Improved method for modelling sediment transport on unpaved roads using KINEROS2 and dynamic erodibility

Abstract

AbstractOne difficulty in modelling sediment transport on unpaved road surfaces stems from the inability of physically based models to simulate the flush of loose surface material that is deposited on the road surface prior to the onset of a storm. This work builds upon a prior modelling methodology (referred to as dynamic erodibility) to simulate time‐varying sediment transport on unpaved mountain roads by loosely coupling a continuous, exponential decay disturbance model with the erosion algorithm in KINEROS2. The method is tested against sediment transport time‐series observed on small‐scale rainfall simulation plots for various slope, antecedent soil wetness and pre‐storm sediment availability conditions. The new method generally improves prediction errors of total sediment output, peak sediment output and fit of the sediment output time‐series. However, for some validation events, the method fails to simulate high initial sediment spikes. This limitation may be a side‐effect of using data from small‐scale plots, but also may signify the need for additional calibration of the disturbance model subcomponent with data from surfaces having a greater variety of pre‐event surface material. Nevertheless, this road erosion modelling approach provides a realistic ‘description’ of time‐varying sediment transport, which is controlled both by the baseline erodibility of the underlying road surface, and importantly, by the removal of a loose, surficial sediment layer by overland flow. Copyright © 2002 John Wiley & Sons, Ltd.