The tectonic setting of porphyry systems is influenced by the subduction style and history that impact the distribution and concentration of copper (Cu), gold (Au), and molybdenum mineralisation. Typically linked to the intrusion of arc-related magma into the upper crust along subduction zones, the formation of porphyry ore deposits is currently understood primarily through geological and geophysical observations of the overriding plate, creating a knowledge gap regarding arc metallogenic processes in convergence zones over time. In this study, we address this gap by investigating the connection between the formation of porphyry Cu-Au deposits and the evolution of subduction zones, utilising a range of features derived from a plate motion model and oceanic crust age grids. Incorporating 47 Cenozoic intrusion-related Cu-Au deposits located in Papua New Guinea and the Solomon Islands, we employ a spatiotemporal mineral prospectivity framework that leverages advanced machine learning methods to map prospective arc terranes. The model successfully predicts all known mineral occurrences in the testing set and identifies the most important features for predicting potential areas of porphyry mineralisation.
We observe that the obliquity angle of the relative motion vector in subduction zones plays a crucial role in distinguishing between mineralised (highly prospective) and barren areas (low prospective). This feature is recognised for its significant influence on a spectrum of geological processes, encompassing fluid flow dynamics, magmatic processes, and stress regimes. This influence extends to the transport of mineralising elements and the creation of favourable conditions for ore deposition, with the range of 25 to 90 degrees correlating with mineralised zones, suggesting that oblique subduction zones are more likely to be rich in mineralisation in the study area. Additionally, the length and curvature of arcs emerge as important features for identifying mineralised areas, with tightly curved arcs associated with higher compressional stress and fractures facilitating magma ascent and porphyry formation. The orthogonal component of the downgoing absolute plate velocity is also identified as a significant feature, with higher magnitudes associated with mineralisation, indicating that rapid convergence rates are optimal for porphyry system formation due to accelerated metasomatism and partial melting processes in the overriding plate.
The seafloor spreading rate of the subducting crust, computed at the time when the crust originally formed, is an additional important feature linked to mineralised areas. This preferentially occurs when crust formed in the range of 25 to 55 mm/yr (half spreading rate) is subducted. At lower spreading rates, there is a higher proportion of serpentinised mantle peridotite, adding water and carbon to the plate, which will be expelled during subduction, contributing to increasing hydrous melting in the mantle wedge and acting as a catalyst for porphyry deposits. In conclusion, the performance of our model underscores the potential of integrating plate motion models and machine learning to advance mineral exploration along subduction zones. This approach holds promise for more efficient, accurate, and sustainable exploration strategies in these geologically active areas.

How to cite: Farahbakhsh, E., Zahirovic, S., McInnes, B. I. A., Polanco, S., Kohlmann, F., Seton, M., and Muller, R. D.: Unveiling the temporal dynamics: A spatiotemporal prospectivity model for porphyry systems in Papua New Guinea and the Solomon Islands, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15856, https://doi.org/10.5194/egusphere-egu24-15856, 2024.

The Goongarrie South, situated northwest of Kalgoorlie in Western Australia, hosts a global resource (60Mt) of lateritic Ni-Co deposits at 1% Ni and 0.07% Co. The Goongarrie lateritic Ni-Co deposit extends over a strike length of 7.5 km and averages approximately 800m in width and 40m in thickness. These deposits originated from the weathering of serpentinized dunite. The lateritic profile is subdivided into lower saprolite, dominated by carbonate, talc, serpentine, chlorite, and mica, and upper ferruginous saprolite, representing the economic Ni-Co laterite, and dominated by goethite and hematite. The ultramafic index of alteration (100 x [(Al₂O₃+Fe₂O₃(_T)/(SiO₂+MgO+Al₂O₃+Fe₂O₃(_T)]) for the Ni-Co laterite is >60, contrasting with <60 in the lower saprolite and saprock. The laterite profile thickens up to 120 m over shear zones. In this study, we investigated the geochemical behavior of gold and some critical metals (REE, PGE, Sc, Ni, Cr, Mn, Co) in the Goongarrie South lateritic nickel deposits, using bulk and in-situ mineralogical, and geochemical methods. Our results show that gold and critical metals are concentrated in different horizons of the lateritic profile. Gold underwent leaching by acidic, halogen-rich, hypersaline groundwater, and has been enriched near the contact between the Ni-Co laterite and lower saprolite and saprock over shear zones, where the pH gradient increased. Additionally, gold is concentrated in paleoredox fronts within the Ni-Co laterite. The Au mineralization is Ag-poor, and occurs as cavity-filling, microcrystalline grains, and aggregates, indicating its supergene origin. Laser ablation ICP-MS mapping indicates that Ni was released from olivine in the lower saprolite and reprecipitated in nimite (Ni chlorite), while Sc is concentrated in serpentine and mica in the lower saprolite. In the Ni-Co laterite, Cr, Ti, V, Sc, Sb, and Y remained immobile; these elements are bound to Fe oxides and clays. Mobile elements such as Ni, Co, Li, Mo, W, Zn, Ce, Ru, and Pb are associated with Mn oxides at the base of the lateritic profile. The ΣREEs content in the laterite profile reaches up to 375 ppm. Cerium shows a weak positive correlation with the rest of REE, while positive Ce anomalies are associated with zones of Mn and Ce oxide enrichment. While there is no known magmatic sulfide mineralization associated with the host rock, chromite with Ru <0.15 ppm and inclusions of gold, Ni-Co-Cu-PGE-bearing sulfides, and sulfarsenides emerge as potential indicators for sulfide saturation. 

How to cite: Salama, W., Schoneveld, L., and Verrall, M.: Geochemical behavior of gold and critical metals in the Goongarrie Ni-laterites, Western Australia, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-1272, https://doi.org/10.5194/egusphere-egu24-1272, 2024.

Even if natural occurrence of Re can be found as rheniite (ReS₂), most Re substitutes to Mo in molybdenite (MoS₂), which explains why Re is usually a by-product of Cu-Mo deposits. In Thrace region (NE Greece), molybdenite can be enriched up to 4.7 wt% in Re in porphyry-epithermal deposits (Voudouris et al., 2013). Now, spectroscopic portable tools (e.g. pXRF) allows to directly detect Re in the field. First qualitative results obtained by pXRF show that is possible to know in which deposits from Greece the molybdenite is the most enriched without the necessity of long and costly laboratory measurements (EDS-SEM or EPMA). The X-250 pXRF (SciAps) used in this study do not include Re in its quantification program. Moreover, the spot diameter (4 mm) is generally larger than the molybdenite size in this area. Hence, Re cannot be quantified and even if it were, the value would be that of Re in the analytical spot and not in molybdenite only. The aims of this study is to (1) directly quantify Re with the pXRF and (2) determine the concentration in Re within the molybdenite.

In Energy-dispersive XRF, there is an interference between the Zn-Kα emission line (8.6389 keV) and the Re-Lα emission line (8.6524 keV). If Zn is quantified and Re not included into the analytical program, the Re signal will be interpreted as Zn quantities. That is the case with our X-250 pXRF. In Thrace region, molybdenite occurs in quartz veins sometimes associated with few feldspar or pyrite but no Zn-bearing minerals. When measuring molybdenite-bearing veins, all the Zn quantitatively measured by the pXRF corresponds to a Re signal. That effect can be corrected by applying a correction factor on the Zn value to convert it into a Re quantity in situ by using calibration curves. A specific user method can also be easily implemented into the tool. In case Zn-bearing minerals are also found in the molybdenite-bearing veins, the in situ method requires a multilinear correction of signal obtained from the ROIs of Zn and Re. That is more difficult to implement and in the meanwhile, one can obtain signals from Zn and Re separately by proceeding to spectral decomposition using the PyMCA software (Solé et al., 2007). Once separated, these signals can be converted into concentrations of each elements with calibration curves. These curves have been built from the measurement of reference samples consisting in chosen proportions of SiO₂ (considered as the matrix), MoS₂, Re and Zn powders. That enabled to evaluate the impact of each parameter on detection limit, precision and accuracy of the Mo, Re and Zn concentrations. The calibration curves were tested by the use of a set of validation samples.

In our case study, Re and Mo are only within molybdenite. The quantity of Re in the analysed area is mainly induced by the quantity of molybdenite, thus Mo, in the same area. The effect of Re-enrichment in molybdenite appears as a second order phenomenon. With these developments, Re-enrichment in molybdenite becomes a mapable parameter.

How to cite: Jatteau, M., Cauzid, J., Tarantola, A., and Voudouris, P.: In situ Rhenium (Re) quantification by pXRF in Molybdenite from porphyry-epithermal deposits (Thrace, NE Greece), EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-12933, https://doi.org/10.5194/egusphere-egu24-12933, 2024.

The Kyzylkum province in Southern Tien Shan, Uzbekistan, which includes the world's largest gold deposit Muruntau, is among the world's major gold provinces. Analysis of all available geological and geophysical data has revealed ten linear trends controlled by regional strike-slip shear zones in the ore-bearing sand-shale sediments O₂-S₁. Analogies were made with gold trends in North-Eastern Nevada (Carlin trend and others). Gold is present predominantly as microinclusions in pyrite and arsenopyrite in both provinces. A significant part of inclusions is represented by invisible nano-sized gold particles.

Opportunities of increasing gold output in the Kyzylkum province are connected with the reduction of losses of "invisible" gold in the process of ore concentration and processing. Innovative technologies and modern laboratory equipment were used to determine nano-sized gold concentration. Another source of increasing gold production is exploration and development of ore-controlling structures overlapped by younger sediments. Remote sensing methods are widely used to identify promising parts of the structures.

How to cite: Antonov, A., Tsoy, V., and Nurtaev, B.: Application of innovative technologies for increasing gold production in Kyzylkum province of Uzbekistan, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-21276, https://doi.org/10.5194/egusphere-egu24-21276, 2024.

The Li mineralization of Valdeflórez (Cáceres province, Spain) is related to an extensive metasomatism of Ordovician metasedimentary rocks due to the circulation of B- and Li-rich magmatic hydrothermal fluids associated probably with a granite dome (Torres-Ruiz, et al., 1996). Li-rich micas mainly, and other Li-rich minerals like amblygonite-montebrasite, appear as consequence of this hydrothermal alteration. The aim of this study is to evaluate the use of the VNIR-SWIR spectroradiometry to the materials classification according to the Li content during the exploitation mining works and to the mining prospection in similar areas. The results of a study conducted by VNIR-SWIR portable spectroradiometry on 335 samples coming from representative cores of the mineralized area are shown. Complementary, a mineralogical study by XRD and chemical analysis by ICP were performed. Visible and infrared wavelength ranges were studied separately, and a classification of the high-resolution spectra was done to compare with the chemical and mineralogical composition of the samples. The spectra were classified into six groups according to their morphology in the near and short-wave IR range, and these groups correspond to a different mineralogy of major components, as logical.  However, there is not a clear relation among these groups and the Li-content because Li is mainly in micas as octahedral substitutions, which influence on the spectral features is neglectable. A higher content in micas does not imply a higher content in Li. Consequently, a spectral signature of Li could not be determined, because Li is not directly related to the content of a certain mineral, micas in this case. As a conventional spectral signature is not useful in this case, because it classified mineralogy but not Li-content, a detailed study of the hight resolution spectra obtaining different parameters (both from the spectra and from the second derivatives of the spectra) was performed. The joint statistical treatment of the hyperspectral, mineralogical, and chemical data allowed us to find a plot of the materials classification according to the Li content from the spectral data which means a very fast procedure for classification that could be automatized though IA.

Acknowledgement: TED2021-130440B-I00 funded by MCIN/AEI/10.13039/501100011033 and European Union NextGenerationEU/PRTR

Torres-Ruiz, J., Pesquera, P., Gil, P., Casas, J. 1996. Tourmalinites and Sn-Li mineralization in the Valdeflores area (Cáceres, Spain). Mineralogy and Petrology 56:209-223.

How to cite: García Romero, E., Barrios, S., Valls, D., and Suárez, M.: FIELD VNIR-SWIR SPECTRORADIOMETRY FOR CLASSIFICATION OF MATERIALS IN THE VALDEFOREZ DEPOSIT (SPAIN) ACCORDING TO THE Li CONTENTS., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7758, https://doi.org/10.5194/egusphere-egu24-7758, 2024.

Recent research has highlighted significant correlations between hyperspectral data and the petrophysical properties of geological formations. Petrophysics acts as the link between geology and geophysics, and is crucial for constraining geophysical inversions, regional characterisation and mineral exploration. In this study, we employ various machine learning methods to predict P-Wave velocities using hyperspectral borehole data, with a focus on cross-validation between different boreholes in the same region. Our dataset includes 4022 paired observations of P-wave velocities, obtained from downhole sonic logging in one borehole, and corresponding hyperspectral data spanning visible-near (380–970 nm), shortwave (970–2500 nm), midwave (2700–5300 nm), and long-wave (7700–12300 nm) spectra, averaged over a 10 cm x 5 cm area. We utilised principal component analysis (PCA) for dimensionality reduction. The initial PCA stage extracted 10 principal components from each sensor type, which were then integrated. A subsequent PCA stage was conducted to reduce inter-sensor correlation, yielding 10 composite features that represent the variability across the complete VNIR-LWIR spectrum.

To validate our model, we conducted tests using 1160 pairs of analogous measurements from a different borehole within the same geological region. The model demonstrated impressive predictive capabilities, particularly with Support Vector Regression (SVR) and Artificial Neural Networks (ANN). The test set yielded R2 scores of 0.758 for SVR and 0.811 for ANN, indicating strong predictive accuracy. Building upon this success, our future work will expand the scope of prediction to include various other petrophysical properties critical to geophysical characterization and mineral exploration, such as S-Wave velocity, magnetic susceptibility, and rock density, properties which are critical for geophysical characterization and mineral exploration.

How to cite: Kamath, A., Kirsch, M., Thiele, S., and Gloaguen, R.: Predicting petrophysical properties from hyperspectral borehole data, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7969, https://doi.org/10.5194/egusphere-egu24-7969, 2024.

Europe holds significant potential for crucial metals essential for renewable energy and digital advancements. However, there is a need for a deeper understanding of the European subsurface, coupled with a requirement to reduce the environmental impact of exploration activities. We employ hyperspectral scanning of legacy drill cores to develop innovative indicators of mineralization in the Central European Kupferschiefer district, home to Europe's largest copper and silver resources. Hyperspectral imaging, capturing spectral reflectance and emission across numerous spectral bands, allows for non-invasive, high-resolution mineral mapping of drill cores. Initial findings showcased the technique's ability to identify critical redox boundaries and alteration minerals, aiding ore deposit characterization. In the framework of two research projects, we scanned 2400 meters of drill core from 87 boreholes across the Spremberg–Graustein Kupferschiefer deposit in Germany. For data acquisition, we deployed a drill-core scanner with a full suite of hyperspectral sensors covering the visible and near-infrared (400 to 970 nm), shortwave (970 to 2500 nm), mid-wave (2700 to 5300 nm), and longwave infrared (7700 to 12300 nm) ranges. The collected data were processed through a novel, open-source software pipeline, which enables i) real-time correction, processing, and analysis, ii) efficient data management and storage, and iii) comprehensive visualization and integrative interpretation of the hyperspectral drill core data. We upscaled mineral abundances across all of the scanned drill cores using a supervised learning model trained on quantitative mineralogical data from select samples. Initial analyses, particularly the visual alignment of hyperspectral derivatives at the base of the Kupferschiefer marker horizon, indicate geographical patterns in dolomite content and correlations between carbonate, clay, and mica compositions and copper grade. The hyperspectral data will eventually be integrated with geological, geophysical, and geochemical constraints to create accurate 3D subsurface and 4D mineral system models, aimed at enhancing our understanding of geological processes and resource management strategies in similar geological settings worldwide.

Acknowledgements: This research has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement nº 101058483 (VECTOR), and from the Geological Survey of Saxony (Sächsisches Landesamt für Umwelt, Landwirtschaft und Geologie, LfULG) under agreement nº 4-0912014LFULG01-88.

How to cite: Kirsch, M., Thiele, S., Madriz, Y., Simões, F., Basoglu, A., Kim, Y., and Gloaguen, R.: Towards hyperspectral exploration vectors in the Central European Kupferschiefer district, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14928, https://doi.org/10.5194/egusphere-egu24-14928, 2024.

Rare earth elements (REE) are key constituents in electronic devices (e.g. smartphones, batteries), being present in both end-user and industrial applications. The rapid innovation cycles of electronic devices, combined with the increasing demand for new technological applications (e.g. mobility and e-cars) pose a challenge for the supply of REE, which are considered as Critical Raw Materials (CRM). This scenario calls for rapid, non-invasive methods that enable the identification of new REE-rich mining resources. Furthermore, the high supply risks associated with CRM such as REE drive technological developments to compensate and overcome market fluctuations by turning previously not mined co-resources into valuable and economic modalities, such as re-mining materials.

We present an investigation focused on the identification of REE in waste rocks and tailing materials from the mine of Siilinjärvi (Finland). The deposit in the area consists of alkaline-carbonatite rocks, with the most important REE-bearing minerals being apatite (average REE concentration: 0.4% (wt%)) and monazite (REE concentration: up to 67% (wt%)). Mining activities focus on extraction of phosphate from fluorapatite, and the chemical reactions involved in this extraction generate phosphogypsum (PG) as a by-product. Literature reports indicate that REE can be incorporated to the PG matrix in the crystallisation process, with the most relevant examples including Nd, Ce, La, Sm, Gd, Tb, Dy, and Eu. 

Our goal is to highlight how the sequential acquisition by multiple optical methods (multi-sensor approach) can trace REE contents for individually identified REE from pristine rocks to processing waste dumped in tailings. Each material type was scanned by two fast hyperspectral imaging (HSI) sensors integrated in a conveyor-belt system:  a reflectance-based HSI sensor operating in the visible to near-infrared and short-wave infrared (Specim AisaFenix); and an innovative laser-induced fluorescence line scan sensor (HSI-LiF, Freiberg Instruments). The optical sensing results were validated by mineralogical methods (mineral liberation analysis (MLA)). MLA results for PG indicate the presence of REE-bearing minerals including gypsum, apatite, and monazite (respective abundances (wt%): 97.4, 0.6, and 0.08).

Optical features characteristic of Nd were identified on rocks and tailings samples by both HSI-reflectance and HSI-LiF sensors. Spectral signatures were detected in HSI-LiF spectra for an additional REE group including Sm, Er, and Pm.

We highlight that efficient, non-invasive optical sensing can detect and re-evaluate tailing materials as a baseline for economic considerations according to market needs. The results confirm that REE detected on the pristine rocks of the mine can be traced through the mineral processing route to be found again in the tailings material. The multi-sensor optical detection based on HSI-reflectance and HSI-LiF, accordingly, provides an efficient non-invasive tool for exploring both mining and re-mining potential by providing immediate results on REE types and their spatial abundance, when employed as scanning techniques.

This investigation was performed within the scope of EIT-funded projects (inSPECTOR and RAMSES-4-CE).

How to cite: de Lima Ribeiro, A., Abend, T., Fuchs, M., Röder, C., Beyer, J., Kärenlampi, K., Xiao Sheng, Y., Heitmann, J., and Gloaguen, R.: REE (re)cycle: a multi-sensor investigation from rocks to tailings, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20261, https://doi.org/10.5194/egusphere-egu24-20261, 2024.

Optical sensors are a key enabler for an in-line, real-time characterisation, quality control and monitoring in industrial, conveyor-based raw material processing. Innovators are actively exploring non-invasive optical sensing to solve current problems in economic, socially acceptable and ecologic resource handling with high efficiency. Despite the evident advantages, integrating available optical technologies into sensor systems poses various challenges. A detailed understanding of physical parameters as well as smart solutions are required to mitigate or circumvent some of the limitations. Realistic solutions for the industry rely on understanding what is possible, where are the key limiting factors and which of the challenges can be overcome in the future.

In this contribution, we present four examples from our HELIOS lab research projects in the field of recycling of society-relevant material streams to discuss the major challenges. Our focus lies on the suitability of optical sensor systems for industrial applications. We emphasize the pathways from scientific setups to industrial demonstrators and highlight the relevant parameters, when operating sensors such as RGB, hyperspectral reflectance imaging (HSI), laser-induced fluorescence (LiF) together with Raman scattering, x-ray fluorescence (XRF) and laser-induced breakdown spectroscopy (LIBS).  

Extremely relevant to the industry is speed (or material throughput). Common conveyor belt speeds of several meters per second imply low signal integration times for the optical sensors (or high frame rates in the case of cameras). While industrial high-speed RGB cameras are well suitable, HSI cameras rely on longer integration times to collect signals with adequate intensities across hundreds of detection bands. Current HSI technology is successful in a variety of conveyor belt applications (2D dynamic setup) at a few meters per second, however, a transfer to applications for material detection in air flows (3D dynamic setup) outlines the trade-off between signal quality and acquisition speed. Similarly, signals of very low intensities as seen in laser-induced fluorescence hyperspectral scanning highlight the multi-parameter trade-off between integration times, acquisition speed and excitation power, where the latter is largely dependent on available optical components.

Most of our consumer products are not made of pristine, pure material but come with coating, as compounds and/or with additives to improve appearance and performance of the materials. For recycling, this poses significant challenges for material separation and processing. Using optical sensors in recycling operations then often implies extrapolating the surface properties as representative of the actual material. We demonstrate with examples from several projects, how coating and additives affect the spectral signatures in polymers (esp. black polymers) and metals (steel and aluminum), and how a combination with additional validation sensors (e.g. Raman, LIBS) can provide important information in materials. This information is essential for an adequate and high-quality recycling process.

The given examples and related research are based on collaborations with the industry and aim at developing and testing new concepts and evaluating corresponding tools for data acquisition and real-time processing in recycling facilities. We gratefully acknowledge project funding for RAMSES-4-CE (KIC RM 19262), Digisort (03XP0337B), Car2Car (19S22007B) and FINEST (KA2-HSC-10).

How to cite: Fuchs, M. C., Lorenz, S., Madriz Diaz, Y. C., de Lima Ribeiro, A., Arbash, E., Beyer, J., Röder, C., Schüler, N., Dornich, K., Heitmann, J., and Gloaguen, R.: Challenges of in-line, sensor-based characterisation of recycling streams, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15719, https://doi.org/10.5194/egusphere-egu24-15719, 2024.

The majority of the world’s known copper reserves is contained in porphyry copper deposits. These deposits are understood to form along subduction zones within the magmatic arc, though the exact contributions to this process of different factors within the subducting and overriding plates are not entirely certain, hampering efforts to develop large-scale prospectivity models for this deposit type. Previous efforts to tackle this problem through the use of data-driven machine learning methods have shown promise, but have been hindered by the relative paucity of fully labelled data required for training classification models. Here we present a suite of models trained using a semi-supervised positive-unlabelled (PU) algorithm, allowing the classifier to be created from data of which only a small subset of one class is labelled: in this context, known porphyry copper deposit locations. These models can be used to create time-dependent prospectivity maps representing the probability of a deposit forming at a given place and time, while model inspection can provide deeper insight into the processes behind the genesis of these deposits, at both a global and regional scale. Furthermore, deep-time erosion rate estimates extracted from global landscape evolution models are used to explore the uplift and erosion histories of known porphyry copper deposits in order to better understand the potential for these deposits to survive to the present day. These two factors of deposit formation and preservation are combined into a powerful prospectivity model, with the potential to facilitate the identification of possible new prospective zones along the world's subduction zones through time and help minimise the environmental and financial costs of mineral exploration.

How to cite: Alfonso, C., Müller, D., Mather, B., and Salles, T.: Predicting global porphyry copper prospectivity using positive-unlabelled machine learning, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6901, https://doi.org/10.5194/egusphere-egu24-6901, 2024.

Mineral Prospectivity Mapping (MPM) is an important tool to identify areas with significant potential to host mineral deposits. Recent advancements in computational sciences, especially the advent of Machine Learning (ML), have enhanced MPM's capabilities. ML techniques enable a higher degree of data integration and extraction compared to traditional statistical methods such as Weights of Evidence, enhancing the accuracy and efficiency of identifying mineral exploration zones. When using ML techniques for MPM, the influence of negative training labels (ie. barren areas with no mineralization) remains a neglected research area, and this study investigates the influence of such label selection to optimize predictive models for Canadian critical mineral exploration.

Previous approaches to ML-based MPM often adopted a random assignment of negative training labels wherever positive training labels were absent. This study aims to refine this method, striving for a more systematic approach in negative label selection. The evolution of MPM, transitioning from traditional statistical methods to modern ML algorithms, signifies a shift towards heightened accuracy and efficiency. Prior research underscores the importance of balanced representation between mineralized and non-mineralized labels in ML models. Techniques such as Synthetic Minority Over-Sampling (SMOTE) and Positive and Unlabelled Learning (PUL) have been highlighted in previous studies, emphasizing the necessity of effectively handling negative training labels to prevent biases and enhance model performance. While SMOTE and PUL synthetically balance datasets by either oversampling minority classes or considering only positive and unlabeled instances, this study focuses on leveraging public exploration data to identify real negative training labels and provide a more authentic representation of non-mineralized areas without synthetic augmentation.

Using datasets compiled by the Geological Survey of Canada containing discoveries & occurrences of magmatic Ni (±Cu ±Co ±PGE), this research incorporates geological, geochemical, and geophysical data from established sources. Public exploration data will be used to identify areas devoid of magmatic Ni (±Cu ±Co ±PGE). These locations will serve as negative training labels for this study. Our particular choice of ML model is a Gradient Boosting Machine (GBM), and validation involves comprehensive evaluation techniques such as confusion matrices and receiver operating characteristic curves to assess model performance.

How to cite: Bay, N., Noh, K., Parsasadr, M., and Swidinsky, A.: Machine Learning-based Mineral Prospectivity Mapping: Exploring the Role of Negative Training Labels to Enhance Predictive Models, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-820, https://doi.org/10.5194/egusphere-egu24-820, 2024.

As conventional hydrocarbon resources become depleted and theoretical innovations in hydrocarbon exploration advance, unconventional resources have gained substantial attention from researchers and explorers in recent decades. In the global energy consumption structure, unconventional shale gas progressively assumes a crucial role in the overall energy landscape. This research is motivated by the high probability of deriving gas accumulations encountered in drilling on the northwestern Anatolia (Akcakoca area) offshore from Paleozoic-aged rocks. Studied black shales from the Silurian Findikli Formation in the Sakarya region of northwestern Anatolia are one of the alternative unconventional resources.

Working with outcrop samples, this work evaluates the unconventional gas potential by performing geochemical characterization of these black shale samples. Studied samples were analyzed by Rock-Eval Pyrolysis. The present-day total organic carbon (TOC_pd) contents range from 0.54 to 1.57 wt.%. High Tmax (up to 504^oC) and low Hydrogen Index (HI) values (4-38 mg HC/g TOC) indicate that these shales are thermally over-mature and seem to be a spent hydrocarbon source rock. The remaining hydrocarbon generative potential (S1+S2) of 0.06–0.49 mg HC/g rock also supports this assessment. The recent organic matter type is Type IV kerogen which can yield limited gas products plotting on the H/C versus O/C atomic ratios on the modified Van Krevelen diagram. According to the interpretive of shale gas potential parameters based on Jarvie’s equation reconstructed these black shales originally may have good to very good source rock potential, with the calculated average original values of TOC_o being 1.72 wt.% and HI_o is 448 mg HC/g TOC. It can be concluded that the characteristics of studied shales seem to be nearly compatible with those of Utica shales in terms of the hydrocarbon generative potential. The studied black shales have lost 95% of their original hydrocarbon potentials, and seem to be a spent hydrocarbon source rock, indicating good risk for gas. However, the source rock may be very deep and deformed from the past to the present day, considering the paleogeographical location and geological evolution of the study area. Further research is required.

Keywords: Unconventional Gas Potential, Source Rock, Geochemical Characteristics, Silurian Black Shales, Northwestern Anatolia (Türkiye)

How to cite: Doner, Z.: Unconventional Gas Potential of Black Shales from Silurian Findikli Formation in northwestern Anatolia, Türkiye: Characterization of Geochemistry, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-738, https://doi.org/10.5194/egusphere-egu24-738, 2024.

The Koillismaa layered igneous complex (KLIC) in northern Finland spans a large distance from the Finnish-Swedish border to the Finnish-Russian border. It has been an area of interest for several decades among geologists due to its potential to host several critical raw materials including cobalt & nickel which are key materials required for lithium-ion batteries. The KLIC intrusion comprises outcropping Koillismaa and Näränkävaara layered complex mafic-ultramafic intrusions that are interestingly connected by a zone of high gravity and magnetic anomalies. Extensive petrophysical and lab studies were conducted by the Geological Survey of Finland on a 1.7 km long deep drillhole located within the area of our interest. A pre-existing low-fold seismic study indicated the reflectivity of the ultramafic rocks. Drillhole study further established the fact that the contacts with mafic intrusion rocks having the potential to host mineralization are causing observable seismic reflections at 1.4 km depth. The ongoing SEEMS DEEP project (2022-2025), an ERA-MIN 3 sponsored project comprises an integrated approach of using seismics and electromagnetics studies to substantially improve the geomodel of the Koillismaa area which will help in better decisions in exploration drilling. In this study, we are focussing on the seismic part of the project. In August 2023, an irregular-sparse 3D seismic and two regional 2D seismic lines, roughly in the direction of E-W and NNE-SSW were acquired under the SEEMS DEEP project. The overall aim of the 2D seismic lines was to constrain large-scale information about the geological architecture of the study area, whereas the 3D survey was conducted to highlight the detailed information from the rock volumes near the deep drill hole. The acquired data, both 2D (~10 Km and ~12 Km) and 3D (~5 Km x 6 Km), are of good quality with reflections visible in the raw data. An 8-tonne Mark IV Vibroseis truck was used as a seismic source for both surveys, with sweep frequency ending at 160 Hz. Almost 3000 single-component receivers (Strydes) were deployed in varying subarctic terrains defined by swamps and forests for the 3D survey. For 2D profiles, over 700 three-component receivers (GSBs) were used next to the existing gravel roads utilized by the Vibroseis truck. For the 3D survey, receiver spacing was kept at 30 m with inline spacing of 200 m. Shot points were located mainly following the existing roads within the forest. For the 2D survey, a uniform receiver spacing of 15 m and a shot spacing of 30 m was used. Seismic data processing was applied with the overall aim of suppressing noise, boosting signal-to-noise ratio, and improving reflectivity in the data. Special emphasis was put on handling the highly heterogeneous near-surface weathering layer. Final results revealed several reflectors at various depths, which have preliminarily been interpreted to originate from mafic intrusions, diabase veins, and faults cross-cutting the intrusion.

How to cite: Singh, B., Amirzadeh, Y., Autio, U., Gόrszczyk, A., Heinonen, S., Malinowski, M., and Wojdyła, M. and the SEEMS DEEP Working Group: First results from the SEEMS DEEP seismic survey conducted over the Koillismaa Igneous Complex, Northern Finland, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17303, https://doi.org/10.5194/egusphere-egu24-17303, 2024.

Geophysical exploration was carried out on karst bauxite deposits in the Posušje area in Bosnia and Herzegovina, which generally represent the problem of researching bauxite deposits in the Dinarides and similar geological models in the Mediterranean. Karst-type bauxite deposits are found in very complex geological models and were deposited in depressions in carbonate bedrock during numerous emersions on the Adriatic carbonate plate. Carbonate or clastic rocks can be found as the hanging wall of the deposits. Due to the complex lithological and structural relations, very irregular shapes of the deposits and their relatively small dimensions, the discovery of karst-type bauxite deposits is a very demanding geophysical task. That is why the published literature falls short in offering solutions for this very complex problem. So, the fundamental question is whether very irregular bauxite deposits whose dimensions are generally small can be detected by geophysical exploration?

Basic near surface geophysical methods, electrical resistivity tomography and seismic refraction, as well as magnetometry were applied in the exploration, which was carried out in two phases. In the first phase, geophysical methods were applied to already discovered bauxite deposits in order to determine whether geophysical responses correlate with bauxite deposits and to evaluate the efficiency of each method. Geophysical measurements were performed at several microlocations, and in the area of Mratnjača were carried out immediately after the deposit discovery, which was subsequently mapped with a very dense network of exploratory boreholes and was very well defined. In the second phase, measurements were performed in an microlocation selected by geological prospecting in order to discover potentially new bauxite deposit.

The characteristic responses of bauxite deposits are expressed on the resistivity models of tomographic profiles as zones of lower resistivities within carbonate rocks, and on the velocity models of refraction profiles as velocity inversions. The responses are much clearer on resistivity models, so the electrical tomography should be considered as a fundamental method in the exploration of karst bauxite deposits. Seismic refraction can contribute to a better characterization of deposits and reduce the interpretation ambiguity, thereby increasing the efficiency of geophysical exploration. In the last case, the electrical tomography can be applied independently to give satisfactory exploration results. On the other hand, the seismic refraction should be combined with the electrical tomography, because in some cases the depth coverage is greatly reduced due to distinct velocity inversions. Unfortunately, magnetometric measurements showed there are no magnetic anomalies that could be associated with bauxite deposits, that is, there are no magnetic minerals in the deposit.

Acknoledgments

This exploration was carried out in the AGEMERA project (Agile Exploration and Geo-modelling for European Critical Raw Materials) which has received funding under the European Union's Horizon Europe research and innovation programme under grant agreement No 101058178.

How to cite: Šumanovac, F., Kapuralić, J., Perković, L., and Pavičić, I.: Assessment of geophysical methods in the discovery of karst bauxite deposits in the Dinarides, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-2491, https://doi.org/10.5194/egusphere-egu24-2491, 2024.

The European Commission classifies graphite as a critical raw material, given its importance in refractory and high-tech applications, including the production of lithium-ion batteries. Many of Austrias graphite deposits are situated in and around the so-called “Drosendorf-Deckensystem” in Lower Austria. As those sites have been extensively mined out in the near-surface and are heavily weathered, the current interest of the potential assessment primarily focuses on the spatial extent of the graphite deposits, with particular interest in the deep continuation of these graphite deposits. Geological cross-sections are a standard approach for understanding subsurface graphite distribution, relying on available structural, geochemical, and/or lithological data. However, such information is primarily derived from surface observations, limited to exposed surfaces or outcrops. Geophysical methods can play a crucial role in extending the interpretation of geological cross-sections both horizontally and vertically. Due to the high electrical conductivity of graphite, electromagnetic and electrical methods like the Transient Electromagnetic (TEM) and the Induced Polarization (IP) are commonly applied for exploration. The IP method complements traditional electrical resistivity methods by measuring not only subsurface conductivity but also variations in the electrical capacitive properties (polarization) at low frequencies. In this study, we explore the feasibility of integrating the interpretation of IP and TEM measurements. The latter is known for its cost-effective coverage of large areas with high resolution and depth of investigation, compared to other geophysical methods. For that purpose, multiple TEM soundings were acquired in a ca. 4 km long profile between Berging and Kochholz in Lower Austria, situated in the geological section called “Drosendorf-Deckensystem”. The inversion of TEM data revealed distinct high-conductivity anomalies (100 – 200 mS/m) at approximately 40 m depth, attributed to the presence of graphite. A parallel comparison with two 160 m – 250 m long Time-Domain IP profiles confirms the graphite presence, attributed by high conductivity and high polarization (σ’ > 200 mS/m, σ” > 20 mS/m). Integrating these results along a geological cross-section improved the delineation and understanding of graphite deposits at depth and their correlation with lithological features in Lower Austria. Furthermore, these findings confirm that acquiring physical property models related to TEM and IP surveys has the potential to enhance mineral exploration.
This research work was co-financed by the Federal Ministry of education, science and research and supervised by the Federal Mining Authority of Austria within the framework of the VLG83 project (Raw Material research initiative).

How to cite: Monsalve Martinez, J. L., Aigner, L., Flores-Orozco, A., Moser, C., Högenauer, P., and Römer, A.: The delineation of graphite deposits in Lower Austria using the Transient Electromagnetic method, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-3882, https://doi.org/10.5194/egusphere-egu24-3882, 2024.

The AGEMERA project [1], which is an acronym for Agile Exploration and Geo-Modelling for European Critical Raw Materials, employs three non-invasive survey methods for mineral exploration: a passive seismic method to assess bedrock hardness and rock type boundaries; an integrated, multi-sensing fixed-wing drone system for measuring conductivity, magnetism, and radioactivity; and a multidetector system based on muon detection for detailed 2D, 3D, and 4D density profiles of large-volume rock bodies (with the 4^th dimension being time). The technologies are designed to map geological structures in scenarios where traditional methods are either environmentally unsound or socially challenging. By the project's conclusion, these methods are anticipated to achieve a Technological Readiness Level (TRL) of 5 within a three-year timeline.

The technologies vary in their operational capacities, including acquisition time, depth penetration, area coverage, and volume assessment. The multi-sensing drone effectively probes to 300-500 meter depth and can survey vast areas, up to hundreds of square kilometres, in a single campaign. Muography, on the other hand, can reach depths of up to 1000 metres and cover large volumes, up to a cubic kilometre. Passive seismic analysis, meanwhile, can survey any area and depth while a larger depth usually implies a lower resolution. While these techniques, especially when combined with deep 3D muography, may require extended periods for data collection, the valuable insights they offer make them a worthwhile investment.

After conducting these innovative, non-invasive geophysical surveys, the findings will be consolidated in a web-based data repository. This repository will be accessible for in-depth analysis to enhance our understanding of critical raw material distribution.

The project receives funding from the Horizon Europe program (Grant agreement ID: 101058178).

[1] AGEMERA project homepage, www.agemera.eu (accessed 9.1.2024)

How to cite: Joutsenvaara, J., Holma, M., Kuusiniemi, P., Pirttijärvi, M., Stimac Tumara, B., Schimmel, M., and Martin, D.: The Horizon Europe AGEMERA Project: Innovative Non-Invasive Geophysical Methodologies for Mineral Exploration, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19207, https://doi.org/10.5194/egusphere-egu24-19207, 2024.