Groundwater is a critical resource for supporting human consumption, stock water, agricultural use, and mineral or energy extraction as well as the environment. However, the quality of groundwater varies enormously from potable to hyper-saline, particularly in the Australian context. To evaluate the suitability of a groundwater resource, the spatial distribution of salinity within an aquifer is typically estimated by measuring the electrical conductivity (EC) of groundwater samples from within boreholes. However, drilling is a logistically and economically challenging task, and hydrogeologists are usually left with a sparse set of measurements from which to infer groundwater salinity over large spatial extents.
Airborne electromagnetic (AEM) surveying is a geophysical technique for estimating the bulk electrical conductivity of the near-surface. Where AEM bulk conductivity are well correlated with groundwater salinity in aquifers, AEM is a useful tool for modelling salinity in the data sparse areas between the boreholes. We present here a probabilistic method for modelling salinity and a case study from the Keep River Plains in the Northern Territory. Co-located probabilistic AEM inversions and EC measurements on pore fluids at coincident locations were fused to calculate an empirical joint probability density function. This function allowed us to estimate salinity away from the bores by sampling the ensemble of AEM conductivities. Unlike deterministic methods that provide a single estimate of salinity, our method generates an ensemble of estimates, which can be used to quantify predictive uncertainty. The results provided by our method can feed into decision making while accounting for uncertainty, enabling remote communities to manage their land and water resources more responsibly.
