Incorporating geological zonal information by cluster analysis into hydraulic tomography in sandbox and field studies

Abstract

Hydraulic tomography (HT) has been proven to be an effective technique for accurately mapping aquifer heterogeneity over the past few decades. Aquifer response data used for model calibration in HT inverse analysis are frequently sparse, so geological structures reflected by the estimated hydraulic conductivity (K) tomograms may be smoothed out and are discrepant from the actual structures. The locations and geometries of high-K and low-K zones, which impact groundwater flow and contaminant transport processes, need to be corrected. Although collecting geological or geophysical data could improve HT results, exploring an alternative approach without collecting additional data may be possible. Therefore, this study explores the possibility of using hierarchical cluster analysis to extract geological zonal information from HT K estimates to characterize the geological structures. For this purpose, this study treats the zonal information as prior information of HT inverse models. We first conducted laboratory sandbox experiments to test the effectiveness of the proposed approach. Afterward, this approach is applied at a highly heterogeneous site in a dam foundation. Results show that K tomograms obtained by the traditional HT inverse analysis can reflect both the interlayer and intralayer heterogeneity, but the geometries of the interlayer heterogeneity must be improved. Integrating the zonal information into the HT inverse model improves the reliability of K estimates to predict drawdowns of different pumping tests. It corrects the geometries of geological structures without additional data.