Abstract: Stubble crushing machine has been widely used in the rotary tillage after wheat harvest. The wheat roots can often be simulated using the discrete element method. It is very necessary to calibrate the mechanical parameters for the better contact performance during crushing. In this study, a series simulations and tests were implemented to verify the wheat root system. The root system of Ningchun 4 wheat plant at mature stage was taken as the research object. The tensile and shear test of single root were carried out. The ultimate stress and elastic modulus of single root were calculated with the different water content (6%, 15% and 27%). The geometric morphology of the roots was characterized in the soil, together with the lateral roots at all levels. Each root was simplified to be straight and evenly distributed. The discrete element model was established for the single and whole root of wheat using the self-developed software, AgriDEM, according to the database of non-spherical particle filling ball modeling function. A flexible root model was established to bend at any position. The 'parallel bonding model' was used as the mechanical model of the connection between adjacent particles inside the roots. The Plackett-Burman and Box-Behnken simulation test were carried out, according to the tensile test and simulation of a single root. The bonding parameters of the 'parallel bonding model' were calibrated and then optimized. Because the soil type was the viscous and compressible loam in the wheat planting area, the 'clay particle model' was used to characterize the soil characteristics. According to the soil accumulation angle test and simulation, the parameters sensitive to the soil accumulation angle were selected by parameter sensitivity analysis. The linear fitting was used to obtain the static friction coefficient, sliding friction coefficient and adhesion coefficient, and the linear relationship between the above parameters, as well as the soil accumulation angle. There was the linear relationship among them. Finally, some parameters were determined for the 'clay particle model' in the soil simulation. The pull-out test of a single root buried in soil was carried out to verify the simulation. The results show that the relative error of the maximum pull-out force of a single root in the simulation and test was within 15%. There were the similar failure modes prone to occur under various test conditions. The high accuracy was also achieved in the single root bonding and soil parameters with the different water contents. There was the similar variation in the average maximum pull-out force after the test and simulation. The relative error between them was also within 15% within the range of theoretical calculation. The accuracy of the whole plant root model was also verified. The finding can provide a strong reference for the simulation analysis of the tillage and soil preparation after wheat harvest.