Multivariate Geostatistical Prediction of Geochemical Measurements for Use in Mineral Prospectivity ModelingReport as inadecuate




Multivariate Geostatistical Prediction of Geochemical Measurements for Use in Mineral Prospectivity Modeling - Download this document for free, or read online. Document in PDF available to download.

Geostatistics, Multivariate Simulation, Mineral Exploration, Mineral Prospectivity Modeling

Black, Warren E

Supervisor and department: Deutsch, Clayton Civil and Environmental Engineering

Examining committee member and department: Hall, Robert Civil and Environmental Engineering Boisvert, Jeff Civil and Environmental Engineering Deutsch, Clayton V Civil and Environmental Engineering

Department: Department of Civil and Environmental Engineering

Specialization: Mining Engeering

Date accepted: 2016-09-29T10:07:32Z

Graduation date: 2016-06:Fall 2016

Degree: Master of Science

Degree level: Master's

Abstract: Traditional approaches to mineral exploration rely on personal experience, conceptual genetic models, past exploration data, and geological characteristics found in analogous target deposit types to locate and evaluate prospective areas. With the increase in both size and complexity of datasets used in mineral exploration, mineral prospectivity modeling MPM provides a means of exploring highly dimensional geological datasets in a meaningful way. When exploring for a specific deposit type, prior knowledge from known mineral deposits within or near the study area and genetic characteristics of the deposit type are used to understand exploration factors that indicate the presence of a mineral deposit i.e., positive information. A concern is that barren locations i.e., negative information are rarely recorded for widespread use by others, yet they are as important as positive locations in training predictive models. It is likely that prospective areas are not being discovered as current methodologies are heuristic in nature and do not consider the full spectrum of the truth. A proposed novel MPM framework provides a means of passing a stochastic multi-element model and other relevant geological data to a transfer function that calculates the probability that a particular mineral deposit type exists at each location. The use of a multi-element geochemical model allows both positive and negative information to be equally represented while avoiding heuristic searches by not using known mineral occurrences as input. In addition, the multiple realizations of the geochemical model permits uncertainty to be transferred to the final probabilistic values at each location. The principle challenge within the proposed framework is the prediction of the required stochastic geochemical model. It is desired to have a flexible multi-element geochemical model that may be used to perform MPM for many deposit types. In hopes of providing a straightforward multivariate simulation framework, novel extensions of the decorrelation and direct cosimulation frameworks that operate in the presence of many secondary data are developed; however, they fail to adequately reproduce input multivariate statistics. The introduction of correlation to the once uncorrelated factors during simulation by the conditioning of secondary data renders the decorrelation framework inadequate. While the modification of direct cosimulation is easy to implement, it is hampered by extreme variance inflation and an inability to reproduce the input correlation structure. As the capabilities of the cokriging and hierarchical framework to model highly dimensional problems had not been demonstrated, they were both implemented in an attempt to predict 42 variables. A linear model of coregionalization consisting of 903 direct and cross-variograms was fitted to the data, however, it did not adequately capture the spatial structure of the input variables. The framework also proved to be very computational expensive. Conversely, the hierarchical framework reasonably reproduces input univariate and collocated multivariate statistics and provides a viable option for simulating multivariate data with many exhaustive secondary data. The proposed MPM framework is demonstrated in a small example workflow that when passed the geochemical model produced by the hierarchical framework and other relevant geological data, predicted three locations with high probability of deposit discovery. All three locations were in very close proximity i.e., within 50-2700 meters to either a showing, drilled prospect, or past-producer, which is a promising sign but requires additional research.

Language: English

DOI: doi:10.7939-R37659M2R

Rights: This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for the purpose of private, scholarly or scientific research. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.





Author: Black, Warren E

Source: https://era.library.ualberta.ca/


Teaser



Multivariate Geostatistical Prediction of Geochemical Measurements for Use in Probabilistic Mineral Prospectivity Modeling by Warren Edward Black A thesis submi ed in partial fulfillment of the requirements for the degree of Master of Science in Mining Engineering Department of Civil and Environmental Engineering University of Alberta © Warren Edward Black, 2016 A Traditional approaches to mineral exploration rely on personal experience, conceptual genetic models, past exploration data, and geological characteristics found in analogous target deposit types to locate and evaluate prospective areas.
With the increase in both size and complexity of datasets used in mineral exploration, mineral prospectivity modeling (MPM) provides a means of exploring highly dimensional geological datasets in a meaningful way.
When exploring for a specific deposit type, prior knowledge from known mineral deposits within or near the study area and genetic characteristics of the deposit type are used to understand exploration factors that indicate the presence of a mineral deposit (i.e., positive information).
A concern is that barren locations (i.e., negative information) are rarely recorded for widespread use by others, yet they are as important as positive locations in training predictive models.
It is likely that prospective areas are not being discovered as current methodologies are heuristic in nature and do not consider the full spectrum of the truth. A proposed novel MPM framework provides a means of passing a stochastic multi-element model and other relevant geological data to a transfer function that calculates the probability that a particular mineral deposit type exists at each location.
The use of a multi-element geochemical model allows both positive and negative information to be equally represented while avoiding heuristic searches by not using known mineral occurrences as input.
In addition, the multiple realizations of the geochemical model permits uncertainty ...





Related documents