Tag
The paper proposes a Physics Informed Recurrent Neural Network (PIRNN) that predicts unobservable physical variables to improve time series forecasting in physical processes, demonstrated through groundwater level predictions.