Including geophysical data in ground water model inverse calibration.

Abstract

A nonlinear regression method is developed that can be used to estimate parameters of a ground waterflow model from a combination of observations of hydrological variables and observations of geophysical properties that are functionally related with the hydraulic conductivity. The procedure estimates: parameters characterizing the hydraulic conductivity field (e.g., zonal or pilot point values); geophysical properties that have been observed and that are functionally related with the hydraulic conductivity parameters; and a few parameters of the function that relates the hydraulic conductivity parameters with the geophysical properties (the type of function is assumed known). A fidelity factor, sigma(r)2, of a term of the minimized objective function reflects the faith one has in the validity of this functional relationship. The estimation methodology has been tested by means of synthetic models. The experimental results demonstrate that the number of estimated hydraulic conductivity parameters can be increased by adding geophysical observations to the set of hydrological observations that are traditionally used for model calibration. The improvement of the estimated hydraulic conductivity field and the simulated hydraulic head field can be significant but is dependent on the number, the locations, and the uncertainty of geophysical observations. The sensitivity of the estimation results to the value of sigma(r) is small for the studied problems except when the uncertainty of geophysical observations is high. In the latter case, a large sigma(r) value was found to be optimal to avoid that hydraulic conductivity estimates are closely tied to corresponding but highly uncertain geophysical observations.