Stochastic Forward and Inverse Modeling: The “Hydrogeophysical” Challenge

Abstract

Successful integration of geophysical and hydrogeological datasets represents a recent and major breakthrough in hydrogeological site characterization. As discussed in Chapter 1 of this volume, the value of integrating these datasets for characterization lies in the extensive spatial coverage offered by geophysical techniques and in their ability to sample the subsurface in a minimally invasive manner. However, this breakthrough is associated with a few difficulties. One difficulty resides in the non-unique relationships that sometimes exist between hydrogeological and geophysical attributes; integration of hydrogeological and geophysical data under non-unique conditions has been investigated by Rubin et al. (1992), Copty et al. (1993), Hubbard et al. (1997) and Hubbard and Rubin (2000). This non-uniqueness can exist even under idealized conditions of error-free measurements in natural systems comprised of multiple hydrogeologically significant units (i.e., Prasad, 2003), and it is only exacerbated by measurement errors. In applications, the situation becomes even more difficult because the rock type at the location associated with the geophysical attribute is almost always unknown, and thus the applicable petrophysical model is also almost always unknown. Another difficulty stems from the disparity between the spatial resolution of the geophysical attributes and the scale that characterizes the hydrogeological attributes, collected for example, through boreholes (c.f., Ezzedine et al., 1999). This scale disparity hinders efforts to develop unique and accurate relations between the two types of measurements, and introduces another source of uncertainty.