The majority of sediment leaving catchments may be from streambank failure. Seepage erosion of unconsolidated sand above a restrictive layer is an important erosion process in incised streams that leads to streambank failure by undercutting banks. The objective of this study was to determine the impact of soil properties on seepage erosion and the resulting streambank failure. Seepage flow and sediment concentrations were measured in situ at eight locations along the banks of a deeply incised stream in northern Mississippi. Using field observations as a guide, the soil profile conditions of a shallow (45 cm) streambank, consisting of 30 cm of topsoil, a 10 cm conductive layer, and a 5 cm restrictive layer, were mimicked in laboratory lysimeter experiments to quantify the hydrologic properties controlling seepage erosion and bank failure under a 40 cm head. The time to flow initiation and the flow rate were linearly related to the slope of the restrictive layer. Seepage erosion began within minutes of flow initiation and resulted in substantial (3 to 34 cm) undercutting of the bank. Sediment concentrations of seeps were as high as 660 g l − − − − −1 in situ and 4500 g l − − − − −1 in the lysimeters. Sediment concentrations were related to the layer slope, thereby indicating the importance of detailed site characterization. The USDA-ARS Streambank Stability model demonstrated the increase in instability of banks due to undercutting by seepage erosion, but failed to account for the sediment loss due to sapping for stable banks and overestimated the sediment loads for failed banks. Published in C a = ψ tan φ b (1) 448 G. V. Wilson et al.

Erosion by lateral, subsurface flow is known to erode streambank sediment in numerous geographical locations; however, the role of seepage erosion on mass failure of streambanks is not well understood. In the absence of an established sediment transport model for seepage erosion, the objectives of this research were to investigate the mechanisms of erosion due to concentrated, lateral subsurface flow and develop an empirical sediment transport model for seepage erosion of noncohesive sediment on near-vertical streambanks. Laboratory experiments were performed using a two-dimensional soil lysimeter of a reconstructed streambank profile packed with three different soil layers to mimic seepage erosion occurring at Little Topashaw Creek ͑LTC͒ in northern Mississippi. Soil samples from LTC streambanks indicated considerable hydraulic conductivity contrast between an overlying silt loam layer ͑SiL͒, highly permeable loamy sand, and confining clay loam layer. Lysimeter experiments were conducted with various upstream water table heads, overburden heights, and lysimeter slopes. Bank failure occurred prior to the total release of negative pore-water pressures in the SiL layer suggesting that such a mechanism was not critical for bank collapse due to seepage erosion. A seepage erosion transport model for conductive, noncohesive soil layers was derived based on a dimensionless sediment discharge and dimensionless seepage flow shear stress. The advantage of this sediment transport model is that it relates sediment flux to seepage discharge from the streambank.

No abstract