Full-Bayesian, simple zoned inversion and transmissivity assessment of the Edwards Aquifer, Texas, USA

摘要

An alternative approach for transmissivity (7) estimation based on hierarchical-Bayesian and Maximum Entropy methods is presented. A stochastic conceptual framework is adopted for the aquifers in question and perturbation techniques are utilized to render a coupled non-linear problem into a linear one. Bayesian updating is used to condition prior estimates of logarithm T [ln(7)] with T measurements; then hydraulic head measurements are incorporated into the updating procedure by adopting a linearized flow equation. This approach was applied to estimate the ln(7) field of the Edwards Aquifer. Simulations with a simple zoned ln(7) distribution were also performed. Simulations showed the head fits alone might not provide enough information to judge what parameter estimation approach is superior to the other and which estimated field approximates reality better; but the T field from Bayesian inversion is of higher resolution, and respects the geology and spring discharges better than the estimation from PEST.