REPRESENTATIVE SAMPLING OF FLUXES IN CARBONATE AQUIFERS

The triple porosity (matrix, fracture, channel) attributes of carbonate aquifers provide challenges to both representative sampling and to an accurate characterization of them. The principal problem is often conceptual because limited sampling and testing may not reveal the true nature of these aquifers. The ideal approach is to consider each of the three components separately and then to combine the results. The matrix typically accounts for most storage in carbonate aquifers but has low hydraulic conductivity (K). Rock cores provide an ideal method of sampling the matrix. At the other extreme, channels and conduits constitute a high-K network that discharges to springs, but wide spacing of conduits make wells less than ideal for sampling them. Nevertheless, downhole tests using video and flow meters often reveal the importance of channeling. Furthermore, both traditional sink to spring tracer tests and traces to and from wells are ideal for establishing flow direction and flow velocity. Additionally, these tests often provide irrefutable evidence that an aquifer has an interconnected network of solutionally-enlarged fractures (i.e. is karstic). Fractures provide the third porosity element; their porosity and K are both intermediate between matrix and channel values and are best sampled with well tests.

Combining the results from the three porosity elements reveals several interesting effects. First, there can be substantial matrix diffusion and substantial ages in the matrix for environmental isotopes. Second, residence time distributions have very high standard deviations, and this becomes apparent when environmental isotope ages and ages of introduced tracers are compared. Third, porous medium approximations are often adequate for flow but not for transport. Comprehensive testing from sites in Kentucky, Florida, Texas, Ontario, Mexico and Turkey provide examples of how the triple-porosity nature of carbonate aquifers has been successfully characterized and incorporated into numerical models.