Hydraulic Tomography Analyses of Different Datasets for Subsurface Heterogeneity Characterization at Various Scales

Abstract

Hydraulic Tomography (HT) has been evaluated to be a robust approach for high-resolution characterization of subsurface heterogeneity. However, geostatistics-based HT may produce overly smooth distributions of hydraulic parameters such as hydraulic conductivity (K) and specific storage (Ss) when hydraulic head data used to constrain the inversion is limited. Furthermore, only a few HT studies have been performed for large-scale field problems due to the difficulty in conducting dedicated HT surveys at large-scale sites, as well as due to uncertainty regarding model conceptualization. This thesis documents four studies designed to investigate the performance of HT in characterizing the spatial distributions of K and Ss at various scales through the inclusion of different types of data for inverse modeling. Study 1 investigates the effect of prior geological information on K and Ss heterogeneity characterization through Transient Hydraulic Tomography (THT) analysis of laboratory sandbox data. Study 2 explores the feasibility of THT analysis of long-term municipal well records for large-scale heterogeneity characterization through synthetic experiments, while Study 3 extends the synthetic study to a field application utilizing data from a wellfield in Kitchener, Ontario, Canada by addressing uncertain initial and boundary conditions for inverse modeling. Finally, Study 4 evaluates the usefulness of field cross-hole flowmeter measurements in mapping spatial K distribution through Steady-State Hydraulic Tomography (SSHT) analysis at a highly heterogeneous field site located on the University of Waterloo. Results from these studies mainly reveal that: (1) the incorporation of prior geological information into geostatistical inverse models improves characterization significantly when pumping tests and drawdown measurements are sparse; however, attention must be paid when constructing geological models for reliable structure information, (2) existing municipal wellfield records could be utilized for large-scale heterogeneity characterization using the approach of HT when uncertainties regarding initial and boundary conditions are well addressed for inverse modeling, and (3) the integration of cross-hole flowmeter measurements with hydraulic head data improves characterization results in terms of revealing K heterogeneity details and predicting independent hydraulic test data. Overall, the body of work presented in this thesis advocates the inclusion of additional datasets that carry non-redundant heterogeneity information for geostatistical inverse modeling and demonstrates the feasibility of utilizing alternative datasets in HT for subsurface heterogeneity characterization at large-scale sites.