In most, but the driest regions in the world, the dominant rivers are effluent; i.e., are gaining water from the ambient groundwater aquifer during the majority of seasons. However, there are instances where large volumes of waste discharge from an industrial plant create an `unnatural' industrial river, shifting the natural balance between the river stream and the surrounding groundwater aquifer in such a way that sections of the river may become an influent or loosing stream during extremely dry seasons, thus creating the potential for aquifer contamination. Hydraulically, for an effluent stream to become influent, the normally positive head gradient from the stream towards the aquifer must reverse itself intermittently.

Fig. 1: The connection between surface water and groundwater

The interaction of the industrial Fenholloway river which is loaded from the outfall of the Buckeye paper mill plant in Taylor county, Florida and the ambient groundwater aquifer are modeled by time series-, flow modeling- and particle tracking methods.

Fig. 2: The Fenholloway river

Fig. 3: The Buckeye paper mill plant
 

The major purpose of the modeling effort is to delineate the possible contamination corridor in the aquifer as may be caused by the infiltration of polluted water from the industrial river. The interrelations of precipitation, river discharge and groundwater data series are first analyzed by methods of structural time series, in order to quantify the interdependence of the groundwater table and river gauge heights, and thus to statistically examine the hydraulic possibility of aquifer contamination. The interaction of the stream and the groundwater is simulated by the USGS MODFLOW model. For the calibration, simulated water tables are compared with the historical records of monitoring wells in the adjacent aquifer. Lateral 'stagnant' points of the water flow based on the transient simulation are connected to delineate the maximal contamination corridor along both sides of the stream.
 

As a second approach, particle tracking simulations with varying water sources and forcing conditions are conducted to compute the dynamic movements of water particles out and along the river banks. A new semianalytical hydraulic stream-aquifer model has been developed for that purpose. Mathematically the model is based on a Green's function analytical solution of the 2-dimensional transient groundwater flow equation for the hydraulic head, using the method of images and the superposition principle. The method of images is used to incorporate assumed or measured river gauge heights as spatially and time-varying Dirichlet boundary conditions, whereas the superposition principle is employed to include the regional groundwater flow. The model encompasses a source/sink term that is related to the effective infiltration recharge of the groundwater aquifer. Initial conditions for hydraulic heads are specified from assumed or measured aquifer water-table elevations fitted by a Dupuit-Forchheimer functional relationship for the water-table change towards the river bed. Using Darcy's law, local flow velocities are computed from the Green's function solution for the head and integrated over time to provide total path-lengths of stream particles through the groundwater aquifer. The positions of the two extreme envelopes on both sides of the stream for a large number of river particles will then delineate an aquifer corridor along the river that, in case of `contaminated' river particles, may be adversely affected by the stream.
 

The sensitivity analysis shows that hydraulic conditions pertinent to the movement of river pollutants into the aquifer occur during long drought periods, due to a deficit of groundwater recharge which entails an overall LOWERING of the groundwater table. Various models are run using observed historical climatic and hydrological conditions. In order to establish possible `worst-case' contamination scenarios and to serve as a guideline for water-quality management, the model is also executed with extreme dry and wet year climatic conditions which, for Florida, are encountered between major El Niño occurrences; i.e. during La Niña years.

Fig.4: La Niña conditions                  Fig. 5: El Niño conditions

Under such exterme conditions the modeling results show that lateral migration of contaminants close to the industrial discharge point might be up to several km in cross-river direction during times when the normally gaining (effluent) stream becomes sectionally a loosing (influent) stream.
 

References:

Koch, M. and H.M. Cekirge, A transient Green's function analytical flow and particle tracking model to quantify a coupled river-aquifer system: Application to the assessment of possible groundwater contamination from a Floridan industrial river, In: Advances in Ground Water Pollution Control and Remediation, M.A. Aral (ed.), NATO ASI Series, Kluwer Academic Publishers, Dordrecht, pp. 127-154, 1996.
 

Sun, H., Koch, M. and Liu, X., Modeling of the interaction of river and groundwater to assess groundwater contamination by time series, flow modeling and particle tracking methods, In: Hydrogeology, Proceedings of the 30th International Geological Congress, Volume 22, Hydrogeology, Fei Jin and N.C. Krothe (eds), VSP, Utrecht, Netherlands, pp. 139-178, 1997, 335p.