Propagation of data and algorithmic uncertainty based on borehole calibration and perturbation – a sensitivity analysis

Pirot, Guillaume1 and Lindsay, Mark1 and Grose, Lachlan2 and de La Varga, Miguel3 and Jessell, Mark1

1Centre for Exploration Targeting, The University of Western Australia, Crawley, Australia, 2School of Earth, Atmosphere and Environment, Monash University, Clayton, Australia, 3Institute for Computational Geoscience and Reservoir Engineering, RWTH Aachen University, Aachen, Germany

Subsurface modelling is a challenge because we have a limited access to direct observations of the desired quantities of interest and because we have an imperfect understanding of geological processes. Thus, each model realization is quite uncertain. This is why we need to consider both our sources of errors and alternative modelling schemes. In order to avoid uncertainty underestimation when the purpose of modelling is decision-making, uncertainties related to observations, algorithms and conceptual representations should be propagated in the generation of stochastic geological realization ensembles.

Here, we focus on the sensitivity of data and algorithmic uncertainties on the resulting geological uncertainty. Indeed, it might not make sense to compare a pie with a cake or a mousse. This is why we leave conceptual uncertainty aside and in the hands of model selection techniques. While data errors can be estimated by repeating some measurements, algorithmic uncertainties might be more complex to define and are not always accessible. To handle that, we propose to rely on the use of pilot-stick perturbations, which consists in adding fictive drill-holes complying with the assumed stratigraphy and the presence or absence of surface geological information.  

To illustrate the method, we perform a sensitivity analysis of the perturbations on a synthetic case, based on a Precambrian basin setting, with three different geological modelling engines. The resulting geological uncertainty is analysed with different indicators based on the cardinality, entropy, connectivity, topology and geostatistics of both lithological formations and their underlying scalar-fields. Preliminary results show the pre-dominant importance of pilot-stick perturbations and their ability to mitigate the smoothing resulting from implicit modelling, in particular at locations where no surface data is available.

Acknowledgement

This work is supported by the ARC-funded Loop: Enabling Stochastic 3D Geological Modelling consortia (LP170100985) and DECRA (DE190100431) and by the Mineral Exploration Cooperative Research Centre whose activities are funded by the Australian Government’s Cooperative Research Centre Programme. This is MinEx CRC Document 2020/46.


Biography

Guillaume joined the Centre for Exploration Targeting in September 2019. He is involved in the ‘Automated 3D Modelling’ MinEx CRC project and as Work-Package 5 leader in the LOOP consortium (loop3d.org), where he develop tools to improve the characterization, the propagation and the reduction of prediction uncertainties.

Constrained Gravity Geometry Inversion with Sparse Seismic Sections

Rashidifard, Mahtab1,3, Giraud, Dr Jérémie1,3, Lindsay, A/Prof Mark1,3, Jessell, Prof Mark1,3, Ogarko, Dr Vitaliy1,2

1Centre for Exploration Targeting, University Of Western Australia, 35 Stirling Highway, WA Crawley 6009, Perth, Australia, 2International Centre for Radio Astronomy Research (ICRAR), University Of Western Australia, 35 Stirling Highway, WA Crawley 6009, Perth, Australia, 3Mineral Exploration Cooperative Research Centre, School of Earth Sciences, University of Western Australia, 35 Stirling Highway, WA Crawley 6009, Perth, Australia

Geological boundary parametrization of the subsurface that is consistent with geological and geophysical observations is the goal of 3D geological modelling. Several approaches have been
developed for geometric inversion and joint inversion of geophysical data sets. Recovering the geological boundary of units using the level-set gravity inversion method has been studied in recent
years with the focus on inversion for different numbers and shapes of the buried bodies. 

We choose to focus on constraining 3D gravity inversion with sparse 2D lower-uncertainty data from seismic images within the modelling area which accounts for the uncertainty of the sparse data to make more reliable geological prediction.

We use a level-set approach to recover the geometry of geological bodies at depth. It is not limited by the number and shape of proposed geological units and is thus appropriate for gravity inversion of the geologically complex Yamarna terrane in the Yilgarn craton, Western Australia. The study focuses on the eastern zone of the area which is mostly composed of greenstones and meta-granitic rocks. As for mineral exploration, Au targets and their near-surface components are small in dimension, thus highresolution models are required for exploration purposes. However, high-resolution data are sparsely distributed within the study area. In this study, 2D seismic sections have been used for constraining the surface evolution of rock unit boundaries during the 3D gravity geometric inversion.

The proposed work is the first we know of that implements a level-set inversion method for data sets with different spatial coverage. Our results indicate that unit boundaries from gravity inversion can be very well constrained when lower-uncertainty, lower coverage data are available. Our approach avoids bias along with the interpretation and uncertainty of the sparse data is also taken into account, showing that the final model is consistent with all available data sets.

Our results suggest that the proposed method has the potential to bring the state of the art a step further towards building a 3D geological model which is consistent with available geological and geophysical data sets. As uncertainty plays an important role in geoscientific data and modelling, geometry inversion of gravity data in this study accounts for uncertainty, meaning that sparse and ill-defined parameters are allowed to vary during the inversion within a predefined threshold limit. The proposed method has resulted in a reduction in ill-posedness of the gravity inversion problem by creating a model with consistency to all high-resolution seismic sections sparsely distributed within the model.

We acknowledge the support of the MinEx CRC and the Loop: Enabling Stochastic 3D Geological Modelling (LP170100985) consortia. The work has been supported by the Mineral Exploration Cooperative Research Centre whose activities are funded by the Australian Government’s Cooperative Research Centre Programme. This is MinEx CRC Document 2020/43.


Biography

I completed my undergraduate and masters in petroleum exploration engineering. I started my Ph.D. at UWA (CET) as a minexCRC student focusing on data fusion methodologies for integrated inversion of geological and geophysical data.  This presentation is part of my PhD project for integrating seismic and gravity with different coverage.  

Reproducible 3D model construction using map2loop

Jessell, Mark1, Ogarko, Vitaliy2, Lindsay, Mark1, Joshi, Ranee1, Piechocka, Agnieszka1,5, Grose, Lachlan3, de la Varga, Miguel4, Fitzgerald, Des6, Aillères, Laurent3, Pirot, Guillaume1

1Mineral Exploration Cooperative Research Centre, Centre for Exploration Targeting, School of Earth Sciences, UWA, Perth, Australia, 2International Centre for Radio Astronomy Research, UWA, Perth, Australia, 3School of Earth, Atmosphere and Environment, Monash University, 4Computational Geoscience and Reservoir Engineering, RWTH Aachen, Germany, 5CSIRO, Mineral Resources – Discovery, ARRC, Kensington WA, Australia

The advent of digital geological maps has not been matched by an uptake of analysing the structural data contained within. At the regional scale, the best predictor for the 3D geology of the near-subsurface is often the information contained in a geological map. This remains true even after recognising that a map is also a model, with all the attendant hidden biases ‘model’ status implies. The difficulty in reproducibly preparing input data for 3D geological models has created a demand for increased automation in the model building process. The information stored in a map falls into three categories of geometric data: positional data such as the position of faults, intrusive and stratigraphic contacts; gradient data, such as the dips of contacts or faults and topological data, such as the age relationships of faults and stratigraphic units.

We present two Python libraries (map2loop and map2model) which combine all these observations with conceptual information, including assumptions regarding the subsurface extent of faults and plutons to provide sufficient constraints to build a reasonable 3D geological model. These algorithms allow automatic deconstruction of a geological map to recover the necessary positional, topological and gradient data as inputs to different 3D geological modelling codes. This automation provides significant advantages: it significantly reduces the time to first prototype models; it produces reproducible models for the same source data, it clearly separates the primary data from subsets produced from filtering via data reduction and conceptual constraints; and provides a homogenous pathway to sensitivity analysis, uncertainty quantification and Value of Information studies. We use examples of the folded and faulted terrains across Australia to demonstrate a complete workflow from data extraction to 3D modelling using three different 3D modelling engines: GemPy, LoopStructural and 3D GeoModeller. 

We acknowledge the support of the MinEx CRC and the Loop: Enabling Stochastic 3D Geological Modelling (LP170100985) consortia. The work has been supported by the Mineral Exploration Cooperative Research Centre whose activities are funded by the Australian Government’s Cooperative Research Centre Programme. This is MinEx CRC Document 2020/48.


Biography

Mark Jessell is a Professor at the Centre for Exploration Targeting at The University of Western Australia. His current scientific interests revolve around integration of geology and geophysics in 3D (the Loop project), and the tectonics and metallogenesis of the West African and Guyanese Cratons (WAXI & SAXI).

LoopResources – Reducing the Mining Footprint

Ailleres, Laurent1, Grose, Lachlan1, Caumon, Guillaume2, Jessell, Mark3

1School of Earth, Atmosphere and Environment, Monash University, Melbourne, Australia, 2RING, Université de Lorraine, Nancy, France, 3CET, University of Western Australia, Australia

Loop is a OneGeology initiative, initiated by Geoscience Australia and funded by Australian Territory, State and Federal Geological Surveys, the ARC and the MinEx CRC with the participation of BHP, Anglo American, GSWA and Micromine. The project is led by Monash University and involves research groups from the University of Western Australia, the RING consortium at the Universite de Lorraine, Nancy, France and RWTH Aachen in Germany. In-kind research is also provided by Natural Resources Canada (Geological Survey of Canada), Geoscience Australia, the British Geological Survey and the BRGM.

We have implemented the use of all structural geological data (e.g.: fault kinematics, fold axial surfaces, fold axes, deformational overprinting relationship) in the modelling process.  We have automated the building of 3D geological models from geological survey served geological data including automatic geological map topological analysis and geological history building. As a proof of concept, users can now draw a polygon on a map and generate 3D models in just a few minutes using the map2loop and LoopStructural libraries. (github.com/Loop3D/LoopStructural & github.com/Loop3D/map2loop-2)

We are integrating geophysical constraints and modelling as early as possible in the modelling workflow.  Model uncertainty is characterised and an integral part of the modelling process. We are in the process of generalising the use of Bayesian modelling to 3D structural geological modelling.

The main outcome of the development of the structural modelling method (LoopStructural) is the definition of structural frames. A structural frame is interpolated throughout the entire modelled volume and is associated to each structural event or object and their corresponding finite strain axes. Each fault has a structural frame (Fig.1) based on the geometry of the fault (faults are not necessarily planar) and the damage zone defined by an ellipsoid representing the decay of the offset in all directions (slowest decay along the throw direction). A fold event has a structural frame defined by a direction normal to the axial surface foliation, a direction parallel to the fold axis and the last axis is parallel to the extension direction. Combining these different events, in a time-aware manner, is the essence of LoopStructural.

LoopResources will be the property estimation library for the Loop platform.  During modelling, the time-aware application of structural frames results in the implicit definition of a curvilinear rectangular coordinate system everywhere in the model and conformable to geological layering.  Using this deformed cartesian coordinate system, we propose to adapt geostatistical and interpolation methods (e.g. kriging) to curvilinear coordinate systems. This will ensure that lithological anisotropies are enforced during resource estimation and property modelling. 

The proposed outcomes improve significantly on current capabilities and will provide a machine-supported decision system for 1) improved domaining, 2) optimal definition of ore blocks of consistent ore grades, geotechnical properties, and crushing requirements, 3) optimal extraction, crushing and processing costs, and 4) increased recovery rates.

In essence, building a much better digital-twin representing the geology, distribution of resources, the geometallurgical and geotechnical characteristics of a mine will ensure better mining of deposits and ultimately better mine footprint.


Biography

Laurent is a structural geologist interested in the evolution of tectonic processes through time and their effect on multi-scale mineralisation processes. He specialises in structural geology and geophysics as well as multi-scale 3D geological modelling applied to minerals systems.  He leads the Loop initiative champions structurally-ruled probabilistic 3D geological modelling.

LoopStructural 1.0: Time aware geological modelling

Grose, Lachlan1, Ailleres, Laurent1, Laurent, Gautier2, Jessell, Mark3

1School of Earth Atmosphere and Environment, Monash University, Melbourne 3800, Australia, 2 Université d’Orléans, CNRS, BRGM, ISTO, UMR 7327, Orleans France, 3 Mineral Exploration Cooperative Research Centre, School of Earth Sciences, UWA, Perth, Australia

LoopStructural is a new open source 3D geological modelling python package (www.github.com/Loop3d/LoopStructural). Geological features are encoded into the geological model using a time aware approach where the relative timing of different deformation features is used to help construct complicated geometries. We use structural frames which are curvilinear coordinate systems based around the major structural feature (e.g. fold axial surfaces, fault surfaces), structural direction (e.g. fold axis, fault slip direction) and where necessary an intermediate direction (e.g. fault extent). This allows for folds and faults to be integrated into the description of the geological features in the implicit models. In this contribution we will use map2loop to automatically extract and augment input data from open access geological datasets from Geological Survey of Western Australia from the Hamersley Basin. The model area will include consists of upright refolded folds of Archean and Proterozoic stratigraphy overlying an Archean basement. The folded stratigraphy is overprinted by NW-SE trending faults. In the model the fault network is modelled first using observations of the fault trace using the estimated displacements from the geological map. The folded stratigraphy is then modelled by building a fold frame that characterises the geometry of the axial surface and the fold axis direction. The fold geometry is modelled by fitting curves representing the fold axis geometry in the axial surface and the fold geometry looking down plunge. We show that by using the fold constraints the geometry of the modelled folds are consistent with the geometries drawn in cross sections.


Biography

Lachlan is a research fellow at Monash University working on the Loop project. His research interests are encoding structural geology in 3D geological modelling algorithms.

About the GSA

The Geological Society of Australia was established as a non-profit organisation in 1952 to promote, advance and support Earth sciences in Australia.

As a broadly based professional society that aims to represent all Earth Science disciplines, the GSA attracts a wide diversity of members working in a similarly broad range of industries.

Conference Managers

Please contact the team at Conference Design with any questions regarding the conference.
© 2020 Conference Design Pty Ltd