Physically-based mathematical modelling of catchment sediment yield

Abstract

A physically-based, distributed sediment yield component has been developed for the SHE hydrological modelling system. This new component models the hillslope processes of soil detachment by raindrop impact, leaf drip impact and overland flow, and transport by overland flow. If the eroded soil reaches a river system it is routed downstream along with any inobilised river bed material. Deposition on land or in a river is simulated and the river bed material size distribution is continuously updated with allowance for armour layer development. The equation developed for soil detachment by raindrop and leaf drip impact was successfully tested using data from a field plot with a range of soybean canopy covers and rainfall intensities. The soil detachment coefficient in this equation was determined for a range of soil types and showed a variation consistent with that which may be expected from a consideration of the physics of a soils resistance to detachment. At present two soil detachment coefficients need calibration. In order to investigate the variation in these coefficient values, as well as to test the component, various applications were carried out. The hilislope sub-component was applied to rainfall simulator plots with a variety of surface conditions. Two sets of calibration parameters, distinguishable on a physical basis according to the degree of soil disturbance, were found to be appropriate for all the plots. To investigate scale effects, parameters calibrated at the rainfall simulator plot scale were transferred to a 1-ha rangeland sub-catchment. With no further calibration, the catchmerit response for four events was poorly simulated for both water and sediment. However, with reasonable variations in the antecedent soil moisture content but no variation in plot calibrated sediment parameters, the sediment yield for two of the four events could be successfully simulated. These applications suggest that parameter transfer is feasible if the sediment yield characteristics at the different scales are similar. Further applications of the hilislope sub-component were carried out for two small agricultural catchments. The sediment response could be simulated to at least the same accuracy as achieved by two existing distributed soil erosion models. The channel sub-component was applied to the East Fork River, Wyoming. Although the complex sediment storage/supply effects could not be reproduced completely, the simulated response was nevertheless of similar accuracy to that achieved by two existing alluvial river models. The new component is considered to be a valuable contribution to sediment yield modelling as a physically-based approach is used for both the hilislope and channel phases of the catchinent sediment system, within the framework of an advanced hydrological modelling system.