Evaluation of infiltration from losing-disconnected rivers using a geophysical characterisation of the riverbed and a simplified infiltration model

• We estimated recharge from a losing-disconnected stream.
• Electrical resistivity surveys were used to characterise the riverbed clay layer.
• Simple model able to reproduce results of numerical model.
• Hydraulic conductivity of clay greatest source of uncertainty.

SummaryDespite their significance to alluvial aquifer water balances, there are few field-derived estimates of infiltration from losing-disconnected rivers. Infiltration was estimated over a 2 km section of Billabong Creek (Murray-Darling Basin, Australia) using a combination of field measurements and modelling techniques. A method was developed whereby in-river and riverbank electrical resistivity surveys were used to map the spatial coverage and thickness of the riverbed clay layer, thought to be the key control on infiltration in this environment. Inverted resistivity measurements were used to generate a surrogate for a spatial map of clay layer thickness, with constraint provided by coring the riverbed at selected high and low resistivity locations to independently determine the thickness of the clay. The survey showed that the clay layer was continuous across the study reach but varied in thickness (∼1 m to >4 m). A simple infiltration model was developed for infiltration through a clay layer and shown to be accurate under steady state conditions when compared to estimates obtained using a variable saturation numerical model for idealised riverbed cross-sections. Infiltration rates across the study reach were estimated to range between 1700 and 7800 m3 km−1 year−1, with an average of 3400 m3 km−1 year−1. A sensitivity analysis showed that infiltration rates were most sensitive to clay hydraulic conductivity, not clay layer thickness. However, it is anticipated that when applied at a larger scale (10–100 km), infiltration rates will be more sensitive to the presence or absence of a clay layer in the riverbed. The proposed methodology can provide independent estimates of infiltration in losing-disconnected rivers at a scale suitable for the calibration of regional groundwater models.