You are on page 1of 11

Proceeding volume of 5th annual international conference Map India 2002, 6-8th Feb., 2002, ew Delhi, pp 405-411.

http:// Mineral potential map by a knowledge driven GIS modelling: an example from Singhbhum copper belt, Jharkhand

Basab Mukhopadhyay

Niladri Hazra

Sujit Ranjan Sengupta

*Swapan Kumar Das Geological Survey of India, Geodata and Database Division CHQ, 27, J. L. Nehru Road, Kolkata *Geological Survey of India, Project Geoinformatics ERO, MSO Building, Salt Lake City, Kolkata

Abstract Mineral resource potential mapping is a complex analytical process, which requires consideration and integration of a number of spatial evidences like geological, geomorphological, wall rock alteration, etc., using the capability of analytical tools of Geographic Information System (GIS). The Singhbhum Copper Belt - a narrow, arcuate highly sheared linear zone in the Singhbhum Precambrian Terrain in Jharkhand, runs over a length of 128 km from Kharswan-Duarpuram in the west to Baharagora in the east. This modelling is an attempt to re-examine the huge inventory of spatial as well as attribute data in GIS datasets (collated from GSI published and unpublished work) in the light of certain evidences recognizable on a regional scale the existing exploration model, favourable for copper mineralisation. The input data for the analysis include i) lithological evidences in the form of lithology, favourable contacts, ii) alteration evidences such as chloritisation/biotitisation/tourmalinisation /sericitisation etc., iii) geophysical anomalies such as aero-magnetic, radiometry, ground geophysics (SP, IP, EM), iv) structural evidences such as lineament, shear-zone and v) geochemical anomalies

such as analytical value (copper) for bed rock samples. A knowledge driven weight on evidence approach was employed to establish relationship between the input datasets and exploration model. In this approach, individual basic layers of evidences are integrated in maps on the basis of a score assigned, according to their influence towards mineralisation. Each element of the input layer used as evidence is assigned a different score (weight) to generate secondary factor maps. In the next phase, the factor maps were combined with different map weight depending on their relevance towards mineralisation. Finally, all the factor maps were integrated to generate a mineral potential map by additive union, using index overlay method. The resulting mineral potential map in probability scale was cross validated by plotting the known mineral deposits, the model shows good match. It further identifies two localities, E and SE of Kanyaluka Gohala area and a large area West of Turamdih. In between these two areas there are small lenticular pockets considered to be the potential site for further copper exploration. Introduction The Singhbhum Copper Belt - an arcuate linear zone in the Precambrian Singhbhum Crustal Province (PSCP), houses over 250 million tonnes of copper ore with variable tenor ranging from 0.5 to 4% of copper. Though the geological environment for the entire belt extending over 128 km from KharswanDuarpuram in the West to Baharagora in the southeast, is a favourable locale for sulphide mineralisation, economic deposits are restricted only over 15% of the total stretch in the central part of the belt. Many of these identified deposits lack the requisite tonnage necessary for mining. In view of the exhaustion of identified deposits for detailed exploration, ore-reserves have remained stagnant over a decade, an examination of the belt applying advanced GIS techniques seems worthwhile to find new locales for mineralisation. Mineral resource potential mapping is a very complex analytical procedure which requires simultaneous consideration of a number of spatial evidences - geological, geomorphological, structural, geochemical, geophysical etc. The capability of Geographic Information System (GIS) to manipulate such classified spatial information through amalgamated layers, makes it a unique tool for delineating potential locales. Flexibility of experimenting with spatial data followed by visualisation of its effect immediately, gives GIS a cutting edge over other contemporary techniques, for modelling mineral deposits. The PSCP has attracted a number of geoscientist through the ages. Five decades of exploration, has generated huge inventory of spatial and attribute data. This paper is an attempt to re-examine the data in the light of certain criteria favourable for mineralisation of copper, which are recognizable on regional scale. The predictive GIS model is based on weights of evidence analysis of lithological, structural, geochemical and geophysical data sets employing knowledge driven GIS approach.

Geology and mineralisation: An overview The Singhbhum Shear Zone is located along the southern fringe of the Proterozoic fold belt of North

Singhbhum which is sandwiched between the early Archean cratonic nucleous represented by Singhbhum and Bonai granite, to the south and Proterozoic Chottanagpur Granite Complex to the north. The intervening gap area between Singhbhum and Chotanagpur crustal provinces is occupied by a curvilinear belt of metasedimentaries belonging to Dhanjori and Singhbhum Group of Proterozoic age. The Singhbhum shear zone which has developed in this Proterozoic belt, is a northerly dipping arcuate ductile shear zone ( Ghosh & Sengupta, 1987) marked by lenticular mylonite zone. The trend of the shear zone changes from E-W in the western part to NNE-SSW in the eastern part. The rocks within the Singhbhum Shear Zone form a tectonic mlange comprising granite mylonite, quartz-mica phyllonite, quartz tourmaline rock and deformed volcanic and volcano clastic rocks (Mukhopadhyay & Deb, 1995). The shear sense indicator suggests a thrust type of deformation (Mukhopadhyay & Deb, 1995). The zone is characterised by mylonitisation, syntectonic with the first phase of deformation. The mineralised zone Singhbhum Copper Belt runs close to the interface of Singhbhum Group of rocks mainly schists and Dhanjori volcano-sedimentary package. The copper mineralisation in this belt consists of number of parallel to sub-parallel discontinuous lodes aligned along the major tectonic grains of the area. The predominant sulphide minerals are chalcopyrite, pyrite and pyrrhotite. The mode of the occurrence varies from massive to braided veins, stringers, dissemination, discordant to sheet like bodies and also as en-echelon veins.Existence of structural and lithological controls for the mineralisation have been suggested by several workers. Sarkar, suggested that sulphide mineralisation in this belt is confined mainly within certain stratigraphic horizon adjacent to the Dhanjori metavolcanics. The remobilization has mainly taken place along shear bands. The general trend of the ore body is controlled by the local trend of the slip planes (report on Project Singhbhum, 1991) . The wall rock alteration in the form of chloritisation, sericitisation, biotitisation, tourmalinisation and albitisation are common. Exploration model for copper in Singhbhum copper belt: Any attempt of deposit modelling for Singhbhum Copper Belt takes into consideration certain general observations made by earlier workers, regarding characteristics of copper mineralisation in relation to lithology, tectonics, mode of emplacement etc. They can be summarised as follows: Bore hole data corroborate that soda granite, chlorite schist and altered basic volcanic rock (talc-chlorite schist) of both Singhbhum and Dhanjori Group act as host for mineralisation in maximum instances. The host rock may be broadly categorised as metasediment, metabasic rocks, granitoid rocks and meta-ultrabasic rocks. Litho contacts served as easy channel for mobilisation of ore during shearing. The planar fabrics generated during shearing are the fundamental planes for ore localisation in macroscopic, mesoscopic and regional scales. The lineaments parallel to the shear zone also served as general conduit for ore mobilisation. High aerogeophysical and radiometric anomalies are important signatures for sub surface mineralisation. The ground geophysical anomaly axes (of IP, SP ,EM and magnetic) also, proved effective in delineating hidden target areas. The wall rock alteration, in the form of chloritisation and biotitisation throughout the belt as well as tourmalinisation in Rakha mines and sericitisation at Mosabani mines point towards hydrothermal nature of copper mineralisation. Presence of bed rock geochemical anomaly, is an indicator of hidden copper mineralisation.

GIS data sets Geological Survey of India has carried out stupendous work in the area (Fig 1) under discussion for last several years by way of mapping on scale 1: 63,360 and extensive Airborne Geophysical Surveys. Electromagnetic, magnetic and scintillometric sensor surveys have also been carried out, followed by ground geochemical and geophysical surveys in selected areas to assess their mineral potential. Nearly 600 boreholes were drilled in the area to study the nature of the ore bodies and estimation of reserve, after detailed geological mapping of prostective locations (on 1:10,000; 1:5,000; 1:2000 scale).

The present paper deals with the result of the analyses of the collated digital data and maps from published and unpublished reports of GSI. Apart from the digital information, the attribute data available in various thematic domains was also consulted. Extensive help was also sought from the synthesis work done under project Singhbhum. The GIS dataset comprise of following layers: Lithological evidences Lithology: Existing field geological maps compiled in scale 1:50,000, and digitised using ARC/INFO GIS. The data was edited and cleaned to polygon coverages and lithological attributes were updated as per present understanding of the litho-units. Favourable contacts: The favourable litho-contacts marked along the lithological boundaries are digitised from the same compiled maps.

Structural evidences Lineament: The major lineaments within the selected area were digitised from interpreted imagery maps (Pers. Com. Mr. D.P. Das, Geologist (Sr.)) and the coverage was clipped to get the desired area. Shear-zone: The most important and controversial factor of the model was digitised as a linear zone from the compilation under Project Geoinformatics. This line represents the modal plane of maximum shearing.

Alteration evidences The main litho-units which underwent alteration as biotitisation, chloritisation, tourmalinisation and sericitisation were selected from the geological maps and digitised. Polygon coverages were prepared from them with an added attribute. Geophysical evidences Aero-magnetic- Contours were digitised as polygons from aero-magnetic anomaly maps and contour values included in the polygon attribute table. Radiometry - Points were digitised from radiometry maps and radiometry counts were included in the PAT table. Ground Geophysics-(anomaly axis of IP, SP,EM and magnetic) -Linear anomaly axis on interpreted ground geophysical maps were digitised .

Geochemical evidences Interpreted geochemical anomalies (from analytical value of bed rock samples) in the form of contour maps available were digitised into polygon coverages. GIS modelling, Weight on evidence in Index overlay method Weight of evidence is the method of combining subjective evidences on the basis of their bearing towards a process quantitatively. The method was originally developed for non-spatial application in medical science and adopted in the late 1980s for mineral potential mapping. The evidence consists of a set of exploration dataset (maps) and weights are estimated from the measured association between known mineral occurrences and the exploration model for a particular terrain. The hypothesis then repeatedly evaluate all possible location of the maps using the weight and in turn produce a mineral potential map in which the evidences of several map layers are combined by a map combination rule, say index overlay method. Thus, this method belongs to those groups, where multicriteria analytical procedure is employed to generate a decision making map. In the index overlay method, each input map (layer of evidence) to be used as evidence is assigned a different score (weight), as well as the map themselves are receiving different weight (Bonham-carter, 1994) depending on the exploration model. The weight can be calculated statistically depending on the number of input mineral occurrences and its relationship with a particular element or theme on a map. Otherwise, weight can be calculated on the basis of relative importance of elements/themes on the input and evidence maps by mineral deposit experts (as expected in knowledge driven approach). Assignment of weight or score to maps by statistical process may be applicable where a unified and definitive exploration model is unknown or the relationship / relative importance between different elements is uncertain. In contrast, an area which is geologically well explored with a relatively wellunderstood exploration model in hand (the present study area), assignment of weight on different themes or maps ought to be through knowledge driven approach. This not only help in developing a clear understanding of relationship between datasets (both geological, geophysical or geochemical ) but this also give flexibility to an exploration geologist to manipulate weight on different elements or evidence maps through geological knowledge about the terrain in different stages of analysis. This is advantageous for developing perhaps a variety of scenarios for different weight schemes, reflecting differences in opinion amongst experts, and allows the evaluation of sensitivity of the mineral potential maps to such differences. This flavour of understanding is generally absent in the data driven statistical approach where the system (the statistical model) directs the exploration and generates decision, which may be irrelevant in the complex geological system. After defining the score by statistical (data driven) or by knowledge driven approach for elements or maps, the average score ( index weight) is then defined by

Where S is the weighted score for an area object, Wi is the weight for the i-th input map, and Sij is the score for the j-th class of the i-th map, the value of j depending on the class actually occurring at the current location (Bonham-carter, 1994). In the present analysis, a knowledge driven intuitive approach was employed by seeking views from exploration geologist or exploration model for finalisation of the weight or score on different themes / evidence maps and finally all these evidence maps were combined by additive union method to generate the mineral potential maps. Probabilistic analysis to assign score was not possible due to, i) very complex evolutionary history of the terrain ii) data density and spacing is not congenial for

carrying out statistical analysis. Thus an intuitive approach with inputs from exploration experts of the region was applied. Weights of evidence were calculated for each of the datasets analysed. Simple point in polygon analysis was performed in case of polygon domains (i.e., lithologic units, geochemical anomaly). In case of linear features, such as lineament, shear-zone buffer analysis was done to create geologically meaningful zones of influence. These zones were then analysed as polygons and weights computed. The final predictive map was calculated by overlaying the images (raster data) created from the weights of evidence, representing the deposit recognition criteria, and each map summed up the weights at every location. Table 1 Lithology-id 1000 1950 3075 3100 2000 3200 1900 3050 2100 Score lithological units 9 9 9 9 8 8 6 6 6 Soda Granite Chlorite Schist ( Singhbhum Group) Chlorite Schist ( Dhanjori Group) Talc chlorite schist Hornblende schist and Epidiorite (Singhbhum Group) Hornblende schist and Epidiorite (Dhanjori group) Younger ultrabasics (Singhbhum Group) Ultrabasics ( Dhanjori Group) Mica schist with Hornblende schist Rest of the rocktype

300/500/600 2 /2200/2300/ 2600/3300/ 3400/3500/ 3600/4100/ 4150/4200 /4300/4400 /4500/4700 /5000/5100

GIS data processing Compilation of the spatial data was followed by processing of data for extraction of evidences critical to the prognostication of copper deposits, through the guidelines imposed by the exploration model. The principal approaches to process the data, prior to data combination are illustrated below: Map reclassification, to simplify the complex geological map to small number of simplified units suitable for modelling. Generation of proximity maps showing distance to linear or polygonal features such as formation contacts, lineaments, shear zone, aeromagnetic and ground geophysical anomalies, etc (buffering). Extraction of bed rock geochemical anomalies of copper Generation of anomaly maps for the wall rock alteration elements Combining the relevant layers of evidences to generate the intermediate Factor Maps namely, Lithological Factor, Geochemical Factor, Structural Factor, Alteration Factor and Geophysical Factor.

Lithological Factor Analysis of host lithology

The sulphide mineralisation in the Singhbhum belt is broadly considered to be confined within the Dhanjori Group of rocks south of the Shear Zone and Singhbhum Group of rocks north of the shear zone. Both overlie the Singhbhum granite unconformably. The Dhanjori Group consists of volcanics, volcanoclastics and lithic tuffs those underwent a low grade metamorphism. The Singhbhum Group consist of high grade mica schist, hornblende schist and quartz granulite at the bottom and comparatively less metamorphosed. The phyllites are considered to be derivative of volcanic tuffs. The sulphide mineralisation is considered to be associated mainly with the meta-volcanics and metatuffs of Singhbhum and Dhanjori Groups, which has been remobilised and concentrated during the later metamorphic processes. Though mineralisation occurs in almost all the rock types of these stratigraphic horizons, the borehole data reveals that soda granite, chlorite and sericite schist and altered basic volcanics are the most favourable host of copper mineralisation. The geological map, originally containing 28 lithounits, was reclassified into the following units with a new attribute lithology-id added to the polygon attribute table (PAT). A new attribute score signifying relative importance of the mapunits on a relative scale of 1 to 10 is also added. The classified units and associated scores are illustrated below (Table 1). The coverage is rasterised on the basis of the attribute score. Analysis of favourable contacts The detailed exploration in this terrain reveals that the contact between metasediments and metavolcanics are generally mineralized to a varied extent. The favourable contacts were buffered at an interval of 250m and two new attributes contact-id and score added to the PAT (Table 2). The coverage is rasterised on the basis of the score. Table 2 Contact-id Score Contact between 1 2 3 4 9 8 6 2 Soda granite, chlorite scist and talc chlorite schist Hornblende schist and epidiorite Ultrabasics and mica schist Contacts with rest of the rocktype

These two layers representing the simplified geological map and proximity to favourable contacts are spatially combined by additive union after giving map score 8 and 6 respectively to produce intermediate map. The intermediate map divided by the total score (8+6 = 14) to produce the Lithological Factor Map (Fig 2). Geophysical Factor Analysis of aerogeophysical data Aerogeophysical surveys carried out by GSI under project Operation Hard rock recorded electromagnetic, magnetic (TF) and radiometric ( total count) values. Aeromagnetic contours were traced out at 250 gamma interval, over relevant portion of the map, to project source of distraction on different lithologies. As raw data was not available, the interpreted aerogeophysical anomaly maps was digitised and polygons containing the anomalies were selected to create the aerogeophysical layer of evidence. The aerogeophysical data represents a range from 700 to 4500 gamma values with a mean of 3108 gamma. An attribute score was added to the PAT table. The score is assigned to each polygon on the basis of the value in that particular polygon divided by 4500. This process generates a scale 0 to 1. Individual anomaly polygons were codified, weighted and rasterised on the basis of the score. Analysis of radiometric data Point radiometric (total count) anomalies were buffered using a buffer distance of 500m to generate the next layer of evidence. The radiometric data represents a range from 2400 to 100 gamma values

with a mean of 500 . An attribute score was added to the PAT table. The score is assigned to each polygon on the basis of the value in that particular polygon divided by 2400. This process generates a scale 0 to 1. Individual anomaly polygons were codified, weighted and rasterized on the basis of the score. Analysis of ground geophysical data The ground geophysical survey was carried out in those areas which exhibited high gravity and aeromagnetic anomalies. The delineated high anomaly axes of Induced potential (IP), self potential (SP) and electro magnetic (EM) were buffered with a distance of 250m to generate the layer of evidence. An attribute score was added to the PAT file (Table3). Individual anomalies are rasterised on the basis of the score. These three layers of evidence (aeromagnetic, radiometry and ground geophysical) were spatially combined to generate intermediate maps after assigning map score 5, 5 and 7 respectively. The intermediate map was divided by the total score (17) to produce the Geophysical Factor Map (Fig 2). Table 3 Ground - id score Characteristics 1 2 9 2 Strong ground geophysical anomaly Low ground geophysical anomaly

Geochemical Factor Geochemical evaluation of an area leads to identification of potential host rock and possible source of mineralisation in that area. Secondary dispersion of elements in soil depends on degree of maturity, nature, thickness and the topographic expression of the area. Since topographic expression is not uniform in the area, a definite relationship of the soil and bed rock chemistry could not be established. Therefore, geochemical data of the bed rock alone was used for contouring to predict the primary dispersion of the copper. Geochemical anomaly maps of copper were digitised and copper values assigned for each polygon. The copper values range from 100 to 4000 ppm. An attribute score was added to the pat file and the value of the score of each polygon is assigned by dividing the copper value by 4000. Individual anomalies are rasterised on the basis of the score to create Geochemical factor map (Fig 2). Structural Factor Analysis of the shear zone The line representing the maximum shearing effect was digitised. The shear line was differentially buffered at intervals of 500m and a new attributes score added to the PAT (Table 4). The shear zone formulated by this process, were rasterised to generate image on the basis of the score. Analysis of lineaments and fault Lineaments interpreted from LANDSAT imagery and regional structures (faults) were buffered using a buffer distance of 100m on each side. The polygons were reclassified according to their orientation and accordingly the score was added to the pat file(Table 5). These two layers of evidence (shear zone and lineaments/faults) were spatially combined to generate intermediate maps after assigning map score 8 and 4 respectively. The intermediate map was divided by the total score (12) to produce the Structural Factor Map (Fig 2).

Table 4 Shear-id 1 2 3 4 5 6 score 10 9 8 7 6 5 Distance from shear zone 500m 1000m 1500m 2000m 2500m 3000m

Table 5 Linea-id score Relation with shear zone 1 2 3 5 3 1 Within 250m of shear zone and parallel to sub-parallel Within 250m of shear zone and not parallel Outside 250m of shear zone

Alteration Factor Wall rock alteration and geochemical haloes are important for discerning hidden ore deposit. Four types of alteration are prevalent in the area. The images generated by presence or absence of these hydrothermal alteration (biotitisation, chloritisation, tourmalinisation and sericitisation) and are later integrated by additive union to generate Alteration factor map (Fig 2). Map Combination The map combination is a process where different layers of evidences are combined with respect to a map combination rule. The predictive model was generated by summing up the various derived maps that provide evidence for copper mineralisation into one or more mineral favourability maps. The overlap combination process of the images involves the weighing and union of evidences by a map combination rule i.e. Index overlay method. The intermediate factor maps are combined to generate binary maps resulted in the highest cumulative weights in the area where all the recognition criteria co-exist. As discussed earlier in the index overlay method each class of every map is given a different score as well as the maps themselves receiving different weights allowing for a more flexible weighting system. The individual score of the map elements have been discussed in the previous section. The following table6 shows the map scores assigned to the individual layers of evidences. Table 6 Factor Map Geological factor Geochemical Factor Weight 7 Derived Map lithological Map Favourable contact 7 Copper anomaly Aero-magnetic Geophysical factor 6 Radiometric Ground Geophysics Structural factor Alteration factor 6 shear zone lineament 3 combined images of wall rock alteration data Weight 8 6 5 5 7 8 4 12 17 Sumw 14

After each evidence layer has been weighted by assigning scores, the input maps were combined in stages. The inference network that summarises the combination process is shown in Figure 2 e.g. during combination, class scores of Lithological map (9,8,6,2) are multiplied by weight 8; favourable contact (8,6,4,2) by 6 respectively. The combined map is normalized by dividing with SUMW i.e., 14. The resulting map is the evidence of lithological factor. Similarly, other factors are also calculated. The mineral potential map which is created by combining the five factor maps (lithological factor * map weight i.e. 7 + geochemical factor * 7 + geophysical factor * 6 + Structural factor * 6 + Alteration factor * 3) has the scores normalized by dividing with the value 29, i.e. SUMW of the Factor maps. This map grades the region according to the degree of probability. The mineral potential map identifies new copper potential zones as well as the known deposit areas (Fig-3). The GIS model was cross-validated by plotting the known mineral occurrences/deposit into the mineral potential map. It is found that 72% of the deposits are lying on the high probability zone. Thus the criteria chosen for this knowledge driven modelling fitted well with the existing ground reality. It is to be noted that a number of mineral potential maps can be generated from the same dataset after choosing different weight scheme or by different map combination rules. The success of the model lies only when more deposits can be found out. So this modeling process is not an end but beginning of a new process.

Conclusion The highly prospective nature of the Singhbhum Copper Belt for hosting mata-sediment hosted hydrothermal copper deposit has been widely known since middle of 19th Century. As many geologists have argued for years to ascertain the mode of origin and control on copper mineralisation, the belt indeed result from a structurally shear controlled copper bearing magmatic activity which ultimately release its load to the favourable locales. The predictive GIS model has unequivocally confirmed these observations. The favourability map (Fig 3) has broadly identified the target areas for evaluation through detailed exploration. The model identifies two broad localities E and SE of Kanyaluka Gohala area and a large area West of Turamdih. In between these two areas there are small linear pockets identified as highest probability zone (cloured as red) can also be treated as the potential site for further exploration. This model and characterisation of the detailed relationship between layers (spatial analysis) can be applied to the adjoining less geologically explored areas. This not only reduces the cost of detailed exploration but also helpful in generating a predictive GIS model in the similar tectono-stratigraphic set up. Acknowledgement The authors express their gratitude to all the officers worked on the Project GeoinformaticsSinghbhum Precambrian on 73J. Directors and other officers, Geodata Division, CHQ are acknowledged for their constructive criticism, providing infrastructure and help during the course of

the study. Authors remain grateful to Dr. M.K. Mukhopadhyay, Director and Dr. Sabyasachi Dasgupta, Geologist(Sr.) of GSI, CHQ for scrutinising the earlier version of the manuscript. References Bonham-Carter, G. F.,1994, Geographic Information System for Geoscientists: Modelling with GIS: Pergamon. Ghosh, S. K. & Sengupta, S., 1987, Progressive development of structures in a ductile shear zone: Journal of Structural Geology: 9 : 277-287. Mukhopadhyay, D. and Deb, G.K., 1995, Structural and textural development in Singhbhum Shear Zone, Eastern India, Proc.Indian.Acad.Sci, v104, no3: 385-408. Officers of GSI, ER, 1991, Unpublished GSI report on Project Singhbhum Synthesis of data of Singhbhum Copper Belt, Singhbhum District, Bihar: Part I & II. Sarkar, S.N., Ghosh, D.K. & Lambert, R.J. St., 1986, Rubidium-Strontium and lead isotopic studies of the soda granites from Musaboni area, Singhbhum Copper Belt. In : Geology and Geochemistry of sulphide ore bodies and associated rocks in Musaboni and Rakha Mines Section in the Singhbhum Copper Belt., 409. Diamond Jubilee Monograph : ISM (Dhanbad):101-110.