Potential For Gold Mineralization Research Articles

Integrated Geophysical Investigation using Aero-radiometric and Electrical Methods for Potential Gold mineralization within Yauri/Zuru Schist Belts, Kebbi State NW Nigeria

This study used aero-radiometric, 2D electrical resistivity tomography (ERT), and induced polarization (IP) methods to delineate gold mineralization potential. The study also confirmed and followed-up on regions with major structural features identified in previous aeromagnetic studies in the area. A half-degree airborne radiometric data of sheet 118_Yelwa from the NGSA, which contains three radioelements (%K, eTh, and eU) were used in this study. These datasets were processed and analyzed with Oasis Montaj Grid Math expression builder to obtain the %K_ratio_eTh and Ternary grid anomalies. The results identified zones E and F as hydrothermally altered regions that may harbour gold mineralization. These findings were consistent with the regions of major structural features identified in previous aeromagnetic studies in the area. The zones were located in the eastern parts of Ngaski, Yauri (Yelwa), Shanga, and Agwara, as well as Magama's northwest region. However, zone F1 (the eastern portion of Ngaski/Yauri) has been further investigated using 2ERT and IP detailed geophysical methods in a dipole-dipole configuration. The results of geoelectric techniques along profiles 1, 2, and 3 identified the major gold mineralization potential zones, which were labeled A1, A, and C. These regions have low/high resistivity and chargeability signatures, and could thus be interpreted as potential target zones for metallic mineral exploration, particularly gold mineralization. The regions are located in the northern part of Mararraba and southwest of the Jinsani areas of Kebbi State.

Deep structure of the Wulong goldfield, Liaodong Peninsula, China, revealed by receiver functions: implications for the tectonic and mineralization dynamics

During the Mesozoic, the North China Craton experienced intense tectonic movements that resulted in the formation of numerous gold deposits on the Liaodong and Jiaodong Peninsulas in northeastern China. To investigate the relationship between deep crustal structure and gold mineralization in the Liaodong Peninsula, we deployed 334 dense seismic stations in the Wulong goldfield (WLGF) with the idea of analysing numerous receiver functions at different array stations. The purpose focused on knowing the potential for gold mineralization in the area. The study revealed the following: (1) The WLGF is characterized by a crustal thickness of approximately 32 km and an average Vp/Vs ratio of 1.76. The high value of the Vp/Vs ratio near the Wulong gold deposit suggests that mantle materials have penetrated into the crust and contributed to the mineralization process. (2) A low-velocity layer located at a depth of 10–18 km below the WLGF seems to support the existence of a potentially brittle-ductile transition zone. Also, hydrothermal magma upwelling channels are observed in the upper crust beneath the Wulong gold deposit. (3) The presence of a discontinuous low-velocity layer in the middle crust beneath the Liaodong Peninsula suggests promising prospects for gold ore exploration. The receiver functions method based on a dense seismic array employed in this study can offer valuable references and guidance for the fine exploration and research of ore deposits in other regions globally.

Mapping of Orogenic Gold Mineralization Potential in the Kushaka Schist Belt, Northcentral Nigeria: Insights from Point Pattern, Kernel Density, Staged-Factor, and Fuzzy AHP Modeling Techniques

Mapping of Orogenic Gold Mineralization Potential in the Kushaka Schist Belt, Northcentral Nigeria: Insights from Point Pattern, Kernel Density, Staged-Factor, and Fuzzy AHP Modeling Techniques

Integrated geophysical Investigation for gold mineralization potential over southern parts of Kebbi state and its Environs, northwestern Nigeria

The potential zones of gold mineralization were identified in this study using aeromagnetic and aero-radiometric methods. The Nigerian Geological Survey Agency (NGSA) provided half-degree airborne magnetic and radiometric datasets covering the southern part of Kebbi State. Magnetic data were subjected to first vertical derivative (1VD), total gradient amplitude (TGA), total horizontal derivative (THD), source edge detection (SED), center for exploration targeting (CET), Euler deconvolution (ED) and source parameter imaging (SPI) to identify favourable structures to gold mineralization. Aero-radiometric data delineation of the region of hydrothermal alteration zones through the K/eTh ratio, K_deviation, F_parameters, and Ternary image analyses were successful. The results of the magnetic data techniques revealed the regions of major structures/or lineaments with gold mineralization attributes, trending in the NE to SW directions and the SE to NE parts of the study area. The depth to the structures of the magnetic source hosting gold mineralization was less than 5 m using algorithm ED and SPI techniques. Normalized radiometric data showed the area of anomalously high and moderate hydrothermal altered zones. The region's designation as a gold field is supported by major fault lines observed on the 1VD, a sequence of bristle fractures from the CET analysis, and high values of K_deviation and F_parameter, all of which are were visible on the Ternary images. The integrated results revealed zones of major structures and hydrothermal regions of gold fields at Agwara, Western Magama, Rijau, Fakai, Bukkuyum, and Borgu in the SE of the study area.

Gold occurrence and mineralization in the Paleozoic Heijianshan Fe–Cu (–Au) deposit, Eastern Tianshan

Gold (e.g., electrum) has been reported in the Paleozoic Heijianshan Fe–Cu (–Au) or iron oxide–copper–gold-like (IOCG-like) deposit in the Eastern Tianshan, but relationships between gold occurrence and sulfides are unclear. The Heijianshan Fe–Cu (–Au) (or IOCG-like) deposit underwent five alteration and mineralization stages: Pre-ore epidote alteration (Stage I), syn-ore magnetite/iron mineralization (Stage II) followed by pyrite alteration (Stage III) and Cu (–Au) mineralization (Stage IV), and post-ore late veins (Stage V). This deposit mainly has two types of pyrite (A and B) based on pyrite–hematite and pyrite–pyrrhotite–chalcopyrite assemblages. They are euhedral to subhedral and mainly homogenous, with the former having porous or inclusion-rich domains. Trace elements incorporating into pyrite at Heijianshan are mainly controlled by compositions (e.g., As, Co, and Cu), temperature (e.g., Ni and Se), and oxygen fugacity (fO2; e.g., As and Se) of hydrothermal fluids, respectively. The time-resolved depth-concentration profiles suggest trace elements in pyrite and chalcopyrite occur as solid solution (e.g., Co, Ni, Zn, As, and Se) and micro- to nano-sized mineral inclusions (e.g., Cu, Bi, and Pb). The gold occurrence at Heijianshan is invisible as solid solution in Stage III pyrite and Stage IV chalcopyrite and visible as micron-sized Stage IV electrum. Moreover, incorporation mechanisms of invisible gold are Au3+ substituting iron or copper in pyrite A and chalcopyrite, respectively, and Au+ incorporating into the vacancy or defect positions of pyrite B. Sulfides texture and geochemical compositions when combined with previous in-situ sulfur isotope suggest that pyrite formation was resulted from interaction between external basinal brines and altered host rocks under different temperature and fO2 conditions. Ongoing fluid-rock reaction led to formation of chalcopyrite ± electrum veins as Cu (–Au) mineralization at Heijianshan. This study suggests that the prolonged interaction contributes to sulfide formation and Cu (–Au) mineralization in these Paleozoic Fe–Cu or IOCG-like deposits formed in basin-related setting, and the Aqishan-Yamansu belt, Eastern Tianshan has gold mineralization potential.

Metallization characteristics and gold ore controlling factors in the south region of Tra Nu - Phuoc Thanh, Quang Nam Province

The gold mineralization in the southern region of Tra Nu-Phuoc Thanh was initially discovered and assessed by the Central Geological Division. However, mineralization characteristics and ore-controlling factors in the Tra Nu-Phuoc Thanh area remain inadequately understood. To assess the gold mineralization potential in this research area, this study conducts a comprehensive synthesis of prior research findings, complemented by the analysis of an additional 50 thin sections, 100 thick sections, SEM scanning electron microscopy, and 10 ICP-MS samples. The research results reveal the presence of gold deposits within the study area, including native gold and electrum. The identified gold-bearing minerals in the deposits encompass quartz, pyrite, galena, sphalerite, pyrrhotite, and goethite. The ore formations predominantly exhibit disseminated, vein, and breccia structures, along with granular, semi-granular, spherical, and solid inclusions. It has been ascertained that the gold ore in the study area belongs to the gold-quartz-sulfur formation and can be categorized into two types: gold-quartz-pyrite ore and polymetallic gold-quartz-sulfur ore. The ore is primarily situated within ancient metamorphic sedimentary rocks of the Kham Duc formation and young granitoid rocks of the Ba Na complex. Typically, the gold ore body occurs in a single-vessel form, with less common branched, vascular zone, and infiltrative vascular forms. All gold ore bodies are aligned along the fault systems, primarily oriented in the northeast-southwest direction. These sub-meridian fault systems serve as conduits for hydrothermal solutions carrying gold ore in the study area.

DETEKSI ZONA MINERALISASI EMAS DI GUNUNG GUPIT, MAGELANG MENGGUNAKAN METODE INDUCED POLARIZATION

Mount Gupit is an area with gold mineralization potential which is part of the Menoreh Mountains, in the Magelang area. Mineralization potential mapping was carried out using geophysics, namely the Induced Polarization method. Induced Polarization Method measurements were carried out in 5 trajectories in a relatively west to east direction and with varying path lengths. The configuration used is a dipole-dipole configuration with a: 10 meters and n: 1-6. Interpretation is carried out on the results of the Resistivity and Chargeability data inversion. The inversion results show a resistivity range from 2 to 400 Ohm-m. Resistivity values are divided into three criteria, namely low with a value of less than 10 Ohm-m as clay rock, medium with a range of (10-75) Ohm-m as pyroclastic breccia and high with a value greater than 75 Ohm-m as intrusion or chunks of andesite. The inversion results show that the chargeability value varies from 2 to 80 ms. This value is divided into three criteria, namely low (<10 ms), medium (10 to 35) ms and high (>35 ms). Low chargeability values are interpreted as zones that are not mineralized so that there are no metal minerals. The mineralized zone is being interpreted from a moderate chargeability value. Meanwhile, a high chargeability value is interpreted as a zone that is strongly mineralized and contains significant metallic minerals. The existence of a mineralized zone stretches from track 2 to track 5 to the north, while the zone has disappeared or been exhausted in track 1 to the south of the research area.

A New Vector for Mapping Gold Mineralization Potential and Proposed Pathways in Highly Weathered Basement Rocks using Multispectral, Radar, and Magnetic Data in Random Forest Algorithm

A New Vector for Mapping Gold Mineralization Potential and Proposed Pathways in Highly Weathered Basement Rocks using Multispectral, Radar, and Magnetic Data in Random Forest Algorithm

The Current State of Research on Secondary Quartzites of the Northern Segment of the Jungar-Balkhash Folded System and Their Au Mineralization (Central Kazakhstan)

In this paper, we outline a study conducted on altered rocks in Central Kazakhstan, in particular, focusing on secondary quartzites. The study reveals a connection between copper-porphyry deposits in Central Kazakhstan and secondary quartzites. Recent research has unveiled gold mineralization in various secondary quartzite massifs, which has spurred further investigations in the region. Over the past three decades, extensive studies have been conducted that have demonstrated similarities among the gold deposits in this area and epithermal gold-silver deposits in volcanic-plutonic arcs around the world. By identifying patterns in the distribution of gold mineralization and developing regional exploration criteria, researchers have comprehensively assessed 48 secondary quartzite massifs and have prioritized promising areas for further exploration. Valuable findings have emerged from detailed examinations of the Akgirek and Birlestik massifs, which have involved drilling exploration wells up to a depth of 250 m. These investigations have shed light on the depth distribution of metasomatites and their association with gold mineralization. Consequently, these findings strengthen our belief in the significant potential for gold mineralization in the secondary quartzites of the Jungar-Balkhash folded system.

Identification of alteration zone and gold mineralization based on magnetic anomaly and 3D model of geomagnetic satellite data inversion of Mount Pongkor Area, West Java

Mount Pongkor is one of the areas in Indonesia with the largest gold mineralization potential in Java. One of the geophysical methods to identify the distribution of gold mineralization zones is the geomagnetic method that utilizes magnetic properties in the subsurface due to the influence of rock magnetization. Geomagnetic research has been conducted at Mount Pongkor, Bogor Regency, West Java with an area of 22 x 17 km as much as 793 magnetic satellite data from the National Oceanic and Atmospheric Administration (NOAA) website that has been corrected daily. This study aims to determine the distribution of alteration zones and mineralization of the study area. The results showed that the RTP map shows the distribution of magnetic anomalies ranging from -4.786 – 4.663 nT, with high anomalies in the north-south direction associated with mineralization zones with anomaly values ranging from 1.881 – 4.663 nT and low anomalies in the north-south direction associated with rock alteration zones with anomaly values ranging from -4,786 - (-2.174) nT. In the 3D inversion model, the alteration zone has an average depth of 350 - 2600 m from topography with susceptibility contrast values ranging from -0.35 - (- 0.25) SI and the mineralization zone has an average depth of 350 - 2800 m from topography with susceptibility contrast values of 0.25 - 0.35 SI.©2022 JNSMR UIN Walisongo. All rights reserved.

Building a Random Forest predictive modeling of mineral perspectivity and Mapping gold mineral prospects in Tam Ky - Phuoc Son, Quang Nam

Tam Ky - Phuoc Son area has great potential for gold mineral with 98 gold occurrences, but the evaluation of the entire gold-mineralization potential of the area is still very limited, while this is considered as a basis for planning, exploration, and mining. The paper uses an Artificial Intelligence model which has a name Random Forest to build predictive modeling of mineral perspectivity and to map the gold mineral prospect of the study area. 12 influencing factors are selected to build the dataset for model training and mapping gold minerals prospect, including Geology, fault systems (NE-SW faults, NW-SE faults, sub meridian faults, sub-latitude faults), Bouguer geophysical anomaly, a geochemical anomaly of silver (Ag), gold ( Au), lead (Pb), zinc (Zn), copper (Cu) and distance to the geologic boundary of complexes related to gold mineralization. The data which are generated from these factors are 12 fuzzy maps. This data combines with 98 occurrences’ locations to create a dataset that is used to train a model of mineral perspectivity using the Random Forest algorithm. After training the model is evaluated by validation. The results of the Random Forest predictive modeling of mineral prospects are well trained with an accuracy of 95.99% on the training set and 83.05 on the validation set, the performance of the model is excellent on both datasets with AUC of 0.993 and 0.95, respectively. Finally, a mineral perspectivity map is built using the trained model. The study area is divided into 3 types of areas: high, medium, and low prospects. The area of high prospect is 982.8 km2, covering 71% of the gold occurrences.

Geophysical assessment for gold mineralization potential over the southern part of Kebbi State using aeromagnetic data

The magnetic signatures over the southern part of Kebbi State and its environs were analyzed together with the geological settings of the area to delineate the structures that may host gold mineralization. The aeromagnetic data used was the survey carried out by Fugro airborne surveys between 2005 and 2010 on behalf of the Federal Government of Nigeria. The reduction to equator (RTE), first and second vertical derivatives (FVD and SVD), Centre for Exploration Targeting (CET), analytic signal (AS), source parameter imaging (SPI) and tilt derivative (TDR) techniques were applied to the magnetic data covering the area. The results of the AS technique revealed that the study area is characterized with high amplitudes of magnetic anomalies (above 0.048 nT/m) and these could be of ferromagnetic minerals such as gold. The FVD, SVD, CET and TDR techniques also helped in delineating the lineaments (such as faults, fractures or shears zones) believed to be associated with alteration zones which play an important role in determining gold mineralized zones. The direction of the orientation of these features/lineaments trended in the NE-SW direction. The faults, fractures or shears zones delineated represent veins of possible mineralization. The depth of occurrence to the causative bodies using SPI algorithms was found to be below 137 m. Structures delineated within the area, when compared with the geological setting of the area, correspond to quartz-mica schist, granite, biotite, gneiss, diorite, medium coarse-grained and biotite hornblende granite. Results from these techniques revealed alteration zones that may host gold. These regions correspond to the following areas: SE parts of Yauri and Shanga, Fakai, Ngaski, Zuru, Magama, Rijau, and the eastern part of Wasagu/Danko and Bukkuyum.

Application of electrical resistivity tomography for mapping gold mineralization potential in Iperindo, Ilesha schist belt, southwestern Nigeria

This study applied Electrical Resistivity Tomography geophysical investigation technique to evaluate the gold mineralization potential of Iperindo in Ilesha Schist Belt, southwestern Nigeria, where commercial exploitation capable of generating revenue and employment for the inhabitants has been challenged by lack / inadequate subsurface geological/geophysical information. The filtered and inverted electrical resistivity data acquired through five (5) 336 m long E – W trending profiles, established 10 m apart from each other, delineate isolated near surface but thick (> 30 m) low resistivity zones, especially at the eastern and western ends of the study area. Some of the delineated low resistivity zones (3 – 200 𝛀m) present vertical sharp edges, likely created by vertical faults that flank the zones on both sides. The low resistivity of these zones could be attributed to the occurrence of conductive material such as gold and associated base metals which probably exist in pegmatitic veins within the zones.  Â

Ore fluid, geochronology and tectonic setting of mesothermal gold metallogeny in southeastern Jilin Province, Northeast China: A case study of the Shajingou gold deposit

The Shajingou gold deposit is a recently discovered medium-scale gold deposit in southeastern Jilin Province, Northeast China. The deposit is located in the easternmost segment of the northern margin of the North China Craton (NCC). This study investigated the mineralization age and tectonic setting of the Shajingou gold deposit based on geology, microthermometry and H–O isotopic analyses of fluid inclusions (FIs), pyrite Re–Os and zircon U–Pb dating, petro-geochemistry and zircon Hf isotopic analyses. The findings are placed in the wider context of mesothermal gold metallogeny in Northeast China.Metallogenesis of the gold deposit can be divided into three stages based on the ore mineral assemblage: quartz–K-feldspar–pyrite (stage I), quartz–polymetallic sulfide–native gold (stage II) and quartz–carbonate (stage III). Four types of primary FI were identified in quartz and calcite minerals: liquid-rich aqueous (L-type), CO2-bearing (C1-type), CO2-rich (C2-type) and CO2-pure (C3-type). From stage I to stage III, the fluid inclusion suite changes from L-type and C2-type in stage I, all four types in stage II and only L-type in stage III. Homogenization temperatures of stages I to III are 318–394 °C, 264–356 °C and 202–276 °C, respectively. Salinities of stages I–III are 7.31–12.42, 2.02–8.94 and 0.53–3.53 wt% NaCl equiv., respectively. The ore-forming fluid system evolved from a H2O–NaCl–CO2 system in stages I and II to a H2O–NaCl system in stage III. During the stage II, the addition of meteoric water led to decreases in temperature and pressure, resulting in an immiscible H2O–NaCl–CO2 fluid system. This immiscibility in turn led to the escape of CO2 and other volatiles, the increase of pH and the decrease in solubility and stability of Au–S clathrates, thus facilitating the deposition of gold and associated polymetallic sulfides.Laser ablation–inductively coupled plasma–mass spectrometry (LA–ICP–MS) zircon U–Pb dating of the Shajingou diorite yielded a weighted mean 206Pb/238U age of 171.7 ± 0.9 Ma (MSWD = 3.0). Re–Os dating of the gold-bearing pyrite yielded an isochron age of 166 ± 29 Ma (MSWD = 5.2). The Shajingou diorite has high concentrations of SiO2, Al2O3, Na2O, Mg#, Cr and Ni, weakly positive Eu anomalies, and is LREE- and LILE-enriched, HREE- and HFSE-depleted, as well as has positive εHf (t) values (+3.0 to +7.1). This indicates the diorite was sourced from the partial melting of a depleted-mantle wedge that had been modified by the subducted- slab-derived fluids. The deposit geology, ore genesis, geochronology and tectonic setting of the Shajingou gold deposit differ from those of other gold deposits in the eastern segment of the XMOB, but are similar to those of gold deposits in the Jiapigou–Haigou gold belt (JHGB) of southeastern Jilin Province at the northern margin of the NCC. These findings suggest the JHGB extends southeastward to the Helong area of Yanbian, which indicates the need for further research on the gold mineralization potential of the southeastern JHGB and further gold prospecting in the region. Comparative studies of Shajingou and other gold deposits in the JHGB have confirmed that a significant tectonic–magmatic–hydrothermal event occurred during the Middle Jurassic due to the subduction of the Paleo-Pacific Plate beneath the Eurasian Plate.

Multivariate Analysis of Stream Sediment Geochemical Data for Gold Exploration in Delijan, Iran

Muteh is a major Iranian gold mine that is located in Delijan, within in the Sanandaj-Sirjan metallogenic zone. Previously, the gold mineralization in Muteh was interpreted to be formed in Precambrian. However, recent geological models propose younger age, i.e., late Eocene, for the gold mineralization, therefore, the geologic units of age and older than late Eocene can be considered as targets for gold exploration. The objective of this paper is to investigate Muteh type gold mineralization potential in the area with respect to the new geological models. Statistical analysis of 644 stream sediment samples was used to determine anomalous concentrations Au and elemental correlations of Au geochemical pathfinders. Using the factor analysis, an eight-factor model was established for the stream sediment geochemical data explaining 74.76% of the total variance. Factor 4 was found to be representative of Al-Ca-Na-K-P, which defines the lithological variation in the data. Factor 6 represents the variation of Pb-Ag-Sb-Ba-Tl, which is related to the hydrothermal mineralization. The factor 7 includes Au-S-Mo and is representative of regional geochemical patterns of gold. The correlation of Au with S and Mo signifies the inclusions of Au in pyrite minerals of Muteh mine. Factor 7 specifies three exploration targets. Moreover, the fourth prospective area representing significant geochemical signatures for gold mineralization was not detected in factor analysis. The results show that data reduction techniques, such as factor analysis, have great potential for geochemical applications; however, it is essential to perform careful analysis in addition to using these techniques. Factor analysis highlights the regional geochemical patterns and may slight the smaller subsystems.

Targeting of Gold Deposits in Amazonian Exploration Frontiers using Knowledge- and Data-Driven Spatial Modeling of Geophysical, Geochemical, and Geological Data

This paper reports the application of weights-of-evidence, artificial neural networks, and fuzzy logic spatial modeling techniques to generate prospectivity maps for gold mineralization in the neighborhood of the Amapari Au mine, Brazil. The study area comprises one of the last Brazilian mineral exploration frontiers. The Amapari mine is located in the Maroni-Itaicaiunas Province, which regionally hosts important gold, iron, manganese, chromite, diamond, bauxite, kaolinite, and cassiterite deposits. The Amapari Au mine is characterized as of the orogenic gold deposit type. The highest gold grades are associated with highly deformed rocks and are concentrated in sulfide-rich veins mainly composed of pyrrhotite. The data used for the generation of gold prospectivity models include aerogeophysical and geological maps as well as the gold content of stream sediment samples. The prospectivity maps provided by these three methods showed that the Amapari mine stands out as an area of high potential for gold mineralization. The prospectivity maps also highlight new targets for gold exploration. These new targets were validated by means of detailed maps of gold geochemical anomalies in soil and by fieldwork. The identified target areas exhibit good spatial coincidence with the main soil geochemical anomalies and prospects, thus demonstrating that the delineation of exploration targets by analysis and integration of indirect datasets in a geographic information system (GIS) is consistent with direct prospecting. Considering that work of this nature has never been developed in the Amazonian region, this is an important example of the applicability and functionality of geophysical data and prospectivity analysis in regions where geologic and metallogenetic information is scarce.

MODELING THE SPATIAL DISTRIBUTION OF MINERAL DEPOSITS USING NEURAL NETWORKS

ABSTRACT. In the context of Geographic Information Systems (GIS), mineral potential mapping can be seen as a process whereby a set of input maps, each representing a distinct geoscientific variable, are combined to produce a single map which ranks areas according to their potential to host deposits of a particular type. However, unlike many spatial classification tasks, the objective is not to assign a pixel to some class, but to assign to each pixel a continuous value representing its mineralization potential. Logistic regression provides one method of modeling conditional probabilities on data in which the target values are binary, e.g., deposit present/deposit absent; however, logistic regression limits the output to be a logistic transformation of the inputs and hence may not be able to capture complex relationships among the input variables. This paper reports on the application of multilayer perceptrons (MLPs) to mapping reef gold mineralization potential in the Castlemaine region of Victoria, Australia. We show that the ability of MLPs to predict the presence of hold‐out test deposits is significantly better than that for logistic regression. We also show that a Bayesian learning approach can be used to automatically select an appropriate value for the weight regularization coefficient without the need of using a cross‐validation procedure. This means that all available data can be used for weight optimization, thus eliminating any sensitivity of the final map to the particular training/validation set partition used.This paper was presented at the 2004 Research Modeling Association World Conference on Natural Resource Modeling in Melbourne, Australia.

Mineral Potential Mapping Using Bayesian Learning for Multilayer Perceptrons

Multilayer perceptrons (MLPs) can be used to discover a function which can be used to map from a set of input variables onto a value representing the conditional probability of mineralization. The standard approach to training MLPs is gradient descent, in which the error between the network output and the target output is reduced in each iteration of the training algorithm. In order to prevent overfitting, a split-sample validation procedure is used, in which the data is partitioned into two sets: a training set, which is used for weight optimization, and a validation set, which is used to optimize various parameters that can be used to prevent overfitting. One of the problems with this approach is that the resulting maps can display significant variability which stems from (i) the (randomly initialized) starting weights and (ii) the particular training/validation set partition (also determined randomly). This problem is especially pertinent on mineral potential mapping tasks, in which the number of deposit cells is a very small proportion of the total number of cells in the study area. In contrast to gradient descent methods, Bayesian learning techniques do not find a single weight vector; rather, they infer the posterior distribution of the weights given the data. Predictions are then made by integrating over this distribution. An important advantage of the Bayesian approach is that the optimization of parameters such as the weight decay regularization coefficient can be performed using training data alone, thus avoiding the noise introduced through split-sample validation. This paper reports results of applying Bayesian learning techniques to the production of maps representing gold mineralization potential over the Castlemaine region of Victoria, Australia. Maps produced using the Bayesian approach display significantly less variability than those produced using gradient descent training. They are also more reliable at predicting the presence of unknown deposits.

Mapping Mineralization Probabilities using Multilayer Perceptrons

Mineral-potential mapping is the process of combining a set of input maps, each representing a distinct geo-scientific variable, to produce a single map which ranks areas according to their potential to host mineral deposits of a particular type. The maps are combined using a mapping function that must be either provided by an expert (knowledge-driven approach), or induced from sample data (data-driven approach). Current data-driven approaches using multilayer perceptrons (MLPs) to represent the mapping function have several inherent problems: they are highly sensitive to the selection of training data; they do not utilize the contextual information provided by nondeposit data; and there is no objective interpretation of the values output by the MLP. This paper presents a new approach by which MLPs can be trained to output values that can be interpreted strictly as representing posterior probabilities. Other advantages of the approach are that it utilizes all data in the construction of the model, and thus eliminates any dependence on a particular selection of training data. The technique is applied to mapping gold mineralization potential in the Castlemaine region of Victoria, Australia, and results are compared with a method based on estimating probability density functions.

Geologically Constrained Fuzzy Mapping of Gold Mineralization Potential, Baguio District, Philippines

An application of the theory of fuzzy sets to the mapping of gold mineralization potential in the Baguio gold mining district of the Philippines is described. Proximity to geological features is translated into fuzzy membership functions based upon qualitative and quantitative knowledge of spatial associations between known gold occurrences and geological features in the area. Fuzzy sets of favorable distances to geological features and favorable lithologic formations are combined using fuzzy logic as the inference engine. The data capture, map operations, and spatial data analyses are carried out using a geographic information system. The fuzzy predictive maps delineate at least 68% of the known gold occurrences that are used to generate the model. The fuzzy predictive maps delineate at least 76% of the “unknown” gold occurrences that are not used to generate the model. The results are highly comparable with the results of previous stream-sediment geochemical survey in the area. The results demonstrate the usefulness of a geologically constrained fuzzy set approach to map mineral potential and to redirect surficial exploration work in the search for yet undiscovered gold mineralization in the mining district. The method described is applicable to other mining districts elsewhere.