Spatial and temporal infiltration dynamics during managed aquifer recharge

Natural groundwater recharge is inherently difficult to quantify and predict, largely because it comprises a series of processes that are spatially distributed and temporally variable. Infiltration ponds used for managed aquifer recharge (MAR) provide an opportunity to quantify recharge processes across multiple scales under semi-controlled conditions. We instrumented a 3-ha MAR infiltration pond to measure and compare infiltration patterns determined using whole-pond and point-specific methods. Whole-pond infiltration was determined by closing a transient water budget (accounting for inputs, outputs, and changes in storage), whereas point-specific infiltration rates were determined using heat as a tracer and time series analysis at eight locations in the base of the pond. Whole-pond infiltration, normalized for wetted area, rose rapidly to more than 1.0 m/d at the start of MAR operations (increasing as pond stage rose), was sustained at high rates for the next 40 d, and then decreased to less than 0.1 m/d by the end of the recharge season. Point-specific infiltration rates indicated high spatial and temporal variability, with the mean of measured values generally being lower than rates indicated by whole-pond calculations. Colocated measurements of head gradients within saturated soils below the pond were combined with infiltration rates to calculate soil hydraulic conductivity. Observations indicate a brief period of increasing saturated hydraulic conductivity, followed by a decrease of one to two orders of magnitude during the next 50 to 75 d. Locations indicating the most rapid infiltration shifted laterally during MAR operation, and we suggest that infiltration may function as a "variable source area" processes, conceptually similar to catchment runoff.