Accurate prediction and monitoring of vadose zone soil water content (VZSWC) is important yet challenging for sustainable water resources management. While machine learning algorithms have shown promise, given limited and noisy subsurface data, they often struggle with overfitting, poor generalization, and physical inconsistency. In this study, we explore the effectiveness of physics-informed neural networks (PINNs) in addressing these challenges by presenting PINN-SM, a physics-informed neural network model tailored for predicting physically consistent VZSWC profiles by incorporating Richardson’s equation as the governing partial differential equation. Our model extends a traditional PINN architecture through inclusion of process-informed input variables and an attention mechanism that captures the time lag in the response of VZSWC to rainfall events. Using data from a monitoring station in Austin, Texas with soil moisture sensors at different depths, PINN-SM demonstrated superior predictive performance compared to the traditional PINN model, reducing RMSE by 80% across all depths. Furthermore, PINN-SM outperformed conventional artificial neural networks (ANNs), achieving approximately 25% lower RMSE across all depths. This performance advantage was particularly pronounced in peak event prediction, where PINN-SM achieved around 50% lower RMSE across all depths. Results indicate that compared to conventional ANNs, PINN-SM can more effectively capture underlying patterns while avoiding overfitting and excel at extreme events’ predictions. By integrating physical constraints with meteorological inputs and modeling temporal dependencies through attention mechanisms, PINN-SM represents a significant advancement in VZSWC prediction, with potential applications in improving subsurface hydrological monitoring and satellite-based forecasting systems.