NUMERICAL MODELING OF INDUCED INFILTRATION AT A WELL FIELD IN SOUTHWESTERN OHIO

Especially in glaciated areas of the US, many public-supply wells are placed near rivers to enhance well yield by means of induced infiltration. An investigation of one such site was previously conducted at a public-supply well field in southwestern Ohio, adjacent to the Great Miami River. Water-quality properties were measured in the river, from two production wells and from sets of monitoring wells along paths of induced flow. Lowered specific conductance of river water during rain events resulted in lowered specific conductance at the monitoring wells at later times. Cross-correlation analysis of specific conductance yielded average lag times between these responses from which were inferred average ground-water traveltimes.

In this study, ground-water traveltimes and paths from the river to production wells are investigated with numerical modeling, and the results are compared to the lag-time analyses. Estimated traveltimes from different parts of the river exhibit large variability. For example, traveltimes to one of the production wells range from 12 and 800 days with a median of 30 days. Median traveltimes from particle tracking far exceed those reported from lag-time analyses, raising the question of what statistical measure is best to compare to observed lag times.

Pulse-source mass-transport simulations were done to better understand the meaning of the lag-time analysis and to more closely mimic the lowered-specific-conductance rain events. Traveltimes based on peak-concentration arrival times at monitoring were in closer agreement with the lag-time analyses. Simulated traveltimes to three wells were 6.8, 8.7 and 16 days while observed lag times were 5.7, 7.5 and 7.9 days; however, simulated peak concentrations at the monitoring wells were less 0.2 times the observed specific-conductance responses. The mass-transport simulations were used to recalibrate the model improving the agreement and lessening the uncertainty associated with the model-parameter values. Simulation and uncertainty analysis results indicate that such models may not account for small-scale heterogeneities or other processes that affect ground-water transport; however, these simulations provide insight into the nature of induced infiltration and the complex mechanisms of ground-water/surface-water interactions.