Heavy metal contamination is a common and complex problem at abandoned mine sites. In Arkansas alone, there are approximately 5000 acres of ‘affected’ mine lands (abandonedmines.gov 2013). These sites can have significant interaction with both surface water and local aquifers. Hard rock mining in particular interacts with groundwater in the form of mine tunnels and mine workings. These tunnels and shafts can change the flow regime of the aquifer, change the oxidation potential, and cause conditions leading to the spread of metal contaminants. When the flow is already complex, such as in karst environments, aquifer and surface water contamination can be difficult to predict. This is further complicated in the interflow zone surrounding a river, and this zone can undergo rapid changes in flow direction and water quality. The Rush Mining District in Northern Arkansas combines both these conditions. A Mississippi-type sulfide deposit, the ore body is located on the shore of the Buffalo National River. The river height has a strong seasonal fluctuation, with individual storm events producing rapid changes in water level. The Everton Dolomite hosts the ore body and is heavily karstified. The ore itself, while highly concentrated in this area, is distributed throughout the Everton (McKnight, 1935) and can cause contamination in areas without mining. This locality was mined primarily for zinc in the form of sphalerite (ZnS) and was closed in the early 1960s. Tracing the transport of zinc and cadmium and modeling the changes in flow direction, geochemistry, and water source over time is the purpose of this project.

Water samples will be obtained from surface water bodies and through monitoring wells. Dye tracing and groundwater modeling (MODFLOW) will be used to simulate the groundwater flow system. Isotope geochemistry of H and O will also be used to identify and separate the water sources. Geochemical analyses of water and precipitated minerals will be acquired from water and solid samples. The geochemical system will be modeled using these inputs and regional data in PHREEQC.