In the study of grain boundary migration of metallic materials using molecular dynamics simulation (MDS), grain boundary mobilities and activation energies are often found to be different from experimentally observed values. To reconcile the discrepancies, tremendous effort has been made to replicate experiment conditions in MDS, e.g.as low a driving force as possible, near zero grain boundary velocity. In the present study, we propose an analytic method that removes effects from non-physical conditions such as high driving force or high temperature. The analytic model presumes that two types of rate limiting events coexist during grain boundary migration. Kinetics parameters, such as activation energies, of the rare events are different and therefore should be modeled separately. Activation energies from this model are closer to experiment than previously reported values. Further, by analyzing the evolution of atomic structures, these two types of rate limiting events correspond to shear coupled migration and grain boundary sliding mechanisms, respectively.