Ground-water flow model of the Boone formation at the Tar Creek superfund site, Oklahoma and Kansas

Extensive mining activities conducted at the Tar Creek Superfund site, one of the largest Superfund sites in the United States, pose substantial health and safety risks. Mining activities removed a total of about 6,000,000 tons of lead and zinc by 1949. To evaluate the effect of this mining on the ground-water flow, a MODFLOW 2000 digital model has been developed to simulate ground-water flow in the carbonate formations of Mississippian age underlying the Tar Creek Superfund site. The model consists of three layers of variable thickness and a grid of 580 rows by 680 columns of cells 164 feet (50 meters) on a side. Model flux boundary conditions are specified for rivers and general head boundaries along the northern boundary of the Boone Formation. Selected cells in layer 1 are simulated as drain cells. Model calibration has been performed to minimize the difference between simulated and observed water levels in the Boone Formation. Hydraulic conductivity values specified during calibration range from 1.3 to 35 feet per day for the Boone Formation with the larger values occurring along the axis of the Miami Syncline where horizontal anisotropy is specified as 10 to 1. Hydraulic conductivity associated with the mine void is set at 50,000 feet per day and a specific yield of 1.0 is specified to represent that the mine void is filled completely with water. Residuals (the difference between measured and simulated ground-water altitudes) has a root-mean-squared value of 8.53 feet and an absolute mean value of 7.29 feet for 17 observed values of water levels in the Boone Formation.

The utility of the model for simulating and evaluating the possible consequences of remediation activities has been demonstrated. The model was used to simulate the emplacement of chat (mine waste consisting of fines and fragments of chert) back into the mine. Scenarios using 1,800,000 and 6,500,000 tons of chat were run. Hydraulic conductivity was reduced from 50,000 feet per day to 35 feet per day in the model cells corresponding to chat emplacement locations. A comparison of the simulated baseline conditions and conditions after simulated chat emplacement revealed little change in water levels, drainage and stream flux, and ground-water flow velocity.

Using the calibrated flow model, particle tracks were simulated using MODPATH to evaluate the simultaneous movement of particles with water in the vicinity of four potential sites at which various volumes of chat might be emplaced in the underground mine workings as part of potential remediation efforts at the site. Particle tracks were generated to follow the rate and direction of water movement for a simulated period of 100 years. In general, chat emplacement had minimal effect on the direction and rate of movement when compared to baseline (current) flow conditions. Water-level differences between baseline and chat-emplacement scenarios showed declines as much as 2 to 3 feet in areas immediately downgradient from the chat emplacement cells and little or no head change upgradient. Chat emplacements had minimal effect on changes in surfacewater flux with the largest simulated difference in one cell between baseline and chat emplacement scenarios being about 3.5 gallons per minute.