Estimating the hydraulic conductivity at the south oyster site from geophysical tomographic data using Bayesian Techniques based on the normal linear regression model

TLDR

The study investigates estimating hydraulic conductivity at the South Oyster Site using GPR and seismic tomography within a Bayesian framework. The authors developed a normal linear regression model within a Bayesian framework to relate GPR tomographic velocity, GPR attenuation, and seismic velocity to hydraulic conductivity, accommodating nonlinearities and uncertainty. The model shows that incorporating GPR and seismic velocity data modestly improves hydraulic conductivity estimates, especially when prior information is scarce, with GPR velocity and seismic velocity outperforming GPR attenuation.

Abstract

This study explores the use of ground penetrating radar (GPR) tomographic velocity, GPR tomographic attenuation, and seismic tomographic velocity for hydraulic conductivity estimation at the South Oyster Site, using a Bayesian framework. Since site‐specific relations between hydraulic conductivity and geophysical properties are often nonlinear and subject to a large degree of uncertainty such as at this site, we developed a normal linear regression model that allows exploring these relationships systematically. Although the log‐conductivity displays a small variation (σ 2 = 0.30) and the geophysical data vary over only a small range, results indicate that the geophysical data improve the estimates of the hydraulic conductivity. The improvement is the most significant where prior information is limited. Among the geophysical data, GPR and seismic velocity are more useful than GPR attenuation.