Time-propagation modeling of the influence of porosity, mineralogy, and water saturation indicates that water saturation exerts a first-order influence on bulk dielectric constant and that mineralogy and porosity exert a second-order influence, though the relationship between these variables is highly interactive. Thus, water saturation is a variable that must be known to some degree for accurate GPR interpretation. Given that capillary properties, and thus water saturations, are likely to vary with lithofacies, modeling in this study indicates that higher water saturations may enhance GPR response. Conversely, overly dry subsurface conditions may eliminate any response. One could argue that some sites should be watered before performing a GPR survey or that optimal survey conditions may follow rainy periods. In addition, TP modeling for conditions at a site may provide information important to the design of the GPR survey.
Modeling performed in the study demonstrates that GPR response varies significantly with typical variations in rock and soil saturation, porosity, and mineralogy. Because of this, interpretation and modeling of GPR data should incorporate consideration of these variables. Conversely, modeling has shown that GPR response can be sensitive to subtle differences in rock properties and therefore can be a powerful tool for investigating subsurface properties. With inverse modeling--that is, using measured field GPR data and TP modeling--it may be possible to back-calculate the rock properties required to produce the observed GPR response. Further, if good correlations exist between porosity or dielectric constant and hydraulic permeability, as found for some rocks in this study, subsurface permeability may be predicted. Given the theoretical sensitivity shown here of GPR response to the fundamental rock properties of porosity, water saturation, mineralogy, and hydraulic permeability, it is likely that refinement and further calibration of this type of modeling will improve and enhance the information obtained from GPR and potentially broaden its application. Beyond interpreting existing data, TP-modeled dielectric-constant values may also be used to populate geologic models in waveform modeling and allow better design of GPR surveys by predicting response under specific conditions.
Kansas Geological Survey
Web version December 3, 2001
http://www.kgs.ku.edu/Current/2001/martinez/martinez11.html
email:lbrosius@kgs.ku.edu