Using temperature modeling to investigate the temporal variability of riverbed hydraulic conductivity during storm events

SummaryUnderstanding the impact of storm events on riverbed hydraulic conductivity is crucial in assessing the efficacy of riverbank filtration as a water-treatment option. In this study, the variability of riverbed hydraulic conductivity and its correlation to river stage during storm events was investigated. Water levels and temperatures were continuously monitored in the river using creek piezometers screened beneath the riverbed, and monitoring wells located on the river bank. The range of values for water levels during the study period was from 161.3 to 163.7 m AMSL while temperatures ranged from 3.75 °C to 24 °C. During the duration of the study the Great Miami River was losing water to the underlying aquifer due to pumping in the adjacent municipal well field. Flow and heat transport were simulated in a groundwater heat and flow program VSH2D to determine the hydraulic conductivity of the riverbed. Hydraulic conductivity was estimated by using it as a calibration parameter to match simulated temperatures to observed temperatures in a monitoring well. Hydraulic heads in the aquifer responded to storm events at the same times but with dampened amplitudes compared to the river stage. The relative responses resulted in increased head gradients during the rising limb of the stage-hydrograph. Heat-flow modeling during five storm events demonstrated that a rise in head gradient alone was not sufficient to produce the temperature changes observed in the wells. Simulated temperatures were fitted to the observed data by varying both river stage (as measured in the field) and riverbed hydraulic conductivity. To produce the best fit temperatures, riverbed hydraulic conductivity consistently needed to be increased during the rising and peak stages of the storm events. The increased conductivity probably corresponds to a loss of fine sediments due to scour during high river stage. Hydraulic conductivity increases during storm events varied from a factor of two (0.0951–0.2195 m/d) to almost one order of magnitude (0.0007–0.00658 m/d). Despite these predicted changes the highest model-predicted hydraulic conductivity value was 0.66 m/d, which is still much lower than the infiltration rate used in sand filtration systems (3.59 m/d). These low values suggest that storm events do not pose a significant risk to the water quality at this well field. There was a direct correlation between the duration of rising limb, rate of change of stage and maximum river stage and the magnitude of change of riverbed hydraulic conductivity.