Incorporating conceptual and interpretation uncertainty to mineral prospectivity modelling

Prospectivity analyses are used to reduce the exploration search space for locating areas prospective for mineral deposits.The scale of a study and the type of mineral system associated with the deposit control the evidence layers used as proxies that represent critical ore genesis processes.In particular,knowledge-driven approaches(fuzzy logic)use a conceptual mineral systems model from which data proxies represent the critical components.These typically vary based on the scale of study and the type of mineral system being predicted.Prospectivity analyses utilising interpreted data to represent proxies for a mineral system model inherit the subjectivity of the interpretations and the uncertainties of the evidence layers used in the model.In the case study presented,the prospectivity for remobilised nickel sulphide(NiS)in the west Kimberley,Western Australia,is assessed with two novel techniques that objectively grade interpretations and accommodate alternative mineralisation scenarios.Exploration targets are then identified and supplied with a robustness assessment that reflects the variability of prospectivity value for each location when all models are considered.The first technique grades the strength of structural interpretations on an individual line-segment basis.Gradings are obtained from an objective measure of feature evidence,which is the quantification of specific patterns in geophysical data that are considered to reveal underlying structure.Individual structures are weighted in the prospectivity model with grading values correlated to their feature evidence.This technique allows interpreted features to contribute prospectivity proportional to their strength in feature evidence and indicates the level of associated stochastic uncertainty.The second technique aims to embrace the systemic uncertainty of modelling complex mineral systems.In this approach,multiple prospectivity maps are each generated with different combinations of confidence values applied to evidence layers to represent the diversity of processes potentially leading to ore deposition.With a suite of prospectivity maps,the most robust exploration targets are the locations with the highest prospectivity values showing the smallest range amongst the model suite.This new technique offers an approach that reveals to the modeller a range of alternative mineralisation scenarios while employing a sensible mineral systems model,robust modelling of prospectivity and significantly reducing the exploration search space for Ni.     

Assessing the impact of conceptual mineral systems uncertainty on prospectivity predictions

The past two decades have seen a rapid adoption of artificial intelligence methods applied to mineral exploration. More recently, the easier acquisition of some types of data has inspired a broad literature that has examined many machine learning and modelling techniques that combine exploration criteria, or ‘features’, to generate predictions for mineral prospectivity. Central to the design of prospectivity models is a ‘mineral system’, a conceptual model describing the key geological elements that control the timing and location of economic mineralisation. The mineral systems model defines what constitutes a training set, which features represent geological evidence of mineralisation, how features are engineered and what modelling methods are used. Mineral systems are knowledge-driven conceptual models, thus all parameter choices are subject to human biases and opinion so alternative models are possible. However, the effect of alternative mineral systems models on prospectivity is rarely compared despite the potential to heavily influence final predictions. In this study, we focus on the effect of conceptual uncertainty on Fe ore prospectivity models in the Hamersley region, Western Australia. Four important considerations are tested. (1) Five different supergene and hypogene conceptual mineral systems models guide the inputs for five forest-based classification prospectivity models model. (2) To represent conceptual uncertainty, the predictions are then combined for prospectivity model comparison. (3) Representation of three-dimensional objects as two-dimensional features are tested to address commonly ignored thickness of geological units. (4) The training dataset is composed of known economic mineralisation sites (deposits) as ‘positive’ examples, and exploration drilling data providing ‘negative’ sampling locations. Each of the spatial predictions are assessed using independent performance metrics common to AI-based classification methods and subjected to geological plausibility testing. We find that different conceptual mineral systems produce significantly different spatial predictions, thus conceptual uncertainty must be recognised. A benefit to recognising and modelling different conceptual models is that robust and geologically plausible predictions can be made that may guide mineral discovery.     

Data- and knowledge-driven mineral prospectivity maps for Canada's North

Data- and knowledge-driven techniques are used to produce regional Au prospectivity maps of a portion of Melville Peninsula, Northern Canada using geophysical and geochemical data. These basic datasets typically exist for large portions of Canada's North and are suitable for a “greenfields” exploration programme. The data-driven method involves the use of the Random Forest (RF) supervised classifier, a relatively new technique that has recently been applied to mineral potential modelling while the knowledge-driven technique makes use of weighted-index overlay, commonly used in GIS spatial modelling studies. We use the location of known Au occurrences to train the RF classifier and calculate the signature of Au occurrences as a group from non-occurrences using the basic geoscience dataset. The RF classification outperformed the knowledge-based model with respect to prediction of the known Au occurrences. The geochemical data in general were more predictive of the known Au occurrences than the geophysical data. A data-driven approach such as RF for the production of regional Au prospectivity maps is recommended provided that a sufficient number of training areas (known Au occurrences) exist.     

Regional prospectivity analysis for hydrothermal-remobilised nickel mineral systems in western Victoria,Australia

Fuzzy logic mineral prospectivity modelling was performed to identify camp-scale areas in western Victoria with an elevated potential for hydrothermal-remobilised nickel mineralisation. This prospectivity analysis was based on a conceptual mineral system model defined for a group of hydrothermal nickel deposits geologically similar to the Avebury deposit in Tasmania. The critical components of the conceptual model were translated into regional spatial predictor maps combined using a fuzzy inference system. Applying additional criteria of land use restrictions and depth of post-mineralisation cover, downgrading the exploration potential of the areas within national parks or with thick barren cover, allowed the identification of just a few potentially viable exploration targets, in the south of the Grampians-Stavely and Glenelg zones. Uncertainties of geological interpretations and parameters of the conceptual mineral system model were explicitly defined and propagated to the final prospectivity model by applying Monte Carlo simulations to the fuzzy inference system. Modelling uncertainty provides additional information which can assist in a further risk analysis for exploration decision making.     

Reducing subjectivity in multi-commodity mineral prospectivity analyses: Modelling the west Kimberley,Australia

Predicting realistic targets in underexplored regions proves a challenge for mineral explorers. Knowledge-driven prospectivity techniques assist in target prediction, and can significantly reduce the geographic search space to a few locations. The mineral prospectivity of the underexplored west Kimberley region was investigated following interpretation of regional gravity and magnetic data. Emphasis was placed on identifying geological structures that may have importance for the mineral prospectivity of the region. Subsurface structure was constrained through combined gravity and magnetic modelling along three transects. Crustal-scale structures were interpreted and investigated to determine their depth extent. These interpretations and models were linked to tectonic events and mineralization episodes in order to map the distribution of minerally prospective regions using a knowledge-driven mineral systems approach. A suite of evidence layers was created to represent geological components that led to mineralization, and then applied to each mineral system where appropriate. This approach was taken to provide a more objective basis for prospectivity modelling. The mineral systems considered were 1) magmatic Ni-sulphide, 2) carbonate-hosted base metals, 3) orogenic Au, 4) stratiform-hosted base metals and 5) intrusion-related base metals (including Sn–W, Fe-oxide–Cu–Au and Cu–Au porphyry deposits). These analyses suggest that a geologically complex belt in the Kimberley Basin at the boundary to the King Leopold Orogen is prospective for magmatic-related hydrothermal mineral systems (including Ni, Au and Cu). The Lennard Shelf is prospective for carbonate-hosted base metals around a feature known as the 67-mile high, and parts of the King Leopold Orogen are prospective for stratiform-hosted base metals. These results show that knowledge-driven mineral system modelling is effective in identifying prospectivity in regional-scale studies of underexplored areas, as well as drastically reducing the search space for explorers working in the west Kimberley.     

三维成矿定量预测系统设计与应用实例研究

隐伏矿体三维成矿定量预测是以多元地学大数据为基础的矿产预测新技术与方法。本文基于隐伏矿体三维成矿定量预测的实际需要及相关方法步骤,采用集成二次开发的方式,设计实现了一套可在三维环境下基于大数据开展定量化成矿预测工作的软件系统。本文阐述了系统的总体架构及开发方式,并对系统中各个功能模块的设计及实现过程进行详细阐述。系统融合了当前三维成矿定量预测研究的最新方法及成果,内含数据库管理、三维地球物理正演、三维空间分析以及三维预测评价等功能模块,能够对多元地学大数据进行集成和分析预测。为了验证系统的适用性和有效性,该系统被应用于长江中下游成矿带钟姑矿田三维成矿定量预测研究,相关成果表明系统的建立不但能够深化和发展三维成矿预测理论,也为新时期基于大数据的隐伏矿体找矿勘探工作提供了新方法及有力工具。     

Data fusion and porphyry copper prospectivity models,southeastern Arizona

Mineral exploration programs commonly use a combination of geological, geophysical and remotely sensed data to detect sets of optimal conditions for potential ore deposits. Prospectivity mapping techniques can integrate and analyse these digital geological data sets to produce maps that identify where optimal conditions converge. Three prospectivity mapping techniques – weights of evidence, fuzzy logic and a combination of these two methods – were applied to a 32,000 km² study area within the southeastern Arizona porphyry Cu district and then assessed based on their ability to identify new and existing areas of high mineral prospectivity. Validity testing revealed that the fuzzy logic method using membership values based on an exploration model identified known Cu deposits considerably better than those that relied solely on weights of evidence, and slightly better than those that used a combination of weights of evidence and fuzzy logic. This led to the selection of the prospectivity map created using the fuzzy logic method with membership values based on an exploration model. Three case study areas were identified that comprise many critical geological and geophysical characteristics favourable to hosting porphyry Cu mineralisation, but not associated with known mining or exploration activity. Detailed analysis of each case study has been performed to promote these areas as potential targets and to demonstrate the ability of prospectivity modelling techniques as useful tools in mineral exploration programs.     

From 2D to 3D: Prospectivity modelling in the Taupo Volcanic Zone,New Zealand

A 2D prospectivity model of epithermal gold mineralisation has been completed over the Taupo Volcanic Zone (TVZ), using the weights of evidence modelling technique. This study was used to restrict a 3D geological interpretation and prospectivity model for the Ohakuri region. The TVZ is commonly thought of as a present-day analogue of the environment in which many epithermal ore deposits, such as in the Hauraki Goldfield, Coromandel Volcanic Zone, are formed. The models utilise compiled digital data including historical exploration data, geological data from the Institute of Geological and Nuclear Sciences Ltd. Quarter Million Mapping Programme, recent Glass Earth geophysics data and historic exploration geochemical data, including rock-chip and stream sediment information. Spatial correlations between known deposits and predictive maps are determined from the available data, which represent each component of the currently accepted mineral system model for epithermal gold. The 2D prospectivity model confirms that the TVZ has potential for gold mineralisation. However, one of the weaknesses of this weights of evidence model is that the studies are carried out in 2D, with an approximation of 3D provided by geophysical and drilling data projected to a 2D plane. Consequently, a 3D prospectivity model was completed over the Ohakuri area, constrained by the results of the 2D model and predictive maps. The 3D model improved the results allowing more effective exploration targeting. However, the study also highlighted the main issues that need to be resolved before 3D prospectivity modelling becomes standard practise in the mineral exploration industry. The study also helped develop a work flow that incorporates preliminary 2D spatial data analysis from the weights of evidence technique to more effectively restrict and develop 3D predictive map interpretation and development.     

混合模糊证据权模型在河北承德煤炭资源预测中的应用

应用模糊逻辑法、加权证据权法相结合的混合模糊证据权模型和GeoDASGIS技术开展了承德煤炭资源预测研究。采用模糊逻辑法对与煤炭矿床有关的证据层进行了系统的处理和分析,并在此基础上采用加权证据权方法编制了成矿后验概率图,最终划分出5个主要的找矿远景区。研究结果不仅对进一步开展预测区优选评价具有重要的参考意义,而且为混合知识驱动与数据驱动的混合预测模型提供了一种可借鉴的有效方法。     

Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New Zealand): A critical tool for use in resource development and deep sea mining

After almost five decades of episodic exploration, feasibility studies are now being completed to mine the deep-water nodular phosphate deposit on the central Chatham Rise. Weights of evidence (WofE) and fuzzy logic prospectivity models have been used in these studies to help in mapping of the exploration and resource potential, to constrain resource estimation, to aid with geotechnical engineering and mine planning studies and to provide background geological data for the environmental consent process. Prospectivity modelling was carried out in two stages using weights of evidence and fuzzy logic techniques. A WofE prospectivity model covering the area of best data coverage was initially developed to define the geological and environmental variables that control the distribution of phosphate on the Chatham Rise and map areas where mineralised nodules are most likely to be present. The post-probability results from this model, in conjunction with unique conditions and confidence maps, were used to guide environmental modelling for setting aside protected zones, and also to assist with mine planning and future exploration planning. A regional scale fuzzy logic model was developed guided by the results of the spatial analysis of the WofE model, elucidating where future exploration should be targeted to give the best chance of success in expanding the known resource. The development work to date on the Chatham Rise for nodular phosphate mineralisation is an innovative example of how spatial data modelling techniques can be used not only at the exploration stage, but also to constrain resource estimation and aid with environmental studies, thereby greatly reducing development costs, improving the economics of mine planning and reducing the environmental impact of the project.     

Fuzzy inference systems for prospectivity modeling of mineral systems and a case-study for prospectivity mapping of surficial Uranium in Yeelirrie Area,Western Australia

A Mamdani-type fuzzy inference system for prospectivity modeling of mineral systems is described. The system is a type of knowledge-driven symbolic artificial intelligence that is transparent, intuitive and is easy to construct by geologists because they are built in natural language and use linguistic values. No examples are used for training the system and expert-opinions are incorporated indirectly in terms of objective mathematical functions, which reduce the possibility of over-emphasizing the known deposits usually used as training data. The cognitive reasoning of the exploration geologist is captured in explicit if–then type of statements written in natural language using linguistic values. Conditional dependencies in the exploration data sets are managed through the use of fuzzy operators. A case study for surficial uranium prospectivity modeling in the Yeelirrie area, Western Australia, is used to demonstrate the approach. In the output prospectivity map, the SE-NW trending Yeelirrie and E-W trending Hinkler's Well palaeochannels show high prospectivity, while other channels show very low prospectivity ranges. The known surficial uranium deposits fall in high prospectivity areas, although minor showings and anomalies in the southern part of the study area fall in low prospectivity areas. A comparison of the prospectivity model with the radiometric image shows that several channels showing high surface uranium concentrations in the NW and NE quadrants may not be prospective.     

Mineral systems approach applied to GIS-based 2D-prospectivity modelling of geological regions: Insights from Western Australia

GIS-based 2D prospectivity modelling of three greenfield geological regions of Western Australia, namely, the West Arunta Orogen, West Musgrave Orogen and Gascoyne Province, was implemented for a range of deposit types including orogenic and intrusion-related gold, volcanic sediment-hosted base-metal sulfides, magmatic nickel–copper and magmatic platinum group element sulfides, iron-oxide copper gold, tin–tungsten, igneous and metamorphic related rare earth elements, surficial uranium and unconformity-related uranium.Conceptual mineral systems models were generated to identify the targeting criteria. The inputs to the models were the spatial proxies derived from 1:100,000 to 1:500,000 scale public domain data. The results showed similar prospectivity patterns for all of the targeted deposit types except sediment-hosted uranium and surficial uranium deposit types. Once a favourable geodynamic architecture is established, it can sustain different mineral systems and produce diverse deposit types depending on the nature of ligands in the source regions and physical–chemical environment in the trap regions through repeated reactivation in the subsequent geological history. A model is proposed to explain the formation of different deposit types at different stages of tectonic evolution of a province. The implication for GIS-based 2D prospectivity modelling at the scale of geological region is that the prospectivity model may not be deposit type specific. Further, prospectivity modelling should be carried out sequentially at progressively finer scales (regional- to district- to camp-scale), using only the targeting criteria that are relevant at the specific scale to delineate targets for specific deposit types.     

宁芜盆地白象山矿区人工神经网络 三维成矿定位预测研究

本文基于三维地质环境,综合白象山矿区积累的地质资料和物探成果,首先开展三维地质建模工作,详细刻画了白象山矿区的三维地质结构;在三维地质模型基础上,利用三维空间分析手段对三维控矿因素进行定量挖掘,提取了多种三维控矿因素;最后采用人工神经网络方法进行三维成矿定位预测。预测结果显示,人工神经网络三维成矿定位预测能很好的定位出已知矿体,同时显示,在已知矿体北部及东部的深边部具有较高的成矿概率,可作为开展进一步找矿勘探的靶区。因此,人工神经网络三维成矿定位预测对于白象山矿区的应用是有效的,可服务于新老矿区的深边部三维成矿定位预测,同时可为隐伏矿、盲矿的成矿预测和优选靶区提供定量、定位新的方法和途径。     

Three-dimensional mineral prospectivity modeling for targeting of concealed mineralization within the Zhonggu iron orefield,Ningwu Basin,China

The Zhonggu iron orefield is one of the most important iron orefields in China, and is located in the south of the Ningwu volcanic basin, within the middle and lower Yangtze metallogenic belt of eastern China. Here, we present the results of new 3D prospectivity modeling that enabled the delineation of areas prospective for exploration of concealed and deep-seated Baixiangshan-type mineralization and Yangzhuang-type mineralization within the Zhonggu orefield; both of these deposits are Kiruna-type Fe-apatite deposits but are hosted by different formations within the Ningwu Basin. The modeling approach used during this study involves 3 steps: (1) combining available geological and geophysical data to construct 3D geological models; (2) generation of 3D predictive maps from these 3D geological models using 3D spatial analysis and 3D geophysical methods; (3) combining all of the 3D predictive maps using logistic regression to create a prospectivity map. This approach integrates a large amount of available geoscientific data using 3D methods, including 3D geological modeling, 3D/2D geophysical methods, and 3D spatial analysis and data integration methods. The resulting prospectivity model clearly identifies highly prospective areas that not only include areas of known mineralization but also a number of favorable targets for future mineral exploration. The 3D prospectivity modeling approach showcased within this study provides an efficient way to identify camp-scale concealed and deep-seated exploration targets and can easily be adapted for regional- and deposit- scale targeting.     

西藏铁格隆南铜(金)矿床三维模型分析与深部预测

铁格隆南铜(金)矿床是近年来在班公湖-怒江成矿带西段多龙矿集区新发现的超大型Cu(Au-Ag)矿床。本文针对铁格隆南矿区深部找矿问题,以现代成矿地质理论和多元地学信息综合分析技术为支撑,以构建矿床找矿模型为指导,依托数据库技术、3S技术、三维建模与可视化技术及地质统计学理论与方法,开展基于矿产地质、地球物理、地球化学等成矿条件及找矿标志的三维地质实体建模与矿化异常三维空间重构,将铁格隆南矿床的预测评价研究拓展到三维空间,揭示了区内成矿地质特征、地球化学及地球物理异常表征,据此探讨了矿床的成因及矿体分布特征。并在此基础上,开展了矿区的地质-地球化学-地球物理综合信息分析与预测评价,以期减少单一信息多解性和成矿条件不确定性,为铁格隆南矿区深部找矿工作提供参考。研究结果表明:在地质找矿模型指导下,基于深部成矿空间三维结构重构基础上的三维地质、地球物理、地球化学异常信息提取与综合分析,可以有效的识别成矿地质体和矿致异常信息,实现深部矿产资源靶区空间定位预测,为深部找矿预测研究提供了新思路。综合分析结果显示铁格隆南矿床深部找矿潜力巨大。     

Geology,geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin,Ghana, West Africa

This paper describes the geology and tectonics of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, as applied to predictive mapping of prospectivity for orogenic gold mineral systems within the basin. The main objective of the study was to identify the most prospective ground for orogenic gold deposits within the Paleoproterozoic Kumasi Basin. A knowledge-driven, two-stage fuzzy inference system (FIS) was used for prospectivity modelling. The spatial proxies that served as input to the FIS were derived based on a conceptual model of gold mineral systems in the Kumasi Basin. As a first step, key components of the mineral system were predictively modelled using a Mamdani-type FIS. The second step involved combining the individual FIS outputs using a conjunction (product) operator to produce a continuous-scale prospectivity map. Using a cumulative area fuzzy favourability (CAFF) curve approach, this map was reclassified into a ternary prospectivity map divided into high-prospectivity, moderate-prospectivity and low-prospectivity areas, respectively. The spatial distribution of the known gold deposits within the study area relative to that of the prospective and non-prospective areas served as a means for evaluating the capture efficiency of our model. Approximately 99% of the known gold deposits and occurrences fall within high- and moderate-prospectivity areas that occupy 31% of the total study area. The high- and moderate-prospectivity areas illustrated by the prospectivity map are elongate features that are spatially coincident with areas of structural complexity along and reactivation during D4 of NE–SW-striking D2 thrust faults and subsidiary structures, implying a strong structural control on gold mineralization in the Kumasi Basin. In conclusion, our FIS approach to mapping gold prospectivity, which was based entirely on the conceptual reasoning of expert geologists and ignored the spatial distribution of known gold deposits for prospectivity estimation, effectively captured the main mineralized trends. As such, this study also demonstrates the effectiveness of FIS in capturing the linguistic reasoning of expert knowledge by exploration geologists. In spite of using a large number of variables, the curse of dimensionality was precluded because no training data are required for parameter estimation.     

GIS-based 3D prospectivity mapping: A case study of Jiama copper-polymetallic deposit in Tibet,China

This paper reports a deposit-scale GIS-based 3D mineral potential assessment of the Jiama copper-polymetallic deposit in Tibet, China. The assessment was achieved through a sequential implementation of metallogenic modelling and 3D modelling of geology, geochemistry and prospectivity. A metallogenic model for the Jiama deposit and 3D modelling workflow were used to construct multiple 3D layers of volumetric and triangular mesh models to represent geology, geochemistry and ore-controlling features in the study area. A GIS-based 3D weights-of-evidence analysis was then used to estimate the subsurface prospectivity for Cu (Mo) orebodies in the area, which led to the identification of three prospective deep-seated exploration targets. Additionally, the geochemical modelling indicates three potential fluid flow pathways based on the 3D zonation of major geochemical elements and their ratios, particularly the Zn/Pb ratios, which support the results of the weights of evidence model.     

Prospectivity for epithermal gold–silver deposits in the Deseado Massif,Argentina

Previous prospectivity modelling for epithermal Au–Ag deposits in the Deseado Massif, southern Argentina, provided regional-scale prospectivity maps that were of limited help in guiding exploration activities within districts or smaller areas, because of their low level of detail. Because several districts in the Deseado Massif still need to be explored, prospectivity maps produced with higher detail would be more helpful for exploration in this region.We mapped prospectivity for low- and intermediate-sulfidation epithermal deposits (LISEDs) in the Deseado Massif at both regional and district scales, producing two different prospectivity models, one at regional scale and the other at district-scale. The models were obtained from two datasets of geological evidence layers by the weights-of-evidence (WofE) method. We used more deposits than in previous studies, and we applied the leave-one-out cross validation (LOOCV) method, which allowed using all deposits for training and validating the models. To ensure statistical robustness, the regional and district-scale models were selected amongst six combinations of geological evidence layers based on results from conditional independence tests.The regional-scale model (1000 m spatial resolution), was generated with readily available data, including a lithological layer with limited detail and accuracy, a clay alteration layer derived from a Landsat 5/7 band ratio, and a map of proximity to regional-scale structures. The district-scale model (100 m spatial resolution) was generated from evidence layers that were more detailed, accurate and diverse than the regional-scale layers. They were also more cumbersome to process and combine to cover large areas. The evidence layers included clay alteration and silica abundance derived from ASTER data, and a map of lineament densities. The use of these evidence layers was restricted to areas of favourable lithologies, which were derived from a geological map of higher detail and accuracy than the one used for the regional-scale prospectivity mapping.The two prospectivity models were compared and their suitability for prediction of the prospectivity in the district-scale area was determined. During the modelling process, the spatial association of the different types of evidence and the mineral deposits were calculated. Based on these results the relative importance of the different evidence layers could be determined. It could be inferred which type of geological evidence could potentially improve the modelling results by additional investigation and better representation.We conclude that prospectivity mapping for LISEDs at regional and district-scales were successfully carried out by using WofE and LOOCV methods. Our regional-scale prospectivity model was better than previous prospectivity models of the Deseado Massif. Our district-scale prospectivity model showed to be more effective, reliable and useful than the regional-scale model for mapping at district level. This resulted from the use of higher resolution evidential layers, higher detail and accuracy of the geological maps, and the application of ASTER data instead of Landsat ETM + data. District-scale prospectivity mapping could be further improved by: a) a more accurate determination of the age of mineralization relative to that of lithological units in the districts; b) more accurate and detailed mapping of the favourable units than what is currently available; c) a better understanding of the relationships between LISEDs and the geological evidence used in this research, in particular the relationship with hydrothermal clay alteration, and the method of detection of the clay minerals; and d) inclusion of other data layers, such as geochemistry and geophysics, that have not been used in this study.     

Improved Wildcat Modelling of Mineral Prospectivity

Wildcat modelling of mineral prospectivity has been proposed for greenfields geologically-permissive terranes where mineral targets have not yet been discovered but a geological map is available as a source of spatial data of predictors of mineral prospectivity. This paper (i) revisits the initial way of assigning wildcat scores (S_c) to predictors of mineral prospectivity and (ii) proposes an improvement by transforming S_c into improved wildcat scores (IS_c) by using a logistic function. This was shown in wildcat modelling of prospectivity for low-sulphidation epithermal-Au (LSEG) deposits in Aroroy district (Philippines). Based on knowledge of characteristics of and controls on LSEG mineralization in the Philippines, the spatial predictors of LSEG prospectivity used in the study are proximity to porphyry plutonic stocks, faults/fractures and fault/fracture intersections. The S_c and IS_c of spatial predictors are input separately to principal components analysis to extract a favourability function that can be interpreted as a wildcat model of LSEG prospectivity. The predictive capacity of the wildcat model of LSEG prospectivity based on the IS_c of geological predictors is roughly 70% higher than that of the wildcat model of LSEG prospectivity based on the S_c of geological predictors. A slight increase of predictive capacity of wildcat modelling of LSEG prospectivity is also achieved when the IS_c of geological predictors are integrated with the IS_c of geochemical anomalies, but not with the S_c of geochemical anomalies. The proposed improvement is significant because if the study district were a greenfields exploration area, then a wildcat model of LSEG prospectivity based on the old wildcat methodology would have caused several LSEG targets to be missed.     

Mapping mineral prospectivity using an extreme learning machine regression

In this research, we conduct a case study of mapping polymetallic prospectivity using an extreme learning machine (ELM) regression. A Quad-Core CPU 1.8 GHz laptop computer served as hardware platform. Almeida's Python program was used to construct the ELM regression model to map polymetallic prospectivity of the Lalingzaohuo district in Qinghai Province in China. Based on geologic, metallogenic, and statistical analyses of the study area, one target and eight predictor map patterns and two training sets were then used to train the ELM regression and logistic regression models. ELM regression modeling using the two training sets spends 61.4 s and 65.9 s; whereas the logistic regression modeling using the two training sets spends 1704.0 s and 1628.0 s. The four trained regression models were used to map polymetallic prospectivity. Based on the polymetallic prospectivity predicted by each model, the receiver operating characteristic (ROC) curve was plotted and the area under the curve (AUC) was estimated. The ROC curves show that the two ELM-regression-based models somewhat dominate the two logistic-regression-based models over the ROC performance space; and the AUC values indicate that the overall performances of the two ELM-regression-based models are somewhat better than those of the two logistic-regression-based models. Hence, the ELM-regression-based models slightly outperform the logistic-regression-based models in mapping polymetallic prospectivity. Polymetallic targets were optimally delineated by using the Youden index to maximize spatial association between the delineated polymetallic targets and the discovered polymetallic deposits. The polymetallic targets predicted by the two ELM-regression-based models occupy lower percentage of the study area (2.66–2.68%) compared to those predicted by the two logistic-regression-based models (4.96%) but contain the same percentage of the discovered polymetallic deposits (82%). Therefore, the ELM regression is a useful fast-learning data-driven model that slightly outperforms the widely used logistic regression model in mapping mineral prospectivity. The case study reveals that the magmatic complexes, which intruded into the Baishahe Formation of the Paleoproterozoic Jinshuikou Group or the Carboniferous Dagangou and Shiguaizi Formations, and which were controlled by northwest-western/east-western trending deep faults, are critical for polymetallic mineralization and need to be paid much attention to in future mineral exploration in the study area.