Identification of an open canal: parametric black-box vs Muskingum approaches

In this paper, two different modelling approaches for open canals based on identification are proposed. One is based on parametric black-box models using classical identification methods, while the other is based on a grey-box approach using a Muskingum model structure. The first approach requires a previous identification of delay that can be determined using correlation or physical analysis. Once the delay is determined, model structure and parameter calibration can be determined using standard identification tools. The second approach establishes a fixed structure for the model and only parameter calibration is needed providing physical meaning. The two proposed approaches are applied on a 2 km open canal and their performances have been compared, emphasising their corresponding advantages and drawbacks.