Longitudinal Fluvial Dispersion of Coarse Particles: Insights From Field Observations and Model Simulations

Abstract

In this study, we use field observations augmented with model simulations to examine gravel dispersion over nine years (2007–2015) in Halfmoon Creek. The observations of flow, entrainment, and dispersion were used to develop a forward model utilizing the Einstein‐Hubbell‐Sayre (EHS) compound Poisson process. The observed mean virtual velocity of the tracer population slows down with cumulative excess energy after the 2010 large event. The forward model deviates from the observations in representation of tails, overpredicts mean displacements, and shows a narrower spatial distribution. The heavy‐tailed resting times indicate prolonged immobilization of some grains, suggesting the preferential movement of other most mobile grains. As such, 34% of most mobile grains constitute 50% of the total entrainments. The consideration of preferential movement explains the longitudinal spread but still overpredicts the displacement after the 2010 event. The model was then explored to consider additional transport‐related mechanisms causing deviations, such as reduction in virtual velocity, entrainment probability, and morphological trapping of meander bends, which helps to adequately recreate the observed dispersive behavior. The available historical flow records used for simulating dispersive behavior over multiple decades reveal an abrupt increase in displacements for exceptionally large events, suggesting the exhumation of deeply buried grains back in transport. The simulation results highlight the need for tracer studies with large sample sizes and improved recovery rates for longer time frames experiencing floods of widely varying magnitudes. Such models, inspired by Einstein's stochastic theory can be valuable for various river research applications.