Efficient Data Assimilation Method for Optimizing Prediction of Flow Field and Sediment Transport in Open Channel

Accurately predicting open‐channel flow under complex conditions remains a challenge. This study introduces a novel synergetic self‐adaptive data assimilation (DA) framework, using a proportional‐integral‐derivative (PID) controller to dynamically calibrate the bed roughness parameter k s in a shallow‐water equation model (via the momentum principle). The PID‐based DA method updates the spatially distributed k s in real time (model run‐time) using observational data. The method was validated using flow data from a laboratory channel with five zones of different bed roughness. The feedback from the controller allows real‐time corrections of prediction errors without relying on linear assumptions or covariance matrices, enhancing the robustness for highly varied flow. Validation of the method using water level data showed excellent performance, achieving the coefficient of correlation and the normalized standard error for most nodes, even those not being directly assimilated. Further validation by including flow velocity as additional data showed beneficial gain, compared to DA using water level data only. Field applicability of the PID‐based DA method was demonstrated using flow and bedload data from the Inn River. This method outperformed the conventional Kalman filtering method in terms of prediction accuracy and efficiency. This method has a good potential for such practical applications as operational forecast of river floods and the development of mitigation strategies, quantification and management of river channel erosion, and cost‐effective operations and maintenance of instream structures.