Hydraulic Conductivity Assessment within a Regional Ground Water Flow Model

Monday, April 12, 2010: 4:50 p.m.
Tabor Auditorium (Westin Tabor Center, Denver)
Tullia Bonomi , Department of Environmental Sciences, University of Milano-Bicocca, Milano, Italy
Paola Canepa , Department of Environmental Sciences, University of Milano-Bicocca, Milano, Italy
Francesca Del Rosso , Department of Environmental Sciences, University of Milano-Bicocca, Milano, Italy

Grid assessment, boundary condition settings and hydraulic conductivity evaluation are fundamental aspects in ground water flow simulation. This study was realized in order to evaluate the importance of detailed hydraulic conductivity distribution in multilayer heterogeneous aquifer flow models at regional scale.

Object of the study is the heterogeneous alluvial aquifer of Lombardy plain (Northern Italy) which extends from the Ticino river (westward) to the Oglio river (towards the East). Prealpine reliefs border the area to the North and the Po River delimits it to the South. The study area is 7000 km2; we discretized the area into 207 rows, 333 columns and 107 changeable thickness layers. Cells are 500 x 500 m. As boundary conditions we used the Ticino, Adda and Oglio rivers to act as head dependent flow boundary type (Cauchy); to the South, the Po river was used as specified head boundary type (Dirichlet). We fixed grid geometry and boundary conditions and we tested different hydraulic conductivity distributions.

The methodology adopted to rebuild heterogeneity is based on georeferenced stratigraphic log (about 16000 items of well data) organized in a specific database (TANGRAM), which permit a value of hydraulic conductivity (K) to be associated with every grain size class; K values were attributed according to literature. K values can be extracted from the database and interpolated with 3D geostatistical techniques. A range of different values of hydraulic conductivity was proposed for every different grain size class in order to interpolate different reasonable K values combinations, in a log –distribution.

Monthly piezometric levels (2001-2004) from Regional Monitoring Network were used as targets for calibration of numerical flow models, implemented by MODFLOW code. We used target analysis to compare results of different flow simulations and to define the influence of heterogeneity on simulations.