Please use this identifier to cite or link to this item:
Type: Thesis
Title: Multivariate statistical analysis of geochemical data to constrain the evolution, mineralization and alteration signatures of IOCG and BiF-hosted deposits
Author: Dmitrijeva, Marija
Issue Date: 2019
School/Discipline: School of Chemical Engineering
Abstract: Iron-oxide copper gold (IOCG) and banded iron formation (BIF)-hosted iron deposits are the dominant styles of mineralization in the Gawler Craton, South Australia. Multivariate geochemical datasets of various types and complexity collected from whole-rock samples and individual mineral grains from these deposits form a previously untapped source of information that can be applied to characterization and genetic modelling of such deposits. Moreover, such datasets carry major implications for new approaches to regional-scale exploration. Despite this, the ever-increasing complexity and size of geochemical datasets necessitates development of bespoke multivariate statistical analysis techniques as a prerequisite to any reproducible quantitative characterization of mineralization and/or alteration signatures, especially when detailed petrographic data are either limited or absent altogether. Left-censored values, i.e., those concentrations that fall below minimum limit of detection (reflecting the analytical limitation of the instrument used for acquisition), are present in all types of geochemical datasets but are amplified in trace element datasets such as those obtained by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). A novel approach for imputation of values below detection limits using copula cumulative density functions (CDFs) is proposed. A case study involving imputation of left-censored data for Te in the presence of Au and Ag values is based upon fitting copula CDFs to the data and iteratively simulating new Te values from copula CDFs using the Metropolis-Hastings algorithm. Within BIF-associated iron ores of the Middleback Ranges, iron oxides display interconversion and replacement reactions between magnetite by hematite and reprecipitation of new generations of platy hematite. Quantitative assessment of such interconversions, when correlated with available textural evidence, allows definition of geochemical signatures associated with ore enrichment. Linear-mixed effects models support the statistically significant difference among prevailing textures. Such models can account for the high degree of correlation among the analyses within a single polished block and therefore respect the hierarchical structure of such data. The observed trace element signatures indicate the potential impact of hydrothermal fluids associated with the ~1.8 Myola Volcanics, ~1.6. Ga Hiltaba Suite granites and younger discordant mafic dykes on ore enrichment process in the Middleback Ranges. Statistical analyses of a large LA-ICP-MS dataset for pyrite shows the independence of several minor and trace elements of economic interest or of relevance to optimized ore processing. Principal Component Analysis shows that Au is contained in zoned pyrite, along with As-Co(- Ni), thus suggesting the presence of invisible Au as both nanoinclusions and within the sulphide lattice. Nevertheless, PC2 essentially discriminates Co-As-Au from Ni, thus suggesting that Co-As zoning in pyrite is not always accompanied by Ni. The clearly defined (by PC1) Ag-Bi-Pb association is likely present in the form of discrete inclusions of telluride minerals. Other elements certainly hosted as minerals inclusions in pyrite and not in the crystal lattice include Zn, Sn, Mn, Ti, Sb and Cu. Application of multivariate statistical analyses to whole-rock data from three different IOCG systems within the Olympic Dam district, coupled with geological 3D modelling, has resulted in recognition and specification of shared but subtly different mineralization and alteration signatures within each system. The novelty of such an approach lies in the direct comparison of lithological and mineralogical zonation with calculated parameters such as geochemical clusters and principal component scores, also taking into consideration an extended suite of elements representative of both protoliths and superimposed mineralization and alteration. Results showed that despite significant differences between the geological settings of individual IOCG systems (host lithologies, presence of mafic dykes, mineralization style etc.), all three (the Olympic Dam deposit and the Acropolis and Wirrda Well prospects) can be readily characterized by the presence of ‘granitophile’ signatures thus confirming a common source of fluids for IOCG mineralization across the Olympic Cu-Au Province. Multivariate statistical analysis is both flexible and adaptable and can readily complement and often outperform conventional approaches to interpreting different types of geochemical data. Although particularly applicable to larger, more complex datasets, this research also emphasizes the caution needed when working with datasets containing large proportions of left-censored values.
Advisor: Ciobanu, Cristiana
Ehrig, Kathy
Metcalfe, Andrew
Dissertation Note: Thesis (Ph.D.) -- University of Adelaide, School of Chemical Engineering & Advanced Materials, 2020
Keywords: IOCG
multivariate statistics
iron oxides
Provenance: This electronic version is made publicly available by the University of Adelaide in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material you wish to be removed from this electronic version, please complete the take down form located at:
Appears in Collections:Research Theses

Files in This Item:
File Description SizeFormat 
Dmitrijeva2019_PhD.pdf67.23 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.