Determination of effective hydrological parameters using experimental stratigraphy

Critical problems facing basin-scale groundwater flow and solute transport studies are the accuracy of geologic framework models and the selection of effective hydrologic parameters. A stratigraphic image was scaled-up to create a synthetic basin-scale hydraulic conductivity (K) map. For selected deposits representative of different depositional environments, numerical, analytical and stochastic methods were used to estimate an effective hydraulic conductivity (K*) and macro-dispersivity (A). Three model representations of the basin were developed: a fully heterogeneous model (FHM) and two geological framework models (GFM). For each model, steady-state topography-driven groundwater flow was computed in the basin in which the transport of a conservative tracer was simulated with random walk particle tracking. We conclude that the analytic-stochastic model prediction of K* becomes accurate when a statistical homogeneity is identified. K* exhibits "scale effect" due to heterogeneity; asymptotic values were reached after data support exceeds one ln(K) correlation range. Asymptotic behavior in longitudinal macro-dispersivity exists only in stratified deposits where the problem scale is greater than the In(K) correlation range. The impact of local dispersion on solute macro-dispersion and the accuracy of tracer-test-inferred K* is also problem scale dependent. In the full-basin simulations, the hydraulic head is not sensitive to heterogeneity; different model representations gave similar head field (non-uniqueness). Regional flow pattern becomes progressively inaccurate as the number of stratigraphic units is reduced. In all models, the basin-scale groundwater velocity is non-stationary and asymptotic behavior in solute spreading is not observed. In the FHM, the tracer plume exhibits alternative expansion and contraction over time. Velocity computed by the GFM is too smooth to capture the detailed solute distributions predicted by the FHM. In summary, reproducing mean flow and transport behavior at the sedimentary basin scale is the most sensitive to the selection of the GFM.