Data assimilation integrates the prior information from numerical model simulations with observational data to achieve an optimal representation of dynamical systems and their uncertainties. High-resolution determination of the crustal stress field is crucial for monitoring and early warning of geodynamic activities. In current numerical model simulations, displacement is typically the only state used. Since the crustal stress field is derived from the spatial derivatives of the displacement field, this approach results in significant errors in simulating the crustal stress field. To enhance the resolution of crustal stress field simulations, an ideal model experiment is established incorporating the velocity-stress wave equation as the model state, and the finite difference method is employed along with a Kalman filter data assimilation for the crustal-stress-field determination. The ideal experiments for 1D P-wave propagation show that the accuracy of the assimilating stress from the model state of velocity-stress vector is much better than that from the model state of displacement-velocity vector, which means that data assimilation could expand the application of the crustal stress fields in monitoring and early warning of earthquake and landslide.