Stochastic simulation of aquifer heterogeneity in a layered basalt aquifer system, eastern Snake River Plain, Idaho

A stochastic approach to modeling aquifer heterogeneity in basalt lava flows was evaluated using different methods. Direct simulation of permeability was not justified with the limited amount of permeability data available. The lithology was interpreted from borehole geophysical logs, and three lithologic categories were used as permeability surrogates: massive basalt and fine-grained sediment, both of low permeability, and high-permeability interflow zones between lava flows. Sequential multiple indicator simulation of lithology was the first modeling method evaluated. Indicator semivariograms of lithology were described with a nested model. The horizontal ranges of the sediment and massive basalt categories were 140 m and 150 m, respectively, and isotropic. The maximum horizontal range of the interflow zones was 110 m, with a horizontal range anisotropy of 7:1, elongated east-west. A 1.74 km2 by 60-mthick volume of aquifer was discretized on a 6 × 6 × 0.6 m grid and simulations were conditioned to borehole lithology. A major limitation of this approach is that permeabilities assigned to the simulated lithologies are not conditional to available well-test permeability data. The second approach was simulated annealing of well-scale permeability, conditioned to limited well-test data and to previously simulated lithology, with a vertical resolution matching the average 6 m measurement support of well-test data. A relationship was demonstrated between well-test permeability (Kb) and number of interflow zones, NZ, in the well-test interval, thereby allowing both lithologic data and simulated lithologic structure to be used as additional conditioning information. The Kb-NZ relationship indicated that interflow zone permeability is lognormal with a relatively small variance. A preliminary structure-imitating model was developed to investigate the feasibility of a variant of an object-based simulation approach. The model mimics the physical geologic structure created by processes such as eruption, inflation, ponding, and sedimentation, by optimizing specified rules and parameters that produce stochastic realizations which reproduce the morphologic and geometric characteristics of lava flows.