Estimating groundwater storage volumes is part of assessing the potential capacity and recovery efficiencies of managed aquifer recharge (MAR) options. However, such storage estimates are challenging, particularly in areas of poor data density- as found in most of Australia. This study reports on recent advances made in storage estimation methods and sensitivity analysis to quantify uncertainties.
Workflows developed by Geoscience Australia for the estimation of groundwater volumes for various salinity categories involve: (1) using salinity measurements of sonic-core pore fluid extracts to define a simple relationship between airborne electromagnetic (AEM) conductivity and groundwater salinity; (2) determining bulk aquifer volumes using hydrostratigraphic textural classes, and potentiometry to map the distribution of saturated coarser-textured aquifer zones in AEM depth slices; and (3) estimating the likely range of effective porosity for texture classes based on a combination of borehole nuclear magnetic resonance (NMR) data and surface nuclear magnetic resonance (SNMR) data.
Sensitivity analysis has been undertaken for both 1D and 3D AEM inversion methods. Variables assessed include: (1) the AEM inversion method; (2) the regularisation used in 1D AEM inversion; (3) the AEM conductivity thresholds for categorising groundwater salinity; and (4) limitations placed by hydrostratigraphy, texture and saturation.
This analysis shows that volume estimates of good quality (<600 mg/L TDS) and acceptable quality (600-1200 mg/L TDS) groundwater are particularly sensitive to the AEM inversion method and regularisation constraints used. For example, tightening the regularisation constraints on 1D AEM inversions resulted in order-of-magnitude fresh groundwater volume reductions. Similar magnitude reductions also occurred for saline (>3000 mg/L TDS) groundwater volumes when textural constraints were included. The impact on volumes from changing the effective porosity ranges or the AEM thresholds used to map groundwater salinity were comparatively low.
This study has highlighted the significant uncertainties surrounding estimation of fresh groundwater volumes using AEM data, particularly in heterogeneous aquifers with variable groundwater salinity distribution. While AEM technologies continue to improve, they resolve conductive targets better than resistive (fresh water) domains. The study has also revealed that careful consideration needs to be given to AEM data processing and the use of spatial filters, while the use of 3D inversion methods provides more accurate estimation of overall conductivity distribution except in near-surface (<15m) geology. Other improvements being explored include the use of stochastic approaches that incorporate a range of sedimentary and petrophysical models to better forecast the range of possible aquifer properties.