Working Toward a Better Representation of Heterogeneities in Regional Groundwater Flow Systems: a Three Dimensional Approach

Groundwater flow models have been increasingly used for multiple purposes, be it evaluation and management of water resources, evaluation of radioactive waste repositories, and others. However, a number of issues affecting the accuracy of obtained results have been a major source of concern and object of discussion over the years. Among these issues are problems associated with the nonuniqueness of model solutions, as well as proper representation of heterogeneities in different formations, particularly in complex regional systems. Another major question that remains unanswered is to whether 2D models are representative of real three-dimensional systems. Castro and Goblet (2003) have recently shown in a detailed 2D model study, conducted in the Carrizo aquifer and surrounding formations of south Texas, that regional groundwater flow systems present a high degree of freedom concerning hydraulic conductivity for models calibrated on measured hydraulic heads. In particular, Carrizo aquifer and overlying confining layer hydraulic conductivities that vary by up to two orders of magnitude lead to similar calculated hydraulic heads. We were interested in knowing if a more detailed representation of this regional system through a 3-dimensional model could place additional constraints and reduce the nonuniqueness problem associated with 2D models. The question of whether a 3-dimensional representation of the system can shed further light on to the degree of heterogeneity present in the different formations is also one of interest. Indeed, and like many other sedimentary sequences, the complexity of depositional processes for the Carrizo aquifer and surrounding formations in a mixture of deltaic, fluvial and marine settings has resulted in a high degree of lithologic heterogeneity. To address some of these questions, we proceeded to the expansion of the initial 2D model (Castro and Goblet, 2003) through successive projections of the initial mesh into the horizontal direction. The resulting 3D model covers an area of ∼ 7,000 Km², it comprises 233 planes, and more than 5 million elements (pentahedrals). Preliminary results show that a high degree of freedom to calibrate the model on hydraulic heads remains, confirming results previously obtained through 2D simulations. Thus, 3D models do not seem to contribute to a reduction of the nonuniqueness problem. By contrast, our results indicate that 3D models are particularly useful in better identifying the presence of heterogeneities within different formations. In particular, 3D simulations show evidence for the presence of variations in hydraulic conductivities not only along the x direction, where decreasing values from northwest to the southeast are observed, but also the presence of a distinctive corridor on the eastern part of the domain where hydraulic conductivity decreases more rapidly, possibly indicating the presence of thicker mixed-load mud-rich alluvial deposits in the area. Three-dimensional models can thus be particularly useful in identifying the presence of heterogeneities in complex regional groundwater flow systems, thus, allowing for a better representation of real systems. Castro M. C. and Goblet P. (2003), Calibration of regional groundwater flow models - working toward a better understanding of site-specific systems, Water Resources Research, 39(6), 1172, doi:10.1029/2002WR001653.