b'AEGC 2023Short abstractsthe permeability from the acquired well logs, which resorts toThese results highlight that variations from fluvial- to wave-the application of time-consuming and expensive laboratorydominated systems represent first-order controls on the techniques using core samples. We present here a methodologycoeval development of deep-water systems. However, while to highlight how machine-learning based predictions ofshelf-to-basin sediment transfer mechanisms influence slope permeability from basic logs can be advantageous to traditionalgradients, inherited palaeotopography and tectonics have workflows and porosity-permeability relationships. From thea direct impact on shelf-margin architecture and associated available core data of two Norwegian offshore wells, relationshipsturbidite systems. By integrating advanced tools in seismic are built between porosity and permeability for the differentinterpretation, quantitative 3D seismic stratigraphy represents lithofacies across the reservoir. An automated outlier detectiona novel approach in assessing at high resolution the controls on algorithm is run to remove erroneous data points. The totaldeep-water sand delivery, and potentially predicting the type porosity log is interpreted from density and neutron combinationand location of reservoirs in deep-water areas based on shelf-which is calibrated with core porosity. Then porosity log ismargin architecture and depositional process regime.used to derive a continuous permeability profile. 70% of basic petrophysical curves like gamma ray, resistivity, density, and sonic logs from these two wells are trained with several regression- Application of natural field airborne electromagnetics based machine-learning algorithms such as support vectorin conductive geo-environments.machine (SVM), random forests (RF), and artificial neural network (ANN) for permeability prediction. The performance of theseAlexander Prikhodko, Aamna Sirohey, Andrew Carpenter and models is validated with the remaining 30% of data. A wide rangeAndrei Bagrianskiof hyper-parameters is tested to ensure that not only the bestExpert Geophysics Limitedmodel is selected based on the mean squared error value and R2 coefficient, but also the right combination of model parametersThe presence of conductive overburden and conductive host is selected. The averaging ensemble method random forestsrocks impose significant limitations on the application of airborne resembles the best solution when tested on a blind well whereelectromagnetic (EM) methods which make use of controlled limited core data exists. The workflow highlights how a machine- source transmitters (active source systems) for subsurface learning based approach helps to provide quick and consistentexploration. The limitations in applicability arise mainly due to results without disregarding the inherent heterogeneity observedthe limited depth of penetration as a consequence of reduced across the reservoirs. skin depth in conductive materials coupled with limitations in the energy transmitted by systems with controlled field sources. Airborne EM methods exploiting natural fields (passive field Shelf-to-basin sediment transfer mechanisms insystems) overcome these limitations affecting active source contrasted tectonic settings: Insights from quantitativesystems to a great degree by utilising the naturally occurring EM 3D seismic stratigraphy. plane waves resulting from thunderstorms and activity in the ionosphere as the primary field source. The advantage of this is Victorien Paumard, Simon Lang, Jennifer Chandra, Anthonythat these naturally occurring fields occur over a broad frequency Gartrell, Andy Jian and John Shepherd range and energise the geological environment over a broad depth range, which in all cases exceeds the depth of investigation The University of Western Australiaof existing active source systems. Current airborne natural field With ~15% of siliciclastic hydrocarbon reservoirs located inEM technologies have a lower limit of approximately 25 Hz deep-water basins, a key challenge is to predict when andcorresponding to the lowest frequency (or greater depth) capable where sands are bypassed to deep-water areas, and howof detection by these systems. The lowest frequency of airborne they are architecturally organised. Quantitative 3D seismicpassive source systems provides a depth of investigation that stratigraphy (QSS) aims at investigating the linkages betweenexceeds the capabilities of active source systems. Based on the hydrodynamic regime along palaeoshorelines, shelf-marginresults of theoretical modeling, including examples that consider architecture and the development of coeval deep-waterrelatively conductive targets located in conductive host rock as systems in a variety of tectonic and climatic settings. Thiswell as beneath conductive overburden, and case studies based approach is underpinned by state-of-the-art, full-volume 3Don field data, the capabilities of the MobileMT technology, based seismic interpretation methods that enable high-resolutionon the natural field EM principle, are demonstrated.seismic stratigraphic analysis. Results obtained from shelf margins developed in late syn-rift (Northern Carnarvon Basin,From pointwise data toward a continuous description Australia), post-rift (Porcupine Basin, Ireland) and foreland fold and thrust belts (NW Borneo) settings will be presented.of 3D stress tensor: an example from the Northern Statistical analysis of these data reveals that overall, fluvial- Bowen Basin.dominated shorelines are typically associated with: (1) slope gradients twice steeper than the wave-dominated shorelines;Mojtaba Rajabi 1, Moritz Zeigler2, Renate Sliwa3 and Joan and (2) longer run-out turbidite systems. However, theseEasterle1relationships become less clear in basins where the slopes are1 University of Queenslandhighly affected by tectonics (i.e., ponded slopes) and/or the2 Helmholtz Centre Potsdam, GFZ German Research Centre for inherited palaeotopography of the basin. Additionally, theGeosciences, Potsdam, Germany turbidite systems developed in NW Borneo present a significant3 Integrated Geoscience Pty Ltd longer run-out distance because the shelf-margin clinoforms are merging with the continental margin clinoforms, thusKnowledge of in-situ stress is important in various aspects promoting the development of architecturally more matureof geoscience and engineering. In-situ stress measurements turbidite systems along higher and longer slopes. provide pointwise data in an area of interest. However, in FEBRUARY 2023 PREVIEW 132'