Evaluation of Prediction Error in Some Particle Size Distribution Models for River Sediments

Particle size distribution is a paramount factor in characterization of sediments and may affect many of its properties such as hydraulic and electrical properties and characteristics related to the transportability by erosive agents. The aim of this study was to compare the capability of some mathematical models in describing particle size distribution of sediments trapped behind check-dams. Twenty sediment samples were taken from check-dams of four seasonal waterways in Anbi and Noshan regions of Urmia, Northwestern Iran. Full particle size distribution of sediments samples were determined by hydrometer method. Different aspects of models performance were evaluated by some efficiency criteria. Hierarchical cluster analysis of models based on selected efficiency criteria indicated that the Weibull, Revised-Weibull, Fredlund, van Genuchten, ORL and ONL are the best models in describing sediment particle size distribution. Among the selected models, Weibull was slightly superior with a mean coefficient of determination of 0.986. Efficiency of selected models, based on R2 value, improved with the increasing of sand content in the sediment samples. Relative error of the selected models was the lowest for particles with 0.005-2 mm in diameter, indicating the high performance of models to predict sand content of sediment samples.