Spectral Characteristics of Hydrothermal Alteration in Zuru, NW Nigeria

Abstract

This study demonstrated the ability of a Landsat-8 OLI multispectral data to identify and delineate hydrothermal alteration zones around auriferous prospects within the crystalline basement, North-western Nigeria. Remote sensing techniques have been widely used in lithological, structural discrimination and alteration rock delineation, and in general geological studies. Several artisanal mining activities for gold deposit occur in the surrounding areas within the basement complex and the search for new possible mineralized zones have heightened in recent times. Systematic Landsat-8 OLI data processing methods such as colour composite, band ratio and minimum noise fraction were used in this study. Colour composite of band 4, 3 and 2 was displayed in Red-Green-Blue colour image to distinguish lithologies. Band ratio ${\frac{4}{2}}$ image displayed in red was used to highlight ferric-ion bearing minerals(hematite, goethite, jarosite) associated with hydrothermal alteration, band ratio ${\frac{5}{6}}$ image displayed in green was used to highlight ferrous-ion bearing minerals such as olivine, amphibole and pyroxenes, while ratio ${\frac{6}{7}}$ image displayed in blue was used to highlight clay minerals, micas, talc-carbonates, etc. Band rationing helped to reduce the topographic illumination effect within images. The result of this study showed the distribution of the lithological units and the hydrothermal alteration zone which can be further prospected for mineral reserves.

1. Introduction

Hydrothermal alteration is a very complex process involving mineralogical, chemical and textural changes, resulting from the interaction of hot aqueous fluids with the rocks through which they pass, under evolving physico-chemical conditions(Pirajno, 1992). Alteration can take place under magmatic sub-solidus conditions by the action and infiltration ofsupercritical fluids into a rock mass. At lower temperature and pressure, ex-solution of gas and aqueous phases constitute hydrothermal solutions which act on the surrounding rocks, producing changes as the result of disequilibrium, largely due to H⁺ and OH^– and other volatile constituents (e.g. CO₂) (Pirajno, 1992). In essence, hydrothermal fluids chemically attack the mineral constituents of the wall rocks, which tend to re-equilibrate by forming newmineral assemblagesthat are in equilibrium with the new conditions.The process is a form of metasomatism, i.e. exchange of chemical components between the fluids and the wall rocks. Therefore, it is also possible that the fluids themselves may change their composition as a result of their interaction with the wall rocks. The main factors controlling alteration processes are: (1) the nature of wall rocks; (2) composition of the fluids; (3) concentration, activity and chemical potential of the fluid components, such as H⁺ , CO₂, O₂, K⁺ , S₂, etc. (Rose et al., 1979). Green et al. (1988) believe that alteration productsin epithermalsystems do not depend so much on wall rock composition as on permeability, temperature and fluid composition. They cite, for example, that in the temperature range of 250-280°C, similarmineral assemblages(e.g. quartz-albite-feldsparepidote-illite-calcite-pyrite) are formed in basalts, sandstone, rhyolite and andesite. Other researchers (Rose etal.,1979),however, emphasizedthe fundamental role played by the nature and composition of wallrocks in hydrothermal alteration processes, particularly in porphyry systems. The action of hydrothermal fluids on wall rocks is by infiltration and/or diffusion of chemical species (Rose et al., 1979).

Recognition of the existence of hydrothermal alteration zones is important, as it is associated with mineralization and its delineation increases the size of the target mineralized zone during mineral exploration (Moradi et al., 2015).Thus, the volume of potential ore may be greater when alteration zones can be identified. This hasled to the development of the concept of large tonnage, low grade mesothermal mineral deposits that include mineralized host rock as well as the narrow mineralized quartz veins that have traditionally been the focus of mining (Rockwell et al., 2008).

2. Background

New generations of advanced remote sensing data have been used by geoscientists over the last decade with the main focus on global dynamics in mineral mapping, environmental geology, geothermal and hydrocarbon exploration. Mineral exploration with satellite images such as Landsat-8 Operational Land Imager(OLI)is useful in the early stages of exploration. Such minerals associated with hydrothermal alteration have been studied in the laboratory forspectralsignatures identifiable also from satellite images based on their reflectance properties asseen in Fig. 1 because thematic mapping of multispectral images from Landsat-8 satellite cover the visible-to-infrared spectrum of hydrothermal alteration prospects(Ahmed et al., 2014).

Fig. 1. (A) Laboratory spectra of alunite, chlorite, kaolinite, muscovite, calcite, and epidote and (B) Laboratory spectra of limonite, jarosite, hematite and goethite (Clark et al., 1993).

The spectralsignature identification of minerals and assemblage of minerals such as hematite, jarosite and goethite formed by hydrothermal alteration, are used to identify patterns of outflows of hydrothermalsystems, which can allow recognition of mineralized zones (Sabins, 1999; Rajesh, 2004). Spectral signatures of minerals and rocks have the basis on the work done by Hunt (1977) that measured in laboratory the spectra of several differentminerals and rocks(spectralsignature). Spectral signatures obtained by Landsat-8 imagery may be different when compared to the laboratory measurements, due to pixelresolution and the presence of different materialsin one pixel. This Landsat image is a product of electromagnetic spectrumhaving a range of frequencies of electromagnetic radiations and their respectivewavelengthandphotonenergies.The frequency range is divided into separate bands (Table 1) that show the reflectance value of materials due to natural incident light on them. Each band has a wavelength bracket that has the dominant spectral value of certain minerals and this helps to distinguish minerals based on their reflectance values.

Table 1. Landsat-8 OLI and TIRS band characteristics (USGS, 2015)

Zuru in central Kebbi, Nigeria is well known for the existence of gold mineral occurrences primarily associated with hydrothermal processes. Figs. 2 and 3 are images of outcropped quartz vein and artisanal mining pits respectively in the area. The ability to discriminate between hydrothermally altered and unaltered rocks are considerable in mineral exploration studies involving multispectral data. In the region of solar reflected light (0.325 to 2.5 μm), many minerals demonstrate diagnostic absorption features due to vibrational overtones, electronic transition, charge transfer and conduction processes(Sabins, 1999;Rajesh, 2004). Kaolinite and alunite are typical constituents of advanced argillic alteration that exhibit Al-OH of between 2.165 μm and 2.2 μm absorption features (Crosta et al., 2009). Porphyritically-altered rocks typically contain varying amounts of chlorite, epidote and calcite, which exhibit Fe, Mg-O-H andCO₃ of 2.31 to 2.33 μm absorption features. Iron oxide/hydroxide minerals such as jarosite and hematite tend to have spectral absorption features in the visible to middle infrared from 0.4 to 1.1 μm of the electromagnetic spectrum (Crosta et al., 2009). Hydrothermal silica mineralstypically consist of quartz, opal and chalcedony. Thermal Infra-red (TIR) emissivity spectral illustrate that quartz and opal contain a prominent feature in the 9.1 μm region (Crosta et al., 2009).

Fig. 2. Side view of the uplifted quartz vein.

Fig. 3. Photograph of one of the artisanal mining pits

Extensive hydrothermal alteration zones and weathering of the sulphide mineralization within the acid volcanicsrepresent a significant mineral province whose spectral features extend between visible and infrared parts (0.4 – 2.5 μm) (Crosta et al., 2009). The main problem in identification of minerals is the interference from vegetation that also has a strong reflectance in the infrared. However, certain bands are very helpful for distinguishing vegetation from hydroxyl and iron oxides. To this end, satellite images provide a superb synoptic view where the location of old workings, manifestations of hydrothermal wall rock alterations and gross structural features can be delineated.

Severalremote sensing studies have been conducted in the research area using ASTER, Landsat-7 ETM+ and Quickbird data during recent years (Ramadan et al., 2010; Amin et al., 2014). Kusky et al. (2003) processedLandsatTMand radarJERS-I SAR(L-Band) imagery to emphasize structural geology features including folded quartzite ridges and plutonsin Itremo area, central Madagascar. The band ratio classification results were fairly accurate,showing a confusionmatrix accuracy of 89%, which corresponded well with geologic maps of the area. Amin et al. (2014) studied hydrothermal alteration mapping using Landsat-8 data in Sar Cheshmeh copper mining district, SE Iran. The result showed that Landsat-8 bands especially bands 2 and 4 in visible and near infra-red, 6 and 7 in shortwave infra-red, and 10 in thermal infra-red contain useful spectral information for porphyry copper exploration purposes.Thisstudy evaluatesthe usefulness of Landsat-8 OLI bands in mapping hydrothermal alteration zones and lithological units associated with auriferous deposits in the mining districts within the Precambrian Basement Complex of North-western Nigeria.

3. Description of the study area

The Pan-African terrain of north-western Nigeria is part of the vast Late Proterozoic - Early Phanerozoic terrain separating theWestAfrican andCongoCratons. It consists of an older crust in whichArchean (ca. 2700 Ma) and early Proterozoic (ca. 2000 Ma) ages have been recorded but were generally reactivated by the Pan-African event (Kogbe, 1979). The Pan-African event (600 ± 150 Ma) yielded regional metamorphism thereby imposing a generally North – South foliation and bringing about the emplacement of granitoids in the region (Kogbe, 1979). Wright (1985) mentioned that a collision type orogeny has been suggested by involving the Pan-African region and theWestAfrican craton, where a subduction zone dipped eastward beneath the Pan-African region. Deformation and metamorphism followed the continental collision around 660 Ma ago with consequent thickening in the Nigeria region.

The study area lies within the Sokoto sector of the Iullemmeden Basin in North-western Nigeria and is characterized by three lithostratigraphic units namely; the supracrustal schist belts, the gneiss-migmatite complex and the Pan-African granitoids (Fig. 4). The gneiss-migmatite complex yielding Archean and Early-proterozoic isotopic ages, possess generally an amphibolite facies grade of metamorphism (Kogbe, 1979). They show complex structural trends and are extensively invaded by granitoid plutons of PanAfricanmagmatism.The schist belt of Late-Proterozoic age were deposited and metamorphosed together with the basement gneiss-migmatite complex during the Pan-African event.

Fig. 4. Geological map of the area.​​​​​​​

The granitoids are mainly syn-to-late tectonic PanAfrican intrusions of granites, granodiorites and diorites with some gabbros and syenites. They range in size from small sub-circular, crosscutting stocks to large elongated concordant batholithic bodies emplaced into both the gneiss-migmatite and supra-crustal rocks during or just after the main phase of Pan-African deformation (Wright, 1985). Structurally, the study area is located within the Precambrian Basement Complex in a highly strained zone where the host rocks are highly foliated, deformed and severely affected by brittle-ductile shear zone (Odeyemi, 1981).The dominant tectonic fabric in the study area is a North – South steeply dipping phyllitic to slaty cleavage, which is mainly axial planar to tight isoclinal folds. Auriferous  mineralization in the study area occurs in a variety of host rocks but is generally related to the supra-crustal schist belts. Several locations of gold mineralization are known within the study area, however, new auriferous occurrence associated with large alteration zones are sought in this study.

4. Data

1) Image Acquisition and Pre-processing

A Landsat-8 OLI/TIRS scene (Path 190, Row 052) dated 4th November (0% cloud cover) was acquired for the year 2017 from the US Geological Survey archive. The Level 1 standard terrain corrected image with image quality of 9 was processed using the Environment for Visualizing Images (ENVI) version 5.1 software and the Environment Systems Research Institute (ESRI) ArcGIS version-10 software. The Landsat-8 image was converted to calibrated variance and then further converted to surface reflectance using the Fast Line-in-sight Atmospheric Analysis Spectral Hypercube (FLAASH) algorithmin the ENVIsoftware for the purpose of atmospheric correction. FLAASH parameters were specific to the Landsat-8 OLI sensor. Finally, the image was spatially subset to the mining province around the Kebbi State eastern border.

2) Image Processing

Several image processing techniques were applied on the data in order to enhance multispectral characteristics of the study area in terms of hydrothermal alteration mapping. The preparation and enhancement of the image was done by contrast stretching method.This processing method is designed to transform multispectral image data format into an image display that either increases contrast between interesting targets and the background or yields information about the composition of certain pixels in the image (Khalid et al., 2014).Colour composite, band ratio and minimum noise fraction (MNF) were also used in this study.

5. Methods

1) Colour composite

Forthe purpose of highlighting geologic features and accentuating textural characteristics at regional scale, a combination of three bands is assigned the natural colours of Red-Green-Blue (RGB) on the Landsat-8 OLI image. In this study, true colour composite (Fig. 5) was generated for the purpose of lithological discrimination. This image which displays band 4, 3 and 2 in RGB composite was produced by importing individual 8-bit gray-scale surface reflectance bands of the multispectral data into ENVI workspace. The operation was then followed by calculating the Optimal Index Factor to select the most informative colour composite image using ILWIS software. Having all the bands ranked according to their respective spectral quality, the brightest bands were then combined and displayed in RGB using the ENVI software. Bands 4, 3 and 2 were chosen because their composite help display the lithological attributes of the area.

Fig. 5. True colour combination RGB for bands 4, 3 and 2​​​​​​​

2) Band ratio

It is possible to divide the digital number(DN) value of one band by the DN of another band in an operation known as band ratio. This procedure isinstrumental in highlighting certain features or rocks that cannot be seen in the individual raw bands. This operation involvesthe selection ofthe band with high reflectance for a mineral as the numerator while setting any other band with high absorption of that same mineral as the denominator in the band ratio interface within the ENVI program. Three distinct ratios were performed namely; \({4 \over 2},\ {5\over 6},\ {6\over 7}\) which were finally assigned to RGB respectively in Fig. 6. The ratio is useful for mapping iron oxides because it has absorption in the blue band region and high reflectance in the red band region.The ratio was used for mapping ferrous-ion minerals due to the high reflectance of those minerals in band 5 and absorption featuresin band 6. Fig. 7 is a stretched image resulting from filter operations performed on the band ratio image. The filtering operation is a contrast stretching procedure which is a relatively straightforward way of enhancing difference in the tone and texture of the image. The contrast stretch used is the histogram equalization which is applied to mask or enhance features with distinct DN ranges, such as low DN associated with water.This operation is performed on the ENVI software for the purpose of reducing dimensionality on the band ratio image as a pre-step towards minimum noise fraction.

Fig. 6. Landsat-8 band ratio (\(\frac{4}{2}\), \(\frac{5}{6}\), \(\frac{6}{7}\)) for RGB.

Fig. 7. Stretched band ratio image of the study area.​​​​​​​

3) Minimum noise fraction (MNF)

This method which was developed by Green et al. (1988) is similar to principal component orthogonalization rotation that results in components ordered in increasing rank of random noise rather than decreasing rank of variance. So, the MNF transform is used to determine the inherent dimensionality ofimage data, to segregate and equalise the noise in the data, and to reduce the computational requirements in the case of subsequent processing. This technique was used in Fig. 8 to identify hydrothermally altered rocks in the image with the help of pan-sharpening technique. All the Landsat-8 OLI bands combined to produce Fig. 8 are of 30 meters resolution. However, the image can be expanded to 15 meters resolution when the panchromatic band is clipped to the data. The process of clipping the panchromatic band 8 to a multispectral mage is feasible using the spectral analyst tool in the ENVI 5.1 platform. The process of clipping the panchromatic band to a multispectral image of lower resolution is known as pan-sharpening technique (Amin et al., 2014).MNF was performed in this study to extract individualmineralspeciesfromamixed pixel spectrum, in theory providing geologist with the capability to map mineral surface composition. Although this image processing method has been mostly applied on hyperspectral data (Amin et al., 2014), they can also be applicable logically to multispectral data. With these image processing methods, pixelsthat havemixed spectralsignatures will be extractable and can be separated from the undesirable background. Thus, mineral abundance maps can be produced free of diluting effects of surrounding environment. MNF approach for analysis ofmultispectral data used in this work isimplementable and documented within the ENVIsoftware system.The analysis approach consists of spectral compression, noise suppression and dimensionality reduction using the Minimum Noise Fraction (MNF) transformation (Amin et al., 2014). MNF component images show steadily decreasing image quality with increasing band number, so images with higher eigenvalues contain higher spectral information (Gupta, 2003). RGB colour combination image was assigned to three high eigenvalues MNF transformed bands.

Fig. 8. Band 1, 2 and 3 from MNF result​​​​​​​

6. Discussion

It is often necessary to display several datasets together to allow for a wholesome interpretation.Also, different presentation of the same dataset, selected to emphasize different characteristics of the data can be combined. For the RGB model, the most common applications are the display of multichannel data such as the different spectral bands in remote sensing data. In mapping the hydrothermal alteration zone, a geologic research was done to ensure that the gold mineralization in the area, which hasled to artisanal pit mining over the years, was a product of hydrothermal fluid deposition. The known gold-bearing quartz veins (Figs. 6-8) are within granite rocksthat are altered, and are result of chemical interactions with hydrothermal fluids responsible for the gold mineralization. The alteration forms a halo around these quartz vein areas providing an exploration target considerably broader than the deposit.

The processed multispectral data highlights areas where assemblage of alterationminerals asiron-bearing minerals and hydrated sulphates occur, discriminating altered from unaltered rocks. In this study, mapping iron oxide haloes was carried out by ratioing band 4 over 2 because iron oxide/hydroxide minerals such as hematite, jarosite and limonite, and sulphuric minerals have high reflectance within 0.63 – 0.69 μm (band 4) and high absorption within 0.45 – 0.52 μm (band 2). Clay and carbonate minerals have absorption features from 2.1 – 2.4 μm (band 7) and reflectance from 1.55 – 1.75 μm (band 6) in Landsat-8 OLI data (Han et al., 2015). Mineralssuch as alunite, and clay mineralssuch asillite, kaolinite and montmorillonite have distinctive absorption features at 2.20 μm and low absorption at 1.6 μm

Band ratios derived from image spectra (\({4 \over 2},\ {5\over 6},\ {6\over 7}\)) and the contrast stretched image in RGB allows the identification of altered rocks, lithological units and vegetation (Figs. 6 and 7 respectively). The alteration mineralized zones (hydrothermal altered rocks) are outlined in the image where they appear as red colour around known and mined gold deposits which are visible in Figs. 6, 7 and 8. From Fig. 6, the alteration appears dominant along a geologic structure trending in the NW – SE direction and extending a distance of about 5 km in length. Since most minerals in veins are mostly hosted in faults, this geologic structure bordering the alteration is of utmost interest for mineralization potentiality.

The boundary between sedimentary (conglomerates) and basement rocks (Precambrian volcanic rocks) are also delimited in the resultant images. From the MNF transformed image, reddish areas are dominated by limonite minerals (hematite, goethite and jarosite), the whitish areas are dominated by Fe-Mg silicate minerals like olivine, pyroxenes and amphiboles while the purple areas are dominated by clay minerals, micas and talco-carbonates. Also, from visual analysis of Fig. 7, a contrast-enhanced image shows vegetation in light orange, granitoids in rose-to-red colour and metasediments in blue to purple colour. However, this image does not distinguish so well between lithologies because of the topography and other interferences, being difficult to define contacts. Band ratios reveal a good technique in lithologicalmapping,since thismethod cancels illumination differences from topography, sun shadowing, etc. The band ratio (\({4 \over 2},\ {5\over 6},\ {6\over 7}\) )showed altered granitoids in red colouration and dark blue for metasediments as represented in Fig. 9.

Fig. 9. Geologic representation of the band ratio image.​​​​​​​

7. Conclusion

This study was conducted using Landsat-8 OLI imagery to determine the spectral characteristics of the hydrothermal alteration zone around the gold mining district in Zuru, North-western Nigeria. The resulting images from the applied remote sensing techniques including colour composite, band ratio and minimum noise fraction were effective in mapping zones of altered rock units.TheFLAASHmethod of atmospheric correction produced the surface reflectance image that was used to process band ratio for mineral prospecting in the study area. The outcome of this study indicates high potentiality of untapped gold mineralization due to the extensive halo mapped around the NW – SE trending fault in the area. However, detailed survey is needed so asto quantitatively evaluate and estimate the resource and reserve in the area.