USING GROUNDWATER FLOW MODELS AND GEOCHEMICAL ANALYSES TO UNDERSTAND SURFACE WATER AND GROUNDWATER INTERACTIONS AT A HISTORIC SITE IN MILLEDGEVILLE, GA

A hydrologic analysis of a farm was conducted as part of the development of an automated surface and ground water-coupled GIS stream restoration design model. The site of the study is the historic Flannery O’Connor farm in Milledgeville, GA. This component of the project focused on understanding groundwater flow into and out of a pond located on the property. Groundwater and surface water interactions were monitored using piezometers and geochemical tracing. Phosphorous and nitrates were used as geochemical tracers of water exchanges between groundwater and surface water. Water levels were monitored with piezometers located in and around the lake, below an existing dam and along Tobler’s Creek. Hydraulic properties of the aquifer were estimated using data from slug tests and permeameters, while depth to bedrock was estimated using electrical resistivity. Creek piezometers and seepage meters were then used in the creek to locate places of up-welling or down-welling. A steady-state model for predicting the water table and groundwater flow from the pond to Tobler’s creek was created in MODFLOW. Particle tracking in MODPATH was used to determine flow directions and trace flow paths between the pond and the creek. Results from MODFLOW were compared with those from the Hydrology tool box in ArcGIS. Being able to get comparable results in ArcGIS would ease the coupling of the stream design model with the groundwater flow. Results of the study indicate that the pond does not receive much overland flow but is always full and losing water to the subsurface. The pond gains water directly from precipitation and from groundwater during rainfall events. Rain events cause the water table to rise and the pond to begin gaining. Groundwater flow out of the lake and the overflow pipe are the main ways the pond loses water to the stream. The flow to the stream creates areas of up-welling along a reach of the creek, which needs to be included in the stream restoration design model. Results from MODFLOW were used as inputs into the GIS stream design model to improve its performance and usefulness. Preliminary results from the ArcGIS hydrology tool have potential of being easily coupled to the stream design model because this tool is in the same program as the model. Since the Hydrology tool is easy to use the time involved in designing a fully coupled stream restoration will be greatly reduced.