An Experimental Design Based Methodology for Calibrating 2D Hidromorphodynamic Numerical Models for Rivers

Abstract

The confidence level of a numerical model is strongly related to the calibration and validation processes and the quality of their adjustment. Typically, numerical 2D models for rivers that involve hydrodynamics, sediment transport and morphology processes are calibrated sequentially under the assumption that the morphology changes do not significantly affect the hydrodynamics. However, in some cases, the morphologic processes modify substantially the hydrodynamic due to the quick bed level accession and erosion as a function of the hydrological regime. These changes require a more efficient procedure reduce the amount of scenarios with indicators that allow to asset the quality of the calibration. This paper shows a methodology proposed to take into account the interaction of hydrodynamic, sedimentology and morphology in rivers based on an experimental design that allow evaluating simultaneously the significance and the sensitivity of the variables in order to select the best set of scenarios based on quantitative and qualitative indicators. The methodology was implemented in the Magdalena River and the Meta River in Colombia, which have different hydromorphodynamic conditions. The calibration and validation was supported with field data of water elevation, velocity vectors obtained by ADCP, flow rate measurements, sediment transport, and bed level changes, showing that is applicable either to straight, meandered, or braided rivers.