Crops adapt to variable soil moisture by varying the distribution of their roots and patterns of water absorption. Meanwhile, proper numerical model for predicting soil moisture in croplands especially under high soil moisture conditions is still under discussion. In this study, we evaluated a numerical model for predicting soil moisture in converted soybean fields. A soybean-growing experiment with a converted field model was conducted under the conditions of (a) G40, in which the groundwater level was at a depth of 40cm from the soil surface during the whole experimental period, and (b) G10-40, in which there was overmoisture during the early growing stage. Then, a numerical simulation of soil water movement for this growing experiment was conducted with the HYDRUS-1D model. At the end of this experiment, approximately 75% and 60% of the total root mass was concentrated in the topsoil layer (0-10cm depth) of G10-40 and G40, respectively; thus, the distribution of soybean root might be strongly affected by soil moisture at the early growing stage. The measured soil water consumption rate was relatively high around 30-90days after seeding in G40, while around 60-90days after seeding in G10-40. The simulation results improved with the consideration of root distribution variation, changes in evapotranspiration rate and LAI values. The dual-porosity model might be appropriate for describing the soil hydraulic characteristics of soils with macropores, but further improvement is needed in determination of the parameters for high soil moisture.