Flow velocity prediction in a vegetated channel using soft computing techniques

Flow vegetation interaction is a complex process affected by physical parameters, such as flow depth, vegetation height, stem characteristics, vegetation density, and submergence conditions. Corelation matrix of dependent and independent variables shows a direct corelation of flow velocity with vegetation density, the height of the vegetation, flow depth and the slope of the channel, and an inverse relation with non-dimensional coefficient of drag, the diameter of the cylindrical vegetation. In this study, three soft computing techniques, namely, group method data handling (GMDH), optimizable Gaussian process regression (GPR), and ensemble tree (ET–B) models with Bayesian optimization, are employed to precisely estimate the flow velocity in vegetated channel. These methods have effective data mining capacity, making them suitable for predicting flow velocity considering many independent parameters. Taylor diagram of results shows superior performance of GMDH (R = 0.982, RMSD = 0.054 for testing) over GPR–B (R = 0.969, RMSD = 0.069 for testing) and ET–B (R = 0.927, RMSD = 0.1 for testing). The convergence of ET–B is faster than GMDH and GPR–B, making it a more computationally efficient model.