Precursors predicted by artificial neural networks for mass balance calculations: Quantifying hydrothermal alteration in volcanic rocks

doi:10.1016/j.cageo.2016.01.003

Computers & Geosciences

Volume 89, April 2016, Pages 32-43

https://doi.org/10.1016/j.cageo.2016.01.003 Get rights and content

Highlights

•
Method for predicting the fresh precursor to hydrothermally altered volcanic rocks.
•
Precursors are predicted using multi-layer perceptron neural networks.
•
Precursor compositions (major oxides) are predicted from immobile elements ratios.
•
Chemical method for quantifying alteration from large datasets of volcanic rocks.
•
New chemical method developed for grassroot or regional mineral exploration.

Abstract

This study proposes an artificial neural networks-based method for predicting the unaltered (precursor) chemical compositions of hydrothermally altered volcanic rock. The method aims at predicting precursor’s major components contents (SiO₂, FeO^T, MgO, CaO, Na₂O, and K₂O). The prediction is based on ratios of elements generally immobile during alteration processes; i.e. Zr, TiO₂, Al₂O₃, Y, Nb, Th, and Cr, which are provided as inputs to the neural networks. Multi-layer perceptron neural networks were trained on a large dataset of least-altered volcanic rock samples that document a wide range of volcanic rock types, tectonic settings and ages. The precursors thus predicted are then used to perform mass balance calculations. Various statistics were calculated to validate the predictions of precursors’ major components, which indicate that, overall, the predictions are precise and accurate. For example, rank-based correlation coefficients were calculated to compare predicted and analysed values from a least-altered test dataset that had not been used to train the networks. Coefficients over 0.87 were obtained for all components, except for Na₂O (0.77), indicating that predictions for alkali might be less performant. Also, predictions are performant for most volcanic rock compositions, except for ultra-K rocks. The proposed method provides an easy and rapid solution to the often difficult task of determining appropriate volcanic precursor compositions to rocks modified by hydrothermal alteration. It is intended for large volcanic rock databases and is most useful, for example, to mineral exploration performed in complex or poorly known volcanic settings. The method is implemented as a simple C++ console program.

Introduction

Grassroots, or regional mineral exploration, necessitates the use of powerful and simple methods to interpret large amounts of chemical data. The interpretation and quantification of hydrothermal alterations in volcanic terrains, for instance, is particularly important as many economic substances are accumulated by the circulation of hydrothermal fluids in areas with magmatic activity, and because altered rocks are one of the main vectors used to explore for such mineralisations.

Geochemical studies of hydrothermally altered rocks aim at determining the amount of elements gained and lost during alteration of an initially fresh rock (i.e. the precursor), by performing mass balance calculations while assuming that one or more elements is immobile during alteration (see Gresens (1967)). Gresens' (1967) approach has been extensively and successfully used (MacLean and Kranidiotis, 1987, Shriver and MacLean, 1993, Cadieux et al., 2006; to name a few).

A major question that arises when using such mass balance equations concerns the choice of an appropriate precursor for the studied altered sample. Precursor’s compositions are usually obtained by analysing “fresh” samples (i.e. samples that lack alteration minerals) that have been geologically related to the altered sample using detailed knowledge of the local geology (e.g. Grant, 1986, Grant, 2005). Such approaches are possible only for detailed deposit-scale studies, if fresh rocks exist and if our level of knowledge of the area render their identification possible. Even so, usually nothing proves that the fresh sample is an exact match for the precursor to the altered sample, and these difficulties can make mass balance calculations un-reliable.

At a more regional scale, on the other hand, these difficulties can be in-surmountable. Thus, orebody targeting using chemical datasets more often rely on alteration indexes (ratio of major elements), which values are sensitive to the precursor’s compositions and could provide miss-leading indications on the intensity of alteration (see discussion in Trépanier et al. (2015)).

Mass balance calculations would be more reliable if a suitable precursor composition to an altered rock could be determined directly from the chemical characteristics of the altered sample, thus providing a precursor adapted to this sample.

This paper proposes a method for estimating the composition in SiO₂, CaO, MgO, FeO^T, K₂O, and Na₂O of fresh precursors to altered volcanic rocks, using ratios of commonly analysed least mobile elements (i.e. Zr, TiO₂, Al₂O₃, Y, Nb, Th, and Cr) and a regression method (i.e. artificial neural network). The method is provided as supplemental material, as a simple C++ program that calculates precursor compositions (this study) prior performing mass balance calculations (using published methods).

Section snippets

Description of the method

In this contribution, a method to predict fresh precursors from the chemical composition (i.e. immobile elements content) of altered volcanic rocks is presented. Predictions are performed for magmatic rocks only because relatively simple relationships, which are controlled by magmatic processes, exist between the trace and major elements content of such rocks. For example, Zr and Si are both incompatible elements that tend to increase during fractional crystallisation, making Zr a proxy for Si

Least-altered volcanic rocks dataset

Relationships between major and immobile elements are evaluated here with neural networks using a large chemical dataset of modern volcanic rocks from around the World. Post-Archean to modern rocks are used because they are well constrained samples for which names, geodynamic settings and freshness can be confidently established; i.e. characteristics that would be harder to determine for more ancient samples. Besides, there might be large differences between ancient (Archean) and modern

Software

This study aims at modelling the chemical composition of fresh precursors to altered rocks, in order to facilitate mass balance calculations. The outputs of the neural networks (i.e. the results of the training process) can however be difficult to use as such, as the software used to train the neural networks (NeuroDimension, 2012) provides outputs files with a “.NSW” extension, which are simple ASCII text files that record the network’s structure, parameters and weights.

To facilitate access to

Neural network validations

Prior to using the method to predict new sample’s precursors, the neural networks’ outputs were validated. The consequences that Zr, Y, Nb, Cr, Th, Al, and Ti mobility might have on the prediction’s accuracy are also tested. Overall, the proposed approach appears accurate, precise, and may tolerate a limited mobility of the immobile elements used by the model.

Method validation using case studies

In this section, several fresh volcanic rocks, which were not included in the Georoc database used to train the neural network, are used to validate the approach (see case studies 1–3). Then, altered rocks from several mineralised areas (i.e. volcanogenic massive sulphide, or VMS, and porphyry deposits) are compiled, their precursors compositions are predicted and mass balance calculations are performed (see case studies 4–6).

Discussion and conclusions

This study proposes a method for predicting the composition in major elements of precursors to hydrothermally altered volcanic rocks, using artificial neural networks. To facilitate its utilisation by the reader, the method is embedded in a C++ console program that also performs mass balance calculations, allowing for the evaluation of alteration, in terms of type and magnitude, from large datasets of volcanic rocks.

Neural networks used here “learned” to predict the composition in major

Computer code

The method presented in this contribution is provided as a computer code (C++ language).

Acknowledgements

The authors wish to address special thanks to the editor and reviewers of the manuscript, Jef Caers and David Lentz, as well as an anonymous reviewer. This project was supported by Canada Economic Development, the Ministère de l'Énergie et des Ressources Naturelles du Québec, the Conférence Régionale des élus Saguenay-Lac-Saint-Jean and companies members of the Consorem. The authors warmly thank their colleague Silvain Rafini for constructive discussions on this project, Geneviève Boudrias for

References (52)

J.H. Bédard
A catalytic delamination-driven model for coupled genesis of Archaean crust and sub-continental lithospheric mantle
Geochim. Cosmochim. Acta
(2006)
J.R. Cann
Rb, Sr, Y, Zr and Nb in some ocean-floor basaltic rocks
Earth Planet. Sci. Lett.
(1970)
J.A. Grant
Isocon analysis: a brief review of the method and applications
Phys. Chem. Earth Parts A/B/C
(2005)
S.-Y. Jiang et al.
Mobility of high field strength elements (HFSE) in magmatic-, metamorphic-, and submarine-hydrothermal systems
Phys. Chem. Earth
(2005)
H.B. Mattsson et al.
Petrogenesis of alkaline basalts at the tip of a propagating rift: evidence from the Heimaey volcanic centre, south Iceland
J. Volcanol. Geotherm. Res.
(2005)
M.T. McCulloch et al.
Geochemical and geodynamical constraints on subduction zone magmatism
Earth Planet. Sci. Lett.
(1991)
M. Meschede
A method of discriminating between different types of mid-ocean ridge basalts and continental tholeiites with the Nb–Zr–Y diagram
Chem. Geol.
(1986)
J.A. Pearce
Geochemical fingerprinting of oceanic basalts with applications to ophiolite classification and the search for Archean oceanic crust
Lithosphere
(2008)
J.A. Pearce et al.
Tectonic setting of basic volcanic rocks determined using trace element analyses
Earth Planet. Sci. Lett.
(1973)
P. Schiano et al.
Application of element concentration variability to the study of basalt alteration (Fangataufa atoll, French Polynesia)
Chem. Geol.
(1993)

J.A. Winchester et al.

Geochemical discrimination of magma series and their differentiation products using immobile elements

Chem. Geol.

(1977)

D.A. Wood

The application of a Th-Hf-Ta diagram to problems of tectonomagmatic classification and to establishing the nature of crustal contamination of basaltic lavas of the British Tertiary Volcanic Province

Earth Planet. Sci. Lett.

(1980)

T.J. Barrett et al.

Chemostratigraphy and hydrothermal alteration in exploration for VHMS deposits in greenstones and younger volcanic rocks

E. Boari et al.

Transition from ultrapotassic kamafugitic to sub-alkaline magmas: Sr, Nd and Pb isotope, trace element and ⁴⁰Ar–³⁹Ar Age data from the Middle Latin Valley Volcanic Field, Roman Magmatic Province, Central Italy

J. Petrol.

(2009)

Cadieux, A.-M., Dubé, B., Williamson, K., Malo, M., Twomey, T., 2006. Characterization of hydrothermal alterations at...

J.F. Cohen

Mineralogy and Geochemistry of Hydrothermal Alteration at the Ann- Mason Porphyry Copper Deposit, Nevada: Comparison of Large-scale Ore Exploration Techniques to Mineral Chemistry (Unpublished Master Thesis)

(2011)

Cohen, J.F., Dilles, J.H., Tosdal, R.M., Halley, S., 2011. Compositional variations in hydrothermal white micas and...

J.H. Dilles

Footprints of Porphyry Cu Deposits: Vectors to the Hydrothermal Center Using Mineral Mapping and Lithogeochemistry

(2012)

Drummond, M.S., Defant, M.J., 1990. A model for trondhjemite‐tonalite‐dacite genesis and crustal growth via slab...

T. Finlow-Bates et al.

The behaviour of so-called immobile elements in hydrothermally altered rocks associated with volcanogenic submarine exhalative ore deposits

Miner. Deposita

(1981)

GEOROC – Geochemistry of Rocks of the Oceans and Continents, 2011. Max-Planck-Institut für Chemie....

J.A. Grant

The Isocon Diagram – a simple solution to gresens’s rquation for metasomatic alteration

Econ. Geol.

(1986)

R.L. Gresens

Composition–volume relationships of metasomatism

Chem. Geol.

(1967)

A.R. Hastie et al.

Classification of altered volcanic island arc rocks using immobile trace elements: development of the Th–Co discrimination diagram

J. Petrol.

(2007)

I.G. Hill et al.

Yttrium: the immobility–mobility transition during basaltic weathering

Geology

(2000)

K. Hornik et al.

Multilayer feedforward networks are universal approximators

Neural Netw.

(1989)

Cited by (22)

Felsic crust development in the Kaapvaal Craton, South Africa: A reference sample collection to investigate a billion years of geological history
2024, Earth-Science Reviews
The crust of the Kaapvaal craton accreted throughout the Archaean over nearly 1 billion years. It provides a unique example of the various geological processes that shape Earth's continental crust, and is illustrated by a reference collection of granitoids and mafic rocks (SWASA collection). This sample collection is fully characterised in term of age, major and trace elements, and documents the following multistage history of the craton. In the Barberton area, the initial stages of accretion (stage B.I, > 3.33 Ga and B.II, 3.28—3.21 Ga) correspond to the formation of a sodic (TTG) crust extracted from a near-chondritic reservoir. Stage B.III (ca. 3.1 Ga) corresponds to reworking of this crust, either through intracrustal melting, or via recycling of some material into the mantle and melting of this enriched mantle. Stage B.IV (2.85—2.7 Ga) corresponds to the emplacement of small, discrete plutons involving limited intracrustal reworking. The Northern Kaapvaal craton corresponds to a mobile belt flanking the Barberton cratonic core to the North. Stage NK.I (> 3.1 Ga) resembles stages B.I and B.II: formation of a TTG crust from a chondritic reservoir. In contrast, stage NK.II. (2.97–2.88 Ga) witnesses probable rifting of a cratonic fragment and formation of greenstone basins as well as a new generation of TTGs with both the mafic and felsic magmatism extracted from an isotopically depleted mantle (super-chondritic) reservoir. Intra-crustal reworking dominates stage NK.III (2.88–2.71 Ga), whereas sanukitoids and related granites, involving a mantle contaminated by recycled crustal material, are common during stage NK.IV (ca. 2.67 Ga).
The SWASA collection are available as a reference collection to investigate the behavior of other systems of interest during a variety of crust evolution modes.
Applying neural networks-based modelling to the prediction of mineralization: A case-study using the Western Australian Geochemistry (WACHEM) database
2023, Ore Geology Reviews
Citation Excerpt :
Multiple attempts have been made to apply machine learning (ML) methods to mineral exploration (Caté et al., 2017; Enkhsaikhan et al., 2021; Grunsky and Arne, 2021; Kirkwood et al., 2016; Prado et al., 2020; Rodriguez-Galiano et al., 2015; Sun et al., 2019; Wang et al., 2020; Wedge et al., 2019; Zuo and Xiong, 2017). Some specific examples include prediction of the presence of gold mineralization in drill core from well logs (Caté et al., 2017), deposit estimation from multi-element geochemical data (Hill et al., 2014), prediction of soil geochemistry from high resolution geophysical and remote sensing data (Kirkwood et al., 2016), and estimation of precursor chemical compositions of hydrothermally altered volcanic rocks from the ratios of commonly analyzed least mobile elements (Trépanier et al., 2016). However, all these approaches utilize either shallow neural networks, or methods such as support vector machines (SVMs), regression trees (RTs), and random forests (RFs).
Geochemical data collected and stored by government and private sectors are becoming increasingly complex and too large for manual interpretation. The use of automatic methods for identification of potentially spurious data and estimation of missing values in very sparse geochemical datasets can significantly improve our understanding of geological systems. Deep neural networks have recently achieved remarkable success in a wide range of applied problems. These methods do not require manual feature engineering and significantly outperform traditional machine learning algorithms when applied to large datasets. We present a deep learning based method for estimation of unknown sample analytes in geochemical data. This approach is entirely data-driven and, once the network is trained, delivers the results in real time by predicting the distribution of an unknown analyte in a single step. A case study on the Western Australian Geochemistry (WACHEM) database demonstrates the efficiency of the method. Base metals as well as many other metals show good predictive capability (average symmetric mean absolute percentage errors range from 20% to 26.1%). Silver, platinum and especially gold are found to be more difficult to predict. The estimation of rock types from geochemical data allows for validation of existing datasets and prediction of rock types directly from geochemical data when there is no other existing information. The results of this study can benefit mineral explorers by indicating exploration targets and highlighting gaps in existing data.
Study on reservoir characteristics and evaluation methods of altered igneous reservoirs in Songliao Basin, China
2022, Journal of Petroleum Science and Engineering
To accurately evaluate altered volcanic reservoirs in the Songliao Basin of China and study the effect of alteration on pore structure, this paper established a methodology for log evaluation based on conventional log data. The volcanic reservoir in the Songliao Basin, China is variable in lithology, complex in pore structure and widespread in alteration phenomena. Because of these problems, the conventional charting method is not good at lithological identification, and the reservoir interpretation conclusions based on interpretation of the conventional logging method are not consistent with the actual logging data, making it difficult to carry out accurate logging evaluation. To solve such problems, this study adopted the particle swarm optimization-support vector machine (PSO-SVM) for carrying out lithological identification. Then, alteration factors were established to evaluate the alteration degree of volcanic reservoirs and obtain the altered clay content by fitting to correct the effect of alteration. Designing the volume model that considered altered clay and selecting the Simandoux model that could correct the influence of clay to calculate relevant petrophysical parameters such as porosity and water saturation, thus formed a methodology of logging evaluation suitable for the altered volcanic reservoir in the study area. At the same time, the study found that for different volcanic rocks, the effects of alteration on pore structure are different. An analysis, in combination with alteration factors and T₂ NMR spectra, showed that for the volcanic reservoir of nonsevere alteration, the higher the alteration degree, the greater the total porosity and the void porosity. However, with increasing alteration degree, the effective porosity of andesite was higher, while that of tuff worsened. The findings may provide a new idea for the analysis of the pore structure of altered volcanic reservoirs.
Carajás Mineral Province - Example of metallogeny of a rift above a cratonic lithospheric keel
2021, Journal of South American Earth Sciences
The Carajás Mineral Province (CMP) is located on the southeastern margin of the Amazonian Craton, in northern Brazil. It is the largest producer of high-grade iron ore and a major world supplier of copper, nickel, manganese and gold. This article presents the CMP as a product of oblique rift evolution (2.76–2.06 Ga) controlled by the Canaã and Cinzento strike slip system. The Carajás rift is hosted in the tonalite-trondhjemite-granodiorite (TTG)-greenstone Rio Maria terrains (3.0–2.85 Ga). It overlies a subcratonic lithospheric mantle (SCLM) keel as indicated by seismic tomography and is filled by the Carajás plutonic-volcanic sedimentary sequences (CPVSS) that host the CMP. The rifting stage (Grão Pará Group) began with intense tholeiitic fissure volcanism with basalt and rhyolite flows, coeval intrusions of alkali granites and mafic-ultramafic bodies, BIFs, shales and restricted tuff beds. The Azul and Águas Claras Formations consist mainly of shales and sandstones deposited during the sag stage. The calc-alkaline signature of the Grão Pará metavolcanic rocks was demonstrated to reflect the composition of the melted TTG basement. The high magnesium and platinum-group elements (PGE) contents of Puma and Onça mafic-ultramafic complexes indicate extensive partial fusion, conditions that are indicative of a plume origin. We are proposing that CMP's IOCG deposits were formed from saline carbonate fluids exsolved from deep-seated metasomatic mantellic magmas (MMM), channelled and concentrated along jogs and step-overs in strike-slip fault systems in the Carajás rift.
The CMP comprises three main ore-forming stages: (1) a plume stage, ca. 2.76–2.73 Ga, creating supergiant exhalative sedimentary biogenic Fe deposits (Serra Norte, Serra sul and Serra Leste), syngenetic exhalative Cu–Au–Zn (Pojuca deposit), exhalative Mn (Antonio Vicente), Ni-PGE-reef deposits in a magma mixing zone of layered intrusions (e.g. Luanga deposit), and Siqueirinho IOCG deposit; (2) a metasome stage including crustal extension and mantle flash melting (60–100 km depth), ca. 2.68–2.50 Ga, forming the most important IOCG deposits like Salobo, Alemão and Igarapé Bahia. and (3) a sag stage, ca. 2.68–2.06 Ga, with sedimentary Mn deposited in a redox-controlled stratified oceanic environment. The deposits related to granitic intrusion, ca. 1.88 Ga, with Cu-(Au–Mo–W–Sn) greisen (Breves deposit) and Cu-(Au–Mo–F) veins (Gameleira and Alvo 118 deposits) are associated with the A-type magmatism that affected the entire central Amazonian Craton. These deposits are, in contrast, related to an intracratonic silicic large igneous province (SLIP) and have no relation to the Archean Carajás cratonic lithosphere keel rift evolution. The CMP is directly related to the thinning of the lithosphere above a mantle plume, beneath a sub-continental lithospheric mantle (SCLM) keel.
Restoring original composition of hydrothermally altered Archean metavolcanic rocks of the Carajás Mineral Province (Brazil): Geodynamic implications for the transition from lid to mobile tectonics
2020, Lithos
Citation Excerpt :
Such a conventional approach has yet to be challenged by new major- and trace-element systematics developed to specifically discriminate Archean volcanic sites. Meanwhile, the ANN-based method of Trépanier et al. (2016) should be developed further in order to propose a full range of model results integrating minor- and trace elements for both predicted precursors and quantified hydrothermal alterations. Precursor contents in high field strength elements (HFSE) and rare earth elements (REE) could then be used to better characterize mantle sources and improve petrogenetic models for Archean metabasalts.
One of the most debated issues about the Carajás Mineral Province in Brazil is the tectonic setting that prevailed at the time of the major Neoarchean (ca. 2.7 Ga) volcanic event, with views ranging from subduction- to rift-related. During this event, thick lava flows of mainly basaltic and basaltic-andesitic composition extruded at the Earth's surface, cooled and underwent significant alteration by hydrothermal fluids rising through deep-seated crustal faults. Archean settings have been inferred for hydrothermally altered volcanic rocks using as a proxy their whole-rock compositions in discrimination diagrams for modern geotectonics. Despite its debatable application to Archean volcanic sites, the same working approach is used in reconsidering the geodynamic view about the Carajás volcanic event. We use an artificial neural networks-based method to perform a full mass balance calculation on a large compiled database of metavolcanic rocks from Carajás. Predicted precursors reproduce a linear and limited FeO*/MgO increase with SiO₂ typical of modern calc-alkaline arc environment. Moreover, the identified least-altered rocks show that they may petrogenetically involve subducted water-bearing sediments. But contrary to what is expected for such a volcanic arc environment, amphibole and magnetite fractionations have not been observed in the precursors. Discrimination diagrams using incompatible trace elements were unable to rule out the subduction hypothesis so other geological and geochemical evidence from the granitoid record of Carajás was taken into consideration. In combination with the Archean granitoid evolution of CMP, we conclude that crustal contamination associated with continental rifting appears to be the most likely process for inducing the calc-alkaline affinity of the Carajás Neoarchean metavolcanic rocks. This calc-alkaline affinity in least-altered volcanic rocks appears to be best reproduced by petrogenetic modelling involving wall-rock assimilation and melting of the Carajás basement (TTG and other K-rich granitoids) by parental magmas derived from depleted mantle source, prior to their fractional crystallization. This volcanic event was coeval with the intrusion of A-type syntectonic granitoids at the transition from a stagnant lid to mobile plate tectonic regime associated with localized translithospheric transpressional rift.
Use of semi-volatile metals as a new vectoring tool for VMS exploration: Example from the Zn-rich McLeod deposit, Abitibi, Canada
2019, Journal of Geochemical Exploration
Citation Excerpt :
Trace elements include metals, metalloids, and incompatibles, such as high field strength elements (HFSE) and the large ion lithophile elements (LILE). The HFSE are commonly immobile and mostly used for quantifying the degree of mobility of lithophile elements (MacLean and Kranidiotis, 1987; Rubin et al., 1993; Trépanier et al., 2016). Semi-volatile metals are a suite of trace elements characterized by a low-melting-point and enhanced mobility in hydrothermal systems (fluid or vapor).
Volcanogenic massive sulfide (VMS) deposits are small targets (100's of meters) hosted in a hectometric to kilometric-scale zoned alteration halo, comprising narrow proximal chlorite and wider distal sericite zones. For exploration, identification of the distal fine-grained sericite halo related to fertile VMS systems is critical but remains challenging. Sericitization is a low-temperature hydrothermal alteration and its geochemical signature could be very similar to clay assemblages produced as the results of simple low-temperature prolonged seawater interaction with volcanic rocks on the seafloor. Currently, classical lithogeochemical methods (e.g., mass balance, element ratios) are not very efficient to decipher sericite alteration related to distal fertile VMS systems. However, semi-volatile metals have good potential as pathfinders, as they have a wider dispersion than major, or trace elements commonly used due to their enhanced mobility in hydrothermal fluids. New analytical improvements, involving four-acid digestion and advanced ICP-MS technology, now allows the accurate determination of semi-volatile metals at the ppb level.
Using the Zn-rich McLeod deposit in the Matagami district (Abitibi, Canada) as a case study, we tested first the representativeness and advantage of the new analytical method, and secondly, the capacity of the semi-volatile metals for vectoring exploration. The new method is both accurate and precise and provided good resolution with low detection limits. The performance of classical lithogeochemical methods was compared with semi-volatile metal distribution (As, Se, Cd, In, Sn, Sb, Te, Hg, Tl, Bi) in the footwall rhyolite. Although the classical tools efficiently identified both the proximal chlorite and distal sericite zones, most of them (except Rb/Sr) cannot efficiently vector, as there is no systematic increase or decrease towards the mineralization. Conversely, overlapping alteration halos were identified using semi-volatile metals. The proximal signature (chlorite zone) is characterized by enrichment of Sb, Li ± Mo, W and U. The distal signature (sericite zone) is defined by successive appearance of higher values of Sn, W, and Tl up to 750 m, 1000 m and 1400 m away from the mineralization, respectively. This zoned spatial distribution of semi-volatile metals clearly allows better localization of a given altered sample within a wide and distal alteration halo, thus providing a solid vectoring tool for VMS exploration. This is especially true since the signal of semi-volatile metals extends far beyond both the visual sericite alteration (up to 200 m from mineralization) and classical lithogeochemical tools (up to 1000 m away). Specifically, the Tl/Co ratio records the widest footprint of the mineralization, up to 1400 m away from the deposit. It is well known that Tl is enriched in Zn-rich VMS. Thallium in the sericite alteration halo is here considered as an indicator for discriminating fertile volcanogenic hydrothermal systems from barren low-temperature convective cells.

View all citing articles on Scopus

View full text

Research paperPrecursors predicted by artificial neural networks for mass balance calculations: Quantifying hydrothermal alteration in volcanic rocks

Highlights

Abstract

Introduction

Section snippets

Description of the method

Least-altered volcanic rocks dataset

Software

Neural network validations

Method validation using case studies

Discussion and conclusions

Computer code

Acknowledgements

Geochim. Cosmochim. Acta

Earth Planet. Sci. Lett.

Phys. Chem. Earth Parts A/B/C

Phys. Chem. Earth

J. Volcanol. Geotherm. Res.

Earth Planet. Sci. Lett.

Chem. Geol.

Lithosphere

Earth Planet. Sci. Lett.

Chem. Geol.

Chem. Geol.

Earth Planet. Sci. Lett.

Chemostratigraphy and hydrothermal alteration in exploration for VHMS deposits in greenstones and younger volcanic rocks

Transition from ultrapotassic kamafugitic to sub-alkaline magmas: Sr, Nd and Pb isotope, trace element and 40Ar–39Ar Age data from the Middle Latin Valley Volcanic Field, Roman Magmatic Province, Central Italy

J. Petrol.

Mineralogy and Geochemistry of Hydrothermal Alteration at the Ann- Mason Porphyry Copper Deposit, Nevada: Comparison of Large-scale Ore Exploration Techniques to Mineral Chemistry (Unpublished Master Thesis)

Footprints of Porphyry Cu Deposits: Vectors to the Hydrothermal Center Using Mineral Mapping and Lithogeochemistry

The behaviour of so-called immobile elements in hydrothermally altered rocks associated with volcanogenic submarine exhalative ore deposits

Miner. Deposita

The Isocon Diagram – a simple solution to gresens’s rquation for metasomatic alteration

Econ. Geol.

Composition–volume relationships of metasomatism

Chem. Geol.

Classification of altered volcanic island arc rocks using immobile trace elements: development of the Th–Co discrimination diagram

J. Petrol.

Yttrium: the immobility–mobility transition during basaltic weathering

Geology

Multilayer feedforward networks are universal approximators

Neural Netw.

Research paper
Precursors predicted by artificial neural networks for mass balance calculations: Quantifying hydrothermal alteration in volcanic rocks

Transition from ultrapotassic kamafugitic to sub-alkaline magmas: Sr, Nd and Pb isotope, trace element and ⁴⁰Ar–³⁹Ar Age data from the Middle Latin Valley Volcanic Field, Roman Magmatic Province, Central Italy