b'Education matters subsurface targets as one homogenousUniversity of Adelaide, Southof groundwater due to its generation of body. The high temperature SQUIDAustraliaMPhil thesis a measurable electrical signal. Previously was able to accurately separate theregarded as a predominately qualitative two targets. Comparison of decayJoe Rugari, University of Adelaide:geophysical tool, contributing factors analysis between survey 1 and 2 sawElectrokinetic methods and applications inincluding advancements in low-cost an order of magnitude improvementAustralian aquifer settings: High-dimensioninstrumentation and processing in the amplitude of magnetic fieldelectrical tomography imaging and neuralcapabilities have meant self-potential variations recoverable by the fluxgatenetwork filtration techniques. surveys can now provide spatially magnetometer. The same comparisonBeing the driest continent in thesignificant quantitative data for a range saw two orders of magnitudeworld, there is a significant relianceof groundwater modelling inputs such as improvement in the variations detectableon groundwater resources withinpermeability.by the high temperature SQUID. Surveymany communities and industries 1 represented poor survey design whilstThe method has been recurrently survey 2 represented optimal surveythroughout Australia. Particularly inreviewed since its early conception in design. Results of the comparisonregional areas with low rainfall andinternational geophysical literature highlight the influence external factorssurface runoff resources, the underlyingthrough to modern times. However, have on receivers ability to accuratelygroundwater availability plays aonly a small quantity of this peer record the secondary electromagneticpivotal role in population capacityreviewed research has been conducted field decay. and economic prosperity. Whilst thewithin Australia. A lesser extent of importance of groundwater resourcespublished literature therefore deals is indisputable, many aspects of its realin particularly with addressing the University of Adelaide, Southworld homeostatic processes, in bothchallenges of both our harsh climate, macro and micro scales, remain difficultand surface and geological conditions. AustraliaPhD thesis to decipher and explain. With our own unique geological and Dennis Conway, University of Adelaide:Within Australias fractured rock aquiferhydrogeological settings, current Advances in Magnetotelluric Modelling:systems, attributed with storage of theand future challenges regarding Time-Lapse Inversion, Bayesian Inversionlargest volume of groundwater resourcessecurement of groundwater resources, and Machine Learning. nationally, there is still only fragmentedand increasingly common practice understandings of several of theirof industrial geotechnical processes This thesis presents advancementsprincipal components and capacities.such as fracking, all research and to the area of magnetotelluric (MT)This is inclusive even of key aquiferfindings are vital contributions to modelling. There are three main aims tocharacteristics, such as total volumefurthering our understanding of this work. The first aim is to implementestimations, regeneration sources, andpotential groundwater applications for an inversion to model time-lapse MT datatheir flow or transportation methods.self-potential methods on home soil. in a temporal dimension. The algorithmImproved modelling capabilities andThis research thesis provides analyses considers the entire dataset at once,techniques based on prominent andof multiple electro-kinetic field with penalisations for model roughnessrobust hydrogeological principals areresearch projects. New self-potential in both the spatial and temporalcontinually emerging from advancingdatasets have been collected in the dimensions. The inversion is tested ontechnologies, new data sources andAdelaide Hills targeting stimulated synthetic data, as well as a case-studyforward thinking. However, within thefractured-rock aquifers up to 40 m from a coal-seam gas dewatering survey.field data retrieval facet of hydrologicalbelow surface - a considerably deep Second is to explore the problem ofresearch a seemingly slower evolution istarget for the method, particularly non-uniqueness in MT data inversiontaking place. A vast quantity of aquiferwithin highly conductive Australian by implementing a 1D Bayesianinformation is still derived directly fromgeological conditions. Previously inversion using an efficient sampler.intrusive observation wells. Although thecollected geophysical datasets from the The implemented model includes aplethora of information these wells canAdelaide Hills have been reprocessed novel way of regularising MT inversionyield in modelling is invaluable, there arefrom two to four-dimensions utilising by allowing the strength of smoothingsome profound limitations that must stillnewly constructed algorithms, to vary between different models. Thebe addressed. then reanalysed with supporting Bayesian inversion is tested on syntheticgeophysical datasets. And finally, and case-study datasets with resultsWells are costly to establish due toa long term (46 day) self-potential matching known data. The third aim is todrilling expenses, can only providemonitoring programme was conducted implement a proxy function for the 3D MTsingle point information, and can alsoat a commercial-use porous media forward function based on artificial neuralbe disruptive to the homeostasis ofaquifer to investigate novel techniques networks. This allows for rapid evaluationthe system. The self-potential methodin both autonomous groundwater of the forward function and the use ofis an electro-kinetic geophysicalflow presence investigation, and evolutionary algorithms to invert formethod that has recently been re- environmental noise filtering resistivity structures. The evolutionaryidentified as an immensely promisingmethodologies for a given self-search algorithm is tested on syntheticgroundwater technique. It is a fast,potential dataset.data sets and a case-study data set frompassive, inexpensive surface technique the Curnamona Province, South Australia.which requires no drilling. UniquelyThis research endeavours to draw Together, these three novel algorithmsand most importantly however, it is thefurther conclusion on the self-potential and software implementations representonly geophysical method that is directlymethods prospective as a value-a contribution to the toolkit of MTsensitive to not only the presence ofadding and commercial viability modelling. groundwater, but also the physical flowmodern geophysical technique in DECEMBER 2019 PREVIEW 40'