Sediment runoff from agricultural watersheds: Variability in suspended sediment transport due to rainfall

Considerable effort has been expended over the past two decades to control erosion from agricultural lands. Yet at the watershed scale, it has been difficult to document the effectiveness of land use practices aimed at reducing erosion. This is largely due to the enormous variability in suspended sediment data and the difficulty in explaining the sources of that variability. My research objective is to identify and quantify the sources of variability that exist in estimates of suspended sediment transport from agricultural watersheds in the Driftless Area of southwestern Wisconsin.;The current estimation techniques used to predict changes in suspended sediment transport in Wisconsin are based on regression analyses relating the characteristics of storm flow to suspended sediment transport. The methodology is statistically-based and limited due to the inability to incorporate spatial and temporal variability in the underlying physical processes of erosion and sediment transport. A more physically-based analysis is needed to explore potential sources of variance and identify key physical processes effecting the variability in sediment transport and analyses of the impacts of land use practices.;This issue is explored through the conceptualization of an approach and the development of a tool to address the spatial and temporal variability in sediment transport. The model is used to explore the effect of including spatial and temporal variability in rainfall on the sediment yield in a basin. Model results identify three sources of variability in the relationship between storm flows and sediment transport: the spatial variability in overbank flows, the spatial variability of antecedent sediment supply in channels, and the spatial variation in the intensity of rainfall.;The model is also used to assess the sensitivity of sediment transport to erosion parameters, channel geometry and rainfall intensity. The insights gathered from this modeling study highlight some important questions for further research.