Electrokinetic Methods and Applications in Australian Aquifer Settings: High-Dimension Electrical Tomography Imaging and Neural Network Filtration Techniques

Abstract

Being the driest continent in the world, there is a significant reliance on groundwater resources within many communities and industries throughout Australia. Particularly in regional areas with low rainfall and surface runoff resources, the underlying groundwater availability plays a pivotal role in population capacity and economic prosperity. Whilst the importance of groundwater resources is indisputable, many aspects of its real world homeostatic processes, in both macro and micro scales, remain difficult to decipher and explain. Within Australia’s fractured rock aquifer systems, attributed with storage of the largest volume of groundwater resources nationally, there is still only fragmented understandings of several of their principal components and capacities. This is inclusive even of key aquifer characteristics, such as total volume estimations, regeneration sources, and their flow or transportation methods. Improved modeling capabilities and techniques based on prominent and robust hydrogeological principals are continually emerging from advancing technologies, new data sources and forward thinking. However, within the field data retrieval facet of hydrological research a seemingly slower evolution is taking place. A vast quantity of aquifer information is still derived directly from intrusive observation wells. Although the plethora of information these wells can yield in modelling is invaluable, there are some profound limitations that must still be addressed. Wells are costly to establish due to drilling expenses, can only provide single point information, and can also be disruptive to the homeostasis of the system. The self-potential method is an electro-kinetic geophysical method that has recently been re-identified as an immensely promising groundwater technique. It is a fast, passive, inexpensive surface technique which requires no drilling. Uniquely and most importantly however, it is the only geophysical method that is directly sensitive to not only the presence of groundwater, but also the physical flow of groundwater due to its generation of a measurable electrical signal. Previously regarded as a predominately qualitative geophysical tool, contributing factors including advancements in low-cost instrumentation and processing capabilities have meant self-potential surveys can now provide spatially significant quantitative data for a range of groundwater modelling inputs such as permeability. The method has been recurrently reviewed since its early conception in international geophysical literature through to modern times. However, only a small quantity of this peer reviewed research has been conducted within Australia. A lesser extent of published literature therefore deals in particularly with addressing the challenges of both our harsh climate, and surface and geological conditions. With our own unique geological and hydrogeological settings, current and future challenges regarding securement of groundwater resources, and increasingly common practice of industrial geotechnical processes such as fracking, all research and findings are vital contributions to furthering our understanding of potential groundwater applications for self-potential methods on home soil. This research thesis provides analyses of multiple electro-kinetic field research projects. New self-potential datasets have been collected in the Adelaide Hills targeting stimulated fractured-rock aquifers up to 40m below surface - a considerably deep target for the method, particularly within highly conductive Australian geological conditions. Previously collected geophysical datasets from the Adelaide Hills have been reprocessed from two to four-dimensions utulising newly constructed algorithms, then reanalysed with supporting geophysical datasets. And finally, a long term (46 day) self-potential monitoring program was conducted at a commercial-use porous media aquifer to investigate novel techniques in both autonomous groundwater flow presence investigation, and environmental noise filtering methodologies for a given self-potential dataset. This research endeavors to draw further conclusion on the self-potential methods prospective as a value-adding and commercial viability modern geophysical technique in Australian groundwater research. Additionally, employing use of artificial neural networks (machine learning) for the self-potential autonomous detection and environmental noise filtration methods, we highlight the current gap in geophysical literature regarding the combination of these techniques. A light is drawn to the combined techniques immensely promising future of potential applications and contributions within the wider electrical geophysics data automation and filtration space. Much akin to our continual pursuit for mineralisation deposits, Australia is searching deeper than ever before for crucial groundwater supplies as shallower sedimentary aquifers are becoming fully utilised or depleted. As we move forward towards this new era of deepening natural resources, we must further develop both old and new tools which can enhance clarity of understanding within these challenging hydrogeological systems.