ABSTRACT
There are multiple interacting physical, chemical and biological parameters, which effect the shear strength of cohesive sediments. Still these processes are not fully understood and analytical approaches can hardly handle the involved uncertainties.
In this context, we used a data-driven neuro-fuzzy model to determine the critical shear stress and a stepwise regression algorithm to select model predictors systematically.
Our analysis of undisturbed sediment samples identified the clay content as the primarily controlling variable for the erosion resistance. Depending on the characteristics of the sampling location also the bulk density was selected as a model predictor. In comparison to analytical models that are available in scientific literature, the fuzzy model achieves higher correlation between measured and computed data. It requires less measured variables and less assumptions have to be met.
