ABSTRACT
This study investigates the impact of hydraulic conductivity uncertainty on the sustainable management of the Lake Karla aquifer using the stochastic optimization approach. Lake Karla basin is an intensely cultivated agricultural region, which faces serious water deficit problems. In 1962, the former Lake Karla was drained causing a series of environmental problems in the area, with the aquifer over-exploitation and its dramatic water table drawdown being the most important. Nowadays projects for the reconstruction of the lake are being implemented. Currently it is the biggest environmental project in the Balkans and aims to reverse all the environmental consequences that have aroused during the last decades from the drainage of the lake. To reverse this situation it is foreseen that a great part of the total irrigation needs will be covered by the new reservoir, thus enabling local authorities to shut down lots of private irrigation wells.
The limited data concerning the spatially varying hydraulic conductivity parameter generates an uncertain hydrogeological environment. This makes the attempt to correctly estimate hydraulic heads and the volumetric budget of the aquifer even harder, either for the historical or for future period. This uncertainty is estimated using the Sequential Gaussian Simulation (SGSIM), for the conditional simulation of hydraulic conductivity. Conditioning is achieved by using the values of K, which are obtained through pump tests at 15 locations. Multiple realizations of the parameter are being generated and groundwater flow is simulated for each of them.
The main goal of the aquifer's sustainable management is to maximize the extracted groundwater volume, under the constraint of restoring the over exploited aquifer. The latter could be achieved by the rehabilitation of the water table to a previous, satisfying level. For that reason the optimization approach is achieved by solving the optimization problem for each one of the multiple stochastic realizations of the aquifer in a future period. The financial risk from water sales reflects hydraulic conductivity uncertainty.
In order to carry out this stochastic optimization procedure, a modeling system consisting of a series of interlinked models was developed: a hydrological model (UTHBAL), a groundwater model (MODFLOW), a lake-aquifer model (LAK3) and a reservoir operation model (UTHRL). The modeling system has been calibrated for the historical period against observed runoff and groundwater hydraulic heads. The Geostatistical Library (GSLIB) is used for the production of stochastic hydraulic conductivity maps and an optimization tool (GWM) is used for the optimization problem. The results prove that the proposed stochastic optimization framework can be a very useful tool for the groundwater sustainable management in an uncertain hydrogeological environment.
