The objective of this research was to explore a cost-effective and non-invasive methodology to characterize the spatial variability of hydraulic conductivity using airborne electromagnetic (AEM) signatures as an alternative to traditional techniques such as borehole sampling. The relationship between AEM measured apparent resistivity and magnetic field strength was explored using a small data set that included 180 natural moisture (NM) content data points and a total data set of 546 grain size distributions that excluded the NM data. The grain size distributions were used to develop a soil indicator (SI) parameter and to estimate the hydraulic conductivity (K*) using pedotransfer functions. Predictive models were developed using three techniques: artificial neural network regression (ANNR), support vector regression (SVR), and artificial neural network classification (ANNC). The sole use of non-invasive parameters to characterize K* proved insufficient. The inclusion of supplemental invasively collected parameters showed ANNR to best characterize the relationship (R2 = 0.64) with the smaller data set, while the SVR model performed best with the total data set (R2 = 0.57). ANNC was shown to be a viable alternative (overall accuracy = 88 percent) when a broad characterization of K* was sufficient. This study lays out a methodology that could be used for future K* characterization using improved data sets.

This content is PDF only. Please click on the PDF icon to access.
You do not have access to this content, please speak to your institutional administrator if you feel you should have access.