Author(s): Sebastian Schwindt; Paul Jacques Grigull; Beatriz Negreiros; Stefan Haun
Linked Author(s): Sebastian Schwindt, Stefan Haun
Keywords: No Keywords
Abstract: Numerical modeling of hydro-morphodynamic processes, encompassing fluvial sediment erosion, transport, and deposition, involves mathematical simplifications translated into adjustable parameters for each case study. This calibration process aims to achieve accurate simulations of measured quantities. Before utilizing a numerical model to predict hydrodynamic or morphodynamic changes in rivers or lakes, additional model validation is required. Validating model quality entails statistically quantifying the accuracy of modeled outputs compared to measured data. A satisfactory validation of a morphodynamic model, focusing on sediment transport, relies on correctly reproducing landscape patterns within model boundary conditions. However, a direct pixel-by-pixel quantification of measured topographic changes may indicate poor model quality despite achieving correct landscape patterns. Thus, the challenge arises of objectively distinguishing morphodynamic models capable of predicting morphological patterns from entirely nonsense models. To address this challenge, a recently developed fuzzy logic approach incorporates blur to pardon some degree of inaccuracy for a fuzzy-statistical assessment of morphodynamic model goodness. In this study, we investigated the performance of fuzzy-logic-based model quality assessment and human perception. Through an online survey, amateurs, somewhat experienced individuals, and experts were asked to subjectively rate correlation levels in six pairs of maps. Expert opinions exhibited increasing agreement with calculated fuzzy correlations, contrary to biases observed in other disciplines. Also, in the survey, experts confirmed the significance of fuzzy-logic metrics for hydro-morphodynamic model validation, enabling precise assessments while considering the inherent impreciseness of these models.
Year: 2024