dielectric tube sensors, temperature sensors and lysimeter, respectively. The data obtained in 2011—2012 were used to fit SFC to optimize the hydraulic parameters, whereas those in 2012—2013 were used to verify the model. The results showed that the model simulations were agreed well with the field measurements. When considering the effects of Es on USWC and ST, the root mean square errors (RMSE) of USWC and ST predictions in 2012—2013 were 0.046m3/m3 and 1.883℃ at 10cm and 0.071m3/m3 and 2.347℃ at 20cm, respectively. In contrast to the condition without Es, the RMSE for USWC and ST predictions in 2012—2013 were 0.059m3/m3 and 2.149℃ at 10cm and 0.081m3/m3 and 2.666℃ at 20cm, respectively. Thus, the improved method of parameters determination based on insitu measured data can ensure the prediction accuracy of the CWHT model. In addition, the accuracy could be further improved with considering the effects of surface evaporation and the effect of surface evaporation on USWC and ST was gradually decreased as the depth increased.