Spatial groundwater dynamics of Zitny ostrov aquifer [Slovakia]

Utilization of methodical possibilities of the hydrodynamic theory of boundaries allowed to solve these problems and to build up the INKANS model, which simulates three dimensional flows in porous media accepting real geology and interaction with surface water streams. INKANS, used for simulation of the three dimensional seepage velocity field, estimates the impact of the hydropower project Gabcikovo on groundwaters of Danube Island accepting the interaction with the river Danube and with the whole (cca 200 km) canal network of the concerned area. Horizontal area of the Danube Island covering about 1450 square km was divided into quadrangles with 84x84=4032 nodal points. Vertical division considers 1 layer, giving 11x4032=44352 computational nodes. The equivalent FEM and FDM models would have to solve simultaneously the algebraic system of linear equations, the number of which amounted to 4x44352, i.e. more than 177,000. The matrix of such system involves more than 3x10E10 of elements, more than 700,000 of which are non-zero ones. The evident impact of the reservoir filling on the groundwater velocity field and free surface movement could be seen. The figure in the paper shows the vector field of horizontal streams - integral of velocity along vertical between bottom and free surface of water. Vectors located along the left boundary of model domain are clearly larger than others due to water level increase in the Gabcikovo reservoir and the seepage channels. The same effect is seen on another fig. showing the actual velocity field in the vertical cross section A-A'. The left vertical boundary is situated nearly by reservoir. The increase of free water surface after start of reservoir filling can be seen in fig. 4. The groundwater increase propagates horizontally by "wave velocity" of about 100 m per day. The points marked by sign x mean the monitored position of free water surface. From INKAS capability point of view it is important that comparison gives good results without any correction of input data by means of "model validation" which is current at many standard models