Submitted to: Journal of the American Water Resources Association
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/15/2003
Publication Date: 4/1/2004
Citation: Lopes, V.L., Canfield, H.E. 2004. Effects of watershed representation on runoff and sediment yield modeling. Journal of the American Water Resources Association 40(2): 311-319.
Interpretive Summary: One problem in predicting soil erosion is deciding how much spatial detail is necessary to represent the variability of runoff-erosion processes on a watershed. Conceptually, there are two mechanisms represented in erosion models that control sediment production and delivery from small watersheds: sediment eroded by raindrop impact, and sediment eroded by flowing water. These mechanisms are known to be spatially dependent. In this study we looked at the effect of watershed spatial representation on runoff and sediment yield predictions from a small watershed in a dryland environment. We used four levels of watershed representation varying from a detailed representation that all rills observable in the field to a less detailed representation excluding all but the largest channels. We found that runoff decreases as complexity in watershed representation decreases. For predictions of sediment production and delivery, however, the effect of spatial detail in watershed representation depends on the size of the particular runoff-generating storm event. For smaller events spatial detail does not affect sediment-yield predictions much. For larger events, such as those that occur at least once every two years or so, greater spatial detail in watershed representation is required in order to properly predict sediment yields and particle size distributions. This conclusion is relevant to contaminant transport predictions because contaminants tend to adsorb onto finer sediment particles.
Technical Abstract: A process-based, distributed runoff-erosion model (KINEROS2) was used to explore the effects of watershed geometric representation (i.e. channel network complexity) on storm runoff and sediment yield simulations in a semiarid environment. We used four different spatial representations of a 4.4 ha experimental watershed, with the most complex representation including all channels identifiable in the field. We concluded that oversimplified representations of watershed geometry greatly influence runoff and sediment yield prediction by inducing excessive infiltration of hillslopes and distorting runoff patterns and sediment fluxes. Runoff decreases systematically with increasing lumping, while sediment yield estimates vary significantly. Lumping had less impact on sediment yield simulations for small events than for large events. Evaluation of the effect of simplification on the particle size distribution of small and large events suggested that larger events entrain more of the coarser sediment stored in rills and channels than smaller events. As such, reducing complexity (i.e. increasing lumping) has the effect of replacing coarser grained channel sources with finer grained hillslope sources, so the sedigraphs from larger events are coarser for more complex geometries and finer for the less complex.