Comparative assessment of a mountain river flow resistance – 1D-: sensitivity and prediction using data-based approaches

Abstract

Mountain -rivers are, by far, the most challenging case to model because of its bed characteristics and their energy dissipation mechanisms depending on its irregular morphology. Resistance, roughness, or friction parameter are equivalent terms. It plays an important role in 1-D open channel models to estimate different variables. Moreover, this parameter contains all the dissipative processes in a mountain river, and it is usually valued through field measurements, existing different methodologies to estimate it. Consequently, it is essential to determine which methodology is the most adequate to predict it. The resistance parameter determined in field is not always the same as the one used in a hydrodynamic model. In this thesis; cascades, plane bed, and step-pool has been studied in the Quinuas river (Ecuador). “Non-dimensional hydraulic geometry equations” (NDHG) were the best option to predict velocity in all the mountain river reaches. The parameters of NDHG varies depending on the author, therefore a methodology based on some field measurements to estimate the NDHG parameters was developed. The differences between model and field resistance coefficient depends on the morphology and flow magnitude. A machine learning technique using the system physics was develop providing optimal results to predict water depths and to calibrate resistance parameter.