Determination of Porosity - Hydraulic Conductivity Relationship Using High-Resolution Geophysical Data and Tracer Test Measurements to Improve Hydrological Predictions

Knowledge of the spatial distribution of hydrological properties at a range of spatial scales is critical for accurate modeling of subsurface contaminant transport. Such knowledge can be obtained using a number of techniques, among them hydrological tests and geophysical measurements. Individually, the data from each of these techniques have a certain capacity to improve the predictability of subsurface hydrological behavior. However, only through integration of these various data types can greatly improved subsurface property models be found that are consistent with all available sources of information. Recently, we have shown that the integration of crosshole ground-penetrating radar (GPR) and borehole porosity log data using conditional stochastic simulations can significantly improve hydrological predictions when it is assumed that the "true" small-scale relationship between hydraulic conductivity (K) and porosity (φ) is well known. This was demonstrated for a variety of hypothetical saturated alluvial aquifers exhibiting varying degrees of continuity and structural complexity. Although the results of this work clearly showed that, by "recapturing" small-scale heterogeneity using geophysics and making use of the correct petrophysical relationship at that scale, we can build much improved hydrological models, it is concerning that (i) such a relationship in practice is unknown, and further scale- and site-dependent, and (ii) the approach does not involve any calibration with hydrological measurements. In addition, we have begun to question just how much small-scale heterogeneity is required to successfully predict hydrological behavior if we consider the idea of calibrating a petrophysical relationship based on hydrological data. Here we explore the possibility of estimating an optimal K/φ relationship to improve hydrological predictions. We examine this issue for the case of porosity / hydraulic conductivity characterization of a saturated heterogeneous aquifer using crosshole ground-penetrating radar (GPR), borehole porosity log data, and tracer test measurements. Based on the assumption that porosity can be linked to hydraulic conductivity, we use a grid search technique to find the K/φ relationship that best predicts the results of a hydrological tracer test at a particular location, when given a porosity field obtained from the integration of the radar and porosity log data. We then test the determined K/φ relationship at an alternate location to examine its effectiveness. Results show that this approach allows us to significantly improve hydrological predictions even at the scale of tomographic resolution.