Predicting streambank seepage flows: Sensitivity to soil properties and layering

Streambank failures result in loss of land and increased stream sediment loads. Variably saturated flow models integrated with bank stability models are being used to predict bank failure; however, understanding of the level of soil characterization necessary to predict when seepage contributes to bank failure is needed. ROSETTA, a pedotransfer function, was used to estimate hydraulic parameters for various levels of soil data (from texture class to saturated hydraulic conductivity, K_s). A two-dimensional groundwater flow code, SEEP/W, was used to model a hypothetical streambank and estimate seepage velocity. Results suggested that K_s and/or percent clay measurements are necessary to adequately predict seepage. A layered bank was also simulated with several combinations of soil types. Results indicated that only one order of magnitude difference in K_s (equivalent to a resistance ratio of 10) may be necessary to induce perching and seepage from the top layer. This research suggested that, if a field reconnaissance is performed to locate streambanks susceptible to perching, K_s and/or bulk density measurements are necessary to indicate the presence of a restrictive layer.