Approximately 80 % of the Australian landmass is blanketed post-mineralisation sedimentary cover and regolith. The cover presents a challenge to successful exploration. The thickness, nature and variability of these cover materials are not well known, particularly in less explored 'green field' regions across Australia.
To facilitate more effective exploration in these buried landscapes Geoscience Australia has developed databases, methodologies and predictive models to better understand the nature of the cover. We aimed to generate 3D surfaces of the depths to major chronostratigraphic interfaces, including the bases of Cenozoic, Mesozoic, Palaeozoic and Neoproterozoic.
Drilling provides the most reliable source of depth information. Geophysical depth estimates derived from, for example, depth to magnetic basement techniques, seismic reflection and airborne electromagnetic profiles also provide important constraints on cover depth. These point depth measurements are stored in a new national database, the Estimates of Geological and Geophysical Surfaces or EGGS (Mathews et al., 2020). Depth information centrally stored in EGGS were used to underpin the generation of chronostratigraphic depth surfaces using point interpolation methods that establishes predictive relationships between the depth estimates and other geological or geophysical information (e.g. distance from outcrop, gravity).
- Information from hundreds of boreholes from stratigraphic, petroleum and exploration company drill holes was interpreted within the focus area covered by the airborne electromagnetic AusAEM1 survey and surrounding regions.
- Stratigraphic horizons were interpreted from pre-competitive datasets acquired during the Exploring for the Future program, such as the airborne electromagnetic AusAEM1 survey (Lee-Cooper and Brodie, 2020; Wong et al., 2020) and the South Nicholson reflection seismic survey (Carr et al., 2020).
- Thousands of depth to magnetic top estimates were modelled and attributed with the inferred stratigraphic interface they reflect, based on the solid geology dataset produced during the Exploring for the Future program.
Collaboration
The machine learning code used to generate cover depth predictions was developed in collaboration with Data61 at CSIRO.
The key external collaborators of this project activity included the Northern Territory Geological Survey (NTGS) and the Geological Survey of Queensland (GSQ). This specifically involved sharing drill hole information and stored exploration company reports.
Outputs
- Chronostratigraphic unit thicknesses from boreholes, magnetic datasets and airborne electromagnetic data are accessible through the EGGS database on the Exploring for the Future Data Discovery Portal.
- Chronostratigraphic cover thickness models over the airborne electromagnetic AusAEM1 survey area are available on the Exploring for the Future Data Discovery Portal. Two different techniques, e.g. minimum curvature interpolation method and machine learning, were used to generate the cover thickness models (Bonnardot et al., 2020). These cover predictions also show the uncertainties of the predictions.
- The machine learning UncoverML codes used to generate cover-thickness models are available through the Geoscience Australia GitHub repository (Wilford et al., 2020).
