Remote Sensing of Environment 148 (2014) 124–133

Contents lists available at ScienceDirect

Remote Sensing of Environment

j ourna l homepage: www.e lsev ie r .com/ locate / rse

Potential of ESA's Sentinel-2 for geological applications

F.D. van der Meer ⁎, H.M.A. van der Werff, F.J.A. van RuitenbeekDepartment of Earth Systems Analysis, Faculty for Geo-Information Science and Earth Observation, University of Twente, Enschede, 7500 AE, The Netherlands

⁎ Corresponding author.E-mail address: f.d.vandermeer@utwente.nl (F.D. van

http://dx.doi.org/10.1016/j.rse.2014.03.0220034-4257/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 December 2013Received in revised form 21 March 2014Accepted 25 March 2014Available online xxxx

Keywords:Sentinel-2SimulationASTERGeologyMineralogyAlterationBand ratios

Sentinel-2 is ESA's medium spatial resolution (10–60 m) super-spectral instrument aimed at ensuring datacontinuity for global land surfacemonitoring of Landsat and SPOT. Several simulation studies have been conductedthat show the potential of Sentinel-2 for estimating biophysical and biochemical parameters such as leaf areaindex, chlorophyll and nitrogen, and spectral products such as the red edge position and NDVI time series,providing data continuity for a number of other operational sensors. This paper aimsat simulating Sentinel-2 prod-ucts that are relevant to the geology and soil science community and we compare these to well established bandratio products fromASTER. As the basis for the simulationwe use airborne hyperspectral imagery from theHyMAPsensor which were spectrally and spatially resampled to the resolutions of ASTER and Sentinel-2 using actualspectral response functions and pixel aggregate, respectively. The simulated image products demonstrate a goodcorrespondence between ASTER and Sentinel-2 VNIR and SWIR bands. A number of band ratios are proposedfor Sentinel-2 to derive the following products: ferric iron, ferrous iron, laterite, gossan, ferrous silicate and ferricoxides. These are compared to established band ratios of real ASTER data as well as simulated (from HyMAP)ASTER data and they correlate favourably. Investigating the spatial patterns reveals that there is a good matchbetween the proposed Sentinel-2 band ratio products and those of ASTER. The resulting band ratio products arecompared to the local geologic map of the imaged hydrothermal area (the Rodalquilar mining area, Cabo deGata volcanic field, SE Spain). It is shown that they support the existing conceptual geologic model of theepithermal deposit.

© 2014 Elsevier Inc. All rights reserved.

1. Introduction

The European Commission (EC) and the European Space Agency(ESA) in a consorted effort to address issues related to the monitoringof the environment have established the European Earth Observationprogramme Copernicus, previously known as the Global Monitoringfor Environment and Security (GMES). The space component is anumber of satellite missions conveniently labelled Sentinel 1 up to 5(Berger, Moreno, Johannessen, Levelt, & Hanssen, 2012). The Sentinel-2mission is a medium spatial resolution, super-spectral instrumentwith a large field of view (290 km), a high revisit capability (5 dayswith two satellites), a high resolution (10 m, 20 m and 60 m) and amoderately large band set (13 spectral bands) aiming to provide globalland coverage. The mission ensures continuity of the SPOT and Landsatmissions and provides operational products such as land-cover maps,land change detection maps, and geochemical/physical variables(Drusch et al., 2012). In preparation of the Sentinel-2 image productsto become available a number of simulation studies have been conductedto explore the potential of Sentinel-2 for a variety of land surfaceparameter estimations including Leaf Area Index (Atzberger & Richter,

der Meer).

2012; Richter, Atzberger, Vuolo, Weihs, & D'Urso, 2009; Richter, Hank,Vuolo, Mauser, & D'Urso, 2012), chlorophyll and nitrogen epclevers2013, biophysical parameters (Frampton, Dash, Watmough, & Milton,2013; Verrelst et al., 2012), and the red edge position (Clevers &Gitelson, 2013; Delegido et al., 2013). In addition, a study has beenperformed on the NDVI time series continuity of Sentinel-2 for a numberof other operational sensors including the National Oceanic andAtmospheric Administration (NOAA)/Advanced Very High ResolutionRadiometer (AVHRR9), Landsat 7 Enhanced Thematic Mapper Plus(ETM+), Satellite Pour l'Observation de la Terre VEGETATION1 (VGT1),the Moderate Resolution Imaging Spectroradiometer (MODIS), andMedium Resolution Imaging Spectrometer (MERIS) (D'Odorico,Gonsamo, Damm, & Schaepman, 2013). Outside the realm of physicalvariables, the application of Sentinel-2 has been tested for water qualitystudies (Salama, Radwan, & van der Velde, 2012) and for mapping ofcoral reef (Hedley, Roelfsema, Koetz, & Phinn, 2012). A comparison ofSentinel-2 and ASTER bands with concurrent satellite sensors is shownin Fig. 1. Based on the spatial and spectral characteristics proposed forSentinel-2 and its similarity to Landsat ETM, SPOT and to a lesser extentASTER, there seems to be great potential of developing products thatcan be used for the geology community. However, currently this isoutside the aim of the Sentinel-2 mission and no study has beenconducted to address the use and usefulness of Sentinel-2 for geology.It is precisely this we aim to achieve with this paper.

HyMap

ASTER

WorldView 3

Sentinel 2

Landsat 8

SPOT 5

0 500 1000 1500 2000 2500Wavelength (nm)

0.0

0.2

0.4

0.6

0.8

1.0

Tra

nsm

ittan

ce

Fig. 1.Proposed bands of the Sentinel-2 sensor, in comparison to the operational SPOT, Landsat 8, ASTER&HyMAPsensors and theproposed bands of theWorldview3 sensor. Atmospherictransmittance is plotted on the Y-axis.

125F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

2. Sentinel-2 mission overview

The Sentinel missions are designed to provide routine observationsfor operational Copernicus services and to provide data continuity ofERS, ENVISAT and multispectral missions such as SPOT, Landsat,ASTER, etc. Sentinel-2 carries a super-spectral instrument with a sun-synchronous 786 km orbit that allows covering all the land surfacesand coastal waters between −56 and +84 degrees latitude with a290 km swath width at a 10 day revisit time at the equator (5 dayrevisit time at the equator based on 2 satellites which will fly simulta-neously) The mission is primarily designed for land cover and associat-ed applications in change-detection mapping. A variety of variables(e.g., leaf chlorophyll content, leaf water content, leaf area index; seeTable 1) are proposed for Sentinel-2 as inputs to earth system models.The provision of long-term data with the Sentinel missions is aprerequisite to allow better understanding of climate variability. Inaddition, long-term data continuity is a pre-requisite for fosteringdownstream services that Sentinel-2 provides for example in the areasof European land use/land cover changes, food security early warningand global change issues. Sentinel-2 will deliver level 2 geochemical/physical data products and level 3 biophysical variables which arephysically consistent and reproducible which implies that with limitedempirical (and ground-based) studies these data products can beincorporated in global process models.

Table 1Spectral position (λ) and bandwidth (Δλ), spatial resolution, heritage and purpose of Sentinel-2(2012) and European Space Agency (2010).

# λ (nm) Δλ (nm) Res (m) Heritage Purpose

1 443 20 60 MODISa, ALIa, LS8a Atm. correction (2 490 65 10 MERISa,LS8, LS7 Vegetation senes3 560 35 10 MERIS,LS8, LS7, SPOT5a Green peak, sens4 665 30 10 MERIS,LS8, LS7 Max. chlorophyll5 705 15 20 MERIS Red edge positio6 740 15 20 MERIS Red edge positio7 783 20 20 MERIS, ALI LAIb; edge of the8 842 115 10 Landsat, SPOT5 LAI8a 865 20 20 MERIS, ALI, Landsat 8 NIR plateau, sens

retrieval of aeros9 945 20 60 MODIS, MERIS Atm. correction (10 1375 30 60 MODIS, LS8 Atm. correction (11 1610 90 20 LS8,SPOT5, LS7 Sensitive to ligni12 2190 180 20 LS8, LS7 Assessment of M

erosion; distincti

a MODIS = Moderate Resolution Imaging Spectrometer, ALI = Advanced Land Imager, LS7SPOT = Satellite Pour l'Observation de la Terre.

b LAI = Leaf Area Index.

3. Geologic remote sensing data products

Geologic remote sensing is based on laboratory spectral analysis ofminerals and rocks conducted by Graham Hunt and John Salisbury(Cooper, Salisbury, Killen, & Potter, 2002; Hunt, 1977; Salisbury,Walter, & Vergo, 1989) in the 1970s through 1980s. A number of articlesreview geologic remote sensing with emphasis on mineral exploration(Sabins, 1999), hyperspectral geology (Cloutis, 1996), mineral resourcemapping (Rajesh, 2004) and combined multi- and hyperspectral analy-sis (van der Meer et al., 2012). In general, the geology community usesmultispectral remote sensing data, similar to Sentinel-2, tomap litholo-gies, structures (fault, folds, fractures) and surface mineralogy whereoften band ratios serve as proxies for mineral assemblages or individualmineral groups. From sedimentology and metamorphic petrology, thekey minerals to describe mineralized zones are known and if these are“spectrally active” they potentially can be mapped with remote sensing(for examples, see Table 2).

In the 1980s Landsat multispectral scanner (MSS) were deployedfor the first time to derive iron oxide maps (Goetz & Rowan, 1981).With the advent of the LandsatThematic Mapper (TM), providingmore spectral details at that time, band ratios were used to separateargillic from non-argillic materials (using TM 5/7) and for mappingferric/ferrous oxides (using TM 3/1). Landsat TM data have beenused widely by the geologic remote sensing community for a wide

bands, according to the Copernicus derived user requirements.Modified after Drusch et al.

aerosol scattering)cing, carotenoid, browning and soil background; atm. correction (aerosol scattering)itive to total chlorophyll in vegetationabsorptionn; consolidation of atmospheric corrections/fluorescence baseline.n; atmospheric correction; retrieval of aerosol loadNIR plateau

itive to total chlorophyll, biomass, LAI and protein; water vapour absorption reference;ol load and typewater vapour absorption)detection of thin cirrus)n, starch and forest above ground biomass; snow/ice/cloud separationediterranean vegetation conditions; distinction of clay soils for monitoring of soilon between live biomass, dead biomass and soil, e.g. for burn scars mapping

= Landsat TM, LS8 = Landsat 8, MERIS = MEdium Resolution Imaging Spectrometer,

Table 2Summary of the main spectrally active minerals in relation to different alteration stylesand environments of formation.

Environment of formation Main spectrally active alteration minerals

High sulphidation epithermal Alunite, pyrophyllite, dickite, kaolinite, diaspore,zunyite, smectite, illite

Low sulphidation epithermal Sericite, illite, smectite, chlorite, carbonatePorphyry: Cu, Cu–Au Biotite, anhydrite, chlorite, sericite, pyrophyllite,

zeolite, smectite, carbonate, tourmalineCarlin-type Illite, dickite, kaoliniteVolcanogenic massive sulphide Sericite, chlorite, chloritoid, carbonates,

anhydrite, gypsum, amphiboleArchean Lode Gold Carbonate, talc, tremolite, muscovite, paragoniteCalcic skarn Garnet, clinopyroxene, wollastonite, actinoliteRetrograde skarn Calcite, chlorite, hematite, illiteMagnesium skarn Forsterite, serpentine-talc, magnetite, calcite

Table 3Sentinel-2 band ratios as an analogue of ASTER band ratios, used as proxies formineralogy.Modified after Kalinowski & Oliver (2004). The table is limited to ratios that fall in thewavelength range of Sentinel-2.

Feature ASTER Sentinel-2

IronFerric Iron, Fe3+ 2/1 4/3Ferrous Iron, Fe2+ 5/3 + 1/2 12/8 + 3/4Laterite 4/5 11/12a

Gossan 4/2 11/4Ferrous silicates (Biotite, chloride,amphibole)

5/4 12/11a

Ferric oxides 4/3 11/8

Carbonates/Mafic mineralsCarbonate/Chlorite/Epidote (7 + 9)/8 –

Epidote/Chlorite/Amphibole (6 + 9)/(7 + 8) –

Amphibole/MgOH (6 + 9)/8 –

Amphibole 6/8 –

Dolomite (6 + 8)/7 –

SilicatesSericite/Muscovite/Illite/Smectite (5 + 7)/6 –

Alunite, Kaolinite, Pyrophyllite (4 + 6)/5 –

Phengitic 5/6 –

Muscovite 7/6 –

Kaolinite 7/5 –

Clay (5 × 7)/62 –

Alteration 4/5 11/12a

Host rock 5/6 –

OtherVegetation 3/2 8/4NDVI (3 − 2)/(3 + 2) (8 − 4)/(8 + 4)

a ASTER bands 5–7 fall within band 12 of Sentinel-2.

126 F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

range of applications including geologic (Fraser, Huggins, Rees, &Cleverly, 1997; Schetselaar, Chung, & Kim, 2000), lithologic (Gad &Kusky, 2006) and structural (Boccaletti et al., 1998; Yesou, Besnus,& Rolet, 1993) mapping, used to characterize volcanic areas interms of mineral potential (Abdelsalam, Stern, & Berhane, 2000;Ferrier, White, Griffiths, Bryant, & Stefouli, 2002; Sabins, 1999).

Since 2000, the Advanced Spaceborne Thermal Emission and Reflec-tance Radiometer, ASTER (Abrams, 2000; Abrams & Hook, 1995;Yamaguchi, Kahle, Tsu, Kawakami, & Pniel, 1998), has been the “workhorse” for the geologic remote sensing community allowing mappingof surface mineralogy. Several ASTER band ratios serve as proxies formineralogy (Cudahy & Hewson, 2002), including some that overlapwith the spectral coverage of Sentinel-2 (see Table 3). In particular theASTER SWIR bands, due to their relative fine spectral resolution, allowmapping of surface mineralogy (Mars & Rowan, 2010).

There is ample literature on the use of ASTER-derived products in ge-ology with emphasis on lithologic mapping (Gomez, Delacourt,Allemand, Ledru, & Wackerle, 2005; Haselwimmer, Riley, & Liu, 2011;Khan, Mahmood, & Casey, 2007; Li, Long, & Liu, 2007; Qiu,Abdelsalam, & Thakkar, 2006; Rowan & Mars, 2003), mineral mappingrelated to active geothermal systems (Vaughan, Hook, Calvin, &Taranik, 2005), and fossil hydrothermal systems (Carranza, vanRuitenbeek, Hecker, van der Meijde, & van der Meer, 2008; Hubbard,Crowley, & Zimbelman, 2003; Mars & Rowan, 2006, 2010; Yamaguchi& Naito, 2003; Zhang, Pamer, & Duke, 2007). Several studies of remotesensing in geology deal with invariant in time (e.g., mineral mappingto aid in mining, lithologic mapping, etc.) where archived satellitedata would be sufficient for geologic mapping purposes. However,geologists increasingly study dynamic phenomena over space andtime (e.g., active geothermal systems, oil and gas seeps, coastal environ-ments, etc.), which requires longer time series of inter/cross calibrateddata products.

A number of studies focused on simulating sensor responses for geo-logic mapping using proxies from airborne data or airborne prototypesimulators. For example, MASTER (theMODIS and ASTER airborne sim-ulator) was a test bed for ASTER to investigate its potential for geologicmapping (Hook,Myers, Thome, Fitzgerald, & Kahle, 2001).More recent-ly, data from the hydrothermal area at Cuprite Nevada was used to as-sess the potential of the new Worldview 3 commercial system,collecting both high spatial resolution as well as multispectral SWIRdata (Kruse & Perry, 2013). Likewise, mineral mapping capabilities ofthe upcoming EnMAP (Stuffler et al., 2007) spaceborne hyperspectralsystem were tested using simulated data sets (Rodgers & Oliver, 2007).

4. Methods

The capabilities of Sentinel-2 for geological mapping were assessedby simulating a dataset from a reflectance-at-surface airbornehyperspectral image, using Sentinel-2 band specifications (Table 1).The image bands and image products created with this dataset were

compared to those derived from an ASTER radiance-at-sensor datasetacquired over the same area, as well as a simulated ASTER dataset creat-ed in the sameway as the Sentinel-2 dataset.We focused on capabilitiesfor discriminating surface mineralogy and geological units separately.

As input to the Sentinel-2 simulation, we use airborne hyperspectraldata from the HyMAP sensor acquired over an hydrothermal alterationarea located in the Sierra del Cabo de Gata in the south-eastern part ofSpain (Rytuba et al., 1990). The spectral coverage of HyMAP is nearlycontinuous in the VNIR and SWIR wavelength regions with small gapsin themiddle of the 1.4 and 1.9 μm atmospheric water bands. The spec-tral coverage, bandwidth and spectral sampling interval for the variousdetectors are: VIS 0.45–0.89 μm, 15–16 nm, 15 nm; NIR 0.89–1.35 μm,15–16 nm, 15 nm; SWIR1 1.40–1.80 μm, 15–16 nm, 13 nm; SWIR21.95–2.48 μm, 18–20 nm, 17 nm. The spatial configuration of the instru-ment gives an Instantaneous Field Of View (IFOV) of 2.5 mrad alongtrack and 2.0 mrad across track, usually resulting in a pixel size in theorder of 5 × 5 m.

TheCabodeGata volcanicfield (Fig. 2) consists of calc-alkaline volcanicrocks (andesites and rhyolites) of late Tertiary agewhich have been exten-sively altered to form an assemblage of metamorphic minerals from highto low temperature as: silica, alunite, kaolinite, montmorillonite andchlorite. The geology, geochemistry and mineralization of the area aredocument in Arribas et al. (1995) while a number of remote sensingstudies have been conducted at the site (Bedini, van der Meer, & vanRuitenbeek, 2009; Debba, Carranza, Stein, & van der Meer, 2009; Van derMeer, 2006; van der Werff et al., 2007).

The HyMAP data were acquired on May 18, 2004 by the GermanAerospace Center (DLR) and processed to reflectance-at-surface valuesat a 4 m spatial sampling interval. This included geometric correctionwith PARGE (Schlapfer & Richter, 2002) and a 5 m interval DEM,followed by an atmospheric correction with the ATCOR 4 (AtmosphericCORrection) software (Richter & Schlapfer, 2002). A spatial subset of theHyMAP imagery was chosen as an experimental dataset for this study.This subset consists of 600 × 600 pixels and covers approximately6 km2 of area (indicated by a dashed box in inset B of Fig. 2).

Almeria

Mediterranean Sea

Cabo de GataVolcanic Field

Cartagena

Spain

Betics

Rif

Lomillacaldera

Los FrailescalderaSan

Jose

Caldera marginLate Miocenesedimentary rocksMiddle Miocenevolcanic rocks

Pliocene and Miocene volcanic rocks

Mesozoic and Paleozoic metamorphic rocks

Ore deposits

2 05’W0 2 00’W0

20 km

3 km

A

N

B

Rodalquilarcaldera

Rodalquilar

Study area

Fig. 2. Location of theRodalquilar gold deposit in the volcanic belt of SE Spain (inset A),modified afterArribas et al. (1995) and Rytuba et al. (1990). The location of the approximately 6 km2

image dataset used in this study is indicated by the dashed box in inset B.

127F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

The HyMAP data were spectrally convolved to the lower resolutionsof the Sentinel-2 and ASTER bands using spectral response functions(Fig. 3). The 5 ASTER bands in the TIR fall outside the wavelengthrange of Sentinel-2 and were ignored in this study. After spectralresampling, the data were spatially resampled from the 4 m HyMAPpixel resolution to the 10, 20 & 60 m resolutions of Sentinel-2 bandsand 15 & 30 m ASTER bands using pixel aggregation (neighbourhoodaveraging). The bands of the simulated datasets were compared withscatterplots, after conversion to a 5 m sampling interval with nearestneighbour resampling to ensure an equal amount of datapoints

400 500 600 70Wavelen

0.0

0.2

0.4

0.6

0.8

1.0

Nor

mal

ized

Spe

ctra

l Res

pons

e

1 2 3 4

21

1400 1600 1800Wavelen

0.0

0.2

0.4

0.6

0.8

1.0

Nor

mal

ized

Spe

ctra

l Res

pons

e

10 11

4

Fig. 3. Spectral response functions of the Sentinel-2 (shaded) and ASTER (l

(Fig. 4). In Fig. 4, we retained the original pixel size of HyMAP andonly spectrally resampled the HyMAP bands to the band settings ofASTER and Sentinel-2. It is obvious from Fig. 4, that there is a goodcorrespondence between the Sentinel-2 and ASTER bands and thatHyMAP is able to spectrally mimic and simulate the broad Sentinel-2as well as ASTER bands.

Spatial degradation follows two steps. The first step requiresmodelling of the transfer function between the initial data and thedesired data and deriving a spatial filter that allows simulation ofthe coarse spatial resolution imagery. The second step involves

0 800 900 1000gth (nm)

5 6 7 8 8a 9

n3

2000 2200 2400gth (nm)

12

5 6 7 8 9

ine) sensors, in the VNIR (top) and SWIR (bottom) wavelength range.

0.0 0.2 0.4 0.6 0.8 1.0S2 band 3

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

1

0.0 0.2 0.4 0.6 0.8 1.0S2 band 4

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

2

0.0 0.2 0.4 0.6 0.8 1.0S2 band 8

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

3

0.0 0.2 0.4 0.6 0.8 1.0S2 band 11

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

4

0.0 0.2 0.4 0.6 0.8 1.0S2 band 12

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

5

0.0 0.2 0.4 0.6 0.8 1.0S2 band 12

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

6

Fig. 4. Scatterplots between simulated Sentinel-2 and simulated ASTER bands, obtained by spectral resampling of HyMAP reflectance data.

128 F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

resampling to the desired pixel size. In the absence of the sensor'sPoint Spread Function (PSF), a square moving average filter of thedesired spatial resolution was used. This approach introduces asmall error, as the bit of information present in the areaneighbouring a pixel in a PSF is now excluded. The bands of the sim-ulated datasets were compared with scatterplots, after conversion to

0.0 0.2 0.4 0.6 0.8 1.0S2 band 3

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

1

0.0 0.2 0.4S2 b

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

2

0.0 0.2 0.4 0.6 0.8 1.0S2 band 11

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

4

0.0 0.2 0.4S2 ba

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

5

Fig. 5. Scatterplots between simulated Sentinel-2 and simulated ASTER bands

a 5m sampling interval with nearest neighbour resampling to ensurean equal amount of datapoints (Fig. 5)

Not taken into account is the lossy compression used onboardSentinel-2. We expect that any impacts on data quality will be in thenoise level over natural scenes and therefore have no impact on studyresults.

0.6 0.8 1.0and 4

0.0 0.2 0.4 0.6 0.8 1.0S2 band 8

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

3

0.6 0.8 1.0nd 12

0.0 0.2 0.4 0.6 0.8 1.0S2 band 12

0.0

0.2

0.4

0.6

0.8

1.0

AS

T b

and

6

, obtained by spectral and spatial resampling of HyMAP reflectance data.

A Field photo

B HyMap true color composite

C Geologic interpretation

D HyMap clay band ratios

Fig. 6. Overview of the study area and the HyMAP airborne hyperspectral image used for data simulation. Top left: Field photograph of the Los Tollos volcanic center near the town ofRodalquilar in the Cabo de Gata National Park (Almeria Province, SE Spain); an area of extensive hydrothermal alteration. Bottom left: 3D perspective with a natural color compositeimage derived fromHyMAP. Top right: Interpretation of the geology of the image area, modified after Arribas et al. (1995). Bottom right: HyMAP band ratio image showing hydrothermalalteration mineralogy. The reddish colors are associated with (high-sulfidation epithermal) gold mineralization. Red colors indicate intense alteration of rocks and the presence ofalteration minerals such as alunite and kaolinite. Green colors are related to less intense alteration at lower temperatures, and are predominantly associated with the mineral illite.Blue colors are unaltered volcanic and sedimentary rocks. Viewing direction is east; width of floodplain is approximately 500 m. Vertical exaggeration of the image products is 2. Pixelsize of the 600 × 600 pixels HyMAP image is 4 × 4 m. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

0.0 0.5 1.0 1.5 2.0 2.5 3.0

S2 ferric iron

0.0

0.5

1.0

1.5

2.0

2.5

3.0

AS

T fe

rric

iron

0.0 0.5 1.0 1.5 2.0 2.5 3.0

S2 ferric oxides

0.0

0.5

1.0

1.5

2.0

2.5

3.0

AS

T fe

rric

oxi

des

0 1 2 3 4 5

S2 gossan

0

1

2

3

4

5

AS

T g

ossa

n

0.0 0.5 1.0 1.5 2.0 2.5 3.0S2 ferrous iron

0.0

0.5

1.0

1.5

2.0

2.5

3.0

AS

T fe

rrou

s iro

n

0.0 0.2 0.4 0.6 0.8 1.0S2 ferrous silicates

0.0

0.2

0.4

0.6

0.8

1.0

AS

T fe

rrou

s si

licat

es

0.0 0.2 0.4 0.6 0.8 1.0S2 ndvi

0.0

0.2

0.4

0.6

0.8

1.0

AS

T n

dvi

Fig. 7. Scatterplots between simulated Sentinel-2 and simulated ASTER band ratio products.

129F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

130 F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

In the remainder of this paper, we first assess the potential capabil-ities of Sentinel-2 to provide data continuity for selected band ratioproducts of ASTER and subsequently we investigate the potential ofSentinel-2 for geologicmapping based on the identification of alterationmineral assemblages.

Sentinel-2 ASTER (simu) ASTER (real)

Ferr

icIr

onFe

rrou

siro

ns

5. Discussion

We used a subset of an ASTER and HyMAP scene covering a geolog-ically varied terrain with marine limestones in the northern part of theimage, hillslopes formed by highly alterated volcanic tuffs containingsulfates and iron oxides in the middle part, and a southern alluvialplanewith erosion products from the surrounding hills formed by al-tered volcanic rocks. This study area is shown in 3D perspective inFig. 6.

Although the design of Sentinel-2 shows similarity with the Landsatseries of sensors, we chose to show the geological capabilities ofSentinel-2 by comparing it to ASTER band ratio products, given thewidespread use of ASTER for geological studies. Specifically we empha-size the potential of Sentinel-2 to provide data continuity for ASTER interms of generating physically consistent and reproducible dataproducts from which surface mineralogy can be distinguished.

Band ratios commonly used for mineral mapping with ASTER wereadapted for Sentinel-2 by selecting matching wavelength ranges. Theband ratioswere then applied to both a simulated (fromHyMAP) datasetand a real ASTER image dataset. The influence of vegetation wassuppressed by masking pixels with NDVI values over 3.0, a thresholdset after visual inspection of the NDVI image and accompanyinghistogram. The masked products created with both simulated datasetswere compared by means of scatterplots after conversion to a 5 msampling interval with nearest neighbour resampling. The maskedimage products made with the simulated datasets and the real ASTERdataset were compared by displaying them in a color scale.

Ferr

icox

ide

Late

rite

Gos

san

Ferr

ous

silic

ates

Fig. 8. Images of band ratio products using simulated Sentinel-2, simulated ASTER and realASTER datasets. The images are masked for vegetation and subsequently colorized( ) in a 2–98% linear stretch. The imaged area is 2.5 × 2.5 km. (For interpretationof the references to color in this figure legend, the reader is referred to the web versionof this article.)

5.1. Evaluation of Sentinel-2 for ratio mapping

We spectrally and spatially resampled the HyMAP image to theASTER and Sentinel-2 band definitions. A comparison is shown inFigs. 4 & 5, where we plot bands that are sensitive in the same wave-length range for the two sensor systems. The apparent scatter (noise)in Fig. 5 is mostly an effect of the difference in spatial resolutionbetween ASTER and Sentinel-2. This noise relates to differences in sys-tem design rather than noise resulting from atmospheric influences,sun angle and scene illumination; however it should be noted that theretrieved relationships are linear, which is important when usingthese as input to ratio images.

Using the simulated ASTER and Sentinel-2 (spectrally and spatiallyresampled) we produced a range of geologic ratio images followingthe band definition in Table 3. A good correspondence between theASTER and Sentinel-2 ratios for ferric/ferrous iron, ferric oxides, ferroussilicates, gossan and NDVI although there is a minimal over predictionfor ASTER in the ferric iron ratio (Fig. 7).

The corresponding ratio images belonging to the scatterplots inFig. 7 are shown in Fig. 8. In Fig. 8, we present ratio images for thespatially/spectrally resampled (and vegetation masked) aster andSentinel-2 images and compare them directly to a real ASTER sceneacquired in June 2002 from the same area. The simulated Sentinel-2 and ASTER products are spatially very comparable. When compar-ing with the real ASTER image the comparison in most cases is goodexcept for the ferrous iron and ferrous silicate product. The ASTERferrous iron product and to a lesser extend the ferrous silicate prod-uct overestimates the iron member in the north western corner ofthe image subset. This might be a result of a change in vegetationgreenness over time.

5.2. Evaluation of Sentinel-2 for geologic mapping

Both simulated datasets and HyMAP mineral maps were comparedwith mapped geological units. This was done by comparing a combinedimage ratio product for the simulated Sentinel-2 data to a number ofHyMAP products as well as to the geologic map of the area (Fig. 9).The simulated Sentinel-2 dataset was visually compared to a geologicalmap and mineral maps to see how well this product resembles highspectral resolution derived alteration maps and how well the Sentinel-2 product reflects the known geology. In Fig. 9A a ratio color compositeimage of three HyMAP ratios is shown in RGB: R2100nm/R2171nm(mimics Alunite), R2171nm/R2206nm (mimics Illite) and R2350nm/R2258nm (mimics Limestone, sediments and shyolitic domes). Themore reddish tones in this image correspond to areas of high-sulfidation epithermal alteration related to localized gold mineraliza-tion. The alteration mineralogy of the volcanic rocks is an assemblage

A HyMap Clay band ratiosR2100/R 2171 ≈ AluniteR2171/R 2206 ≈ IlliteR2350/ R2258 ≈ Limestone and seds.

B HyMap MRSFF classificationPyrophylliteAlunite/DickiteAluniteKaoliniteIlliteLimestoneUnclassified

C S2 Sabins band ratiosBand 11/12 ≈ Hydroxyl bearingBand 04/02 ≈ All iron oxidesBand 04/11 ≈ Ferrous iron oxides

D Arribas’ geological map10: Ignimbrite11: Rhyolitic domes12: Ignimbrite13: Volcanoclastic & lacustrine seds14: Outflows and intrusive volcanics19: Reef complex21: Quaternary deposit

Fig. 9. Geological and mineralogical interpretation of the Rodalquilar area. The imaged area is 2.5 × 2 km.

131F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

dominated by alunite and kaolinite. Greenish colors reflect less intenseand lower temperatures of alteration associated with predominantlythe mineral illite. Unaltered volcanic and sedimentary rocks (mostlylimestones) appear in blue. Surface (alteration) mineralogy mapswere obtained using endmember spectra from theUSGS spectral librarywith HyMAP SWIR bands using the multi-range spectral feature fittingprocedure (Clark, Gallagher, & Swayze, 1990). The resulting mineralmap is shown in Fig. 9B showing pyrophyllite, alunite/dickite, kaolinite,illite, and limestone. For comparison, the published Spanish geologicmap (Arribas, 1993) is shown in Fig. 9D.

Fig. 9C shows a so-called Sabins ratio false color composite image(Sabins, 1999) for the simulated Sentinel-2 image with in RGB: Band11/12 (mimics LandsatTM 5/7), Band 04/02 (mimics LandsatTM 3/1)and Band 04/11 (mimics łandsat TM 3/5). This color scheme showshydroxyl bearing alteration minerals in predominantly reddish colors,iron oxides in predominantly greenish colors and ferrous iron oxidesin predominantly bluish colors.

Sentinel-2 picks up both the illite aswell as the iron oxide rich alteredclay minerals (green in Fig. 9A–C). However, illite is often confused with

Quaternary deposits (unit 21 in Fig. 9D). Carbonaterocks (unit 19 inFig. 9D) are mapped in the Sentinel-2 band ratios 11/12 and 04/11 (thepurple tones in Fig. 9C), but these are not well distinguished from clasticsediments other than by presence of iron-rich volcanic products. In thesimulated Sentinel-2 image mineral alteration (red tones in Fig. 9C) isclearly associated with the iron rich volcanic rocks (units 10 & 12 inFig. 9D).

6. Conclusions

We simulated ESA's Sentinel-2 products that are relevant to thegeology (and soil science) community and compared these to wellestablished band ratio products from ASTER. Airborne hyperspectralimagery from the HyMAP sensor over the Rodalquilar mining area(Cabo de Gata volcanic field, SE Spain), an area of extensive hydrother-mal alteration, is used as input to the simulation.

The simulated image products demonstrate a good correspondencebetween ASTER and Sentinel-2 VNIR and SWIR bands. A number ofband ratios are proposed for Sentinel-2 to derive the following

132 F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

products: ferric iron, ferrous iron, laterite, gossan, ferrous silicate andferric oxides. These are compared to established band ratios of realASTER data as well as simulated (from HyMAP) ASTER data and theycorrelate favourably. Investigating the spatial patterns reveals thatthere is a good match between the proposed Sentinel-2 products andthose of ASTER. The resulting products are also compared to the localgeologic map of the imaged hydrothermal area and it is shown thatthey support the existing conceptual geologic model of the epithermaldeposit.

Acknowledging that many geologic systems are dynamic, datacontinuity and sensor cross and intercalibration remains an importantissue. Time gaps in the optical remote sensing data series in particularto fact that no follow up of ASTER is foreseen other than HySPIRI(Abrams, Pieri, Realmuto, & Wright, 2013) make the community toexplore all sources of information. Hence this study shows that theSentinel-2 mission (Drusch et al., 2012) can fill these gaps and canprovide data continuity for a part of the r products.

Acknowledgement

The authors are grateful to Michael Berger and Ferran Gascon Roca(ESA) for providing the most recent spectral response functions ofSentinel-2.

References

Abdelsalam, M. G., Stern, R. J., & Berhane, W. G. (2000). Mapping gossans in arid regionswith Landsat TM and SIR-C images: The Beddaho Alteration Zone in northern Eritrea.Journal of African Earth Sciences, 30, 903–916.

Abrams,M. (2000). The Advanced Spaceborne Thermal Emission and Reflection Radiometer(ASTER): Data products for the high spatial resolution imager on NASA's Terra platform.International Journal of Remote Sensing, 21, 847–859.

Abrams, M., & Hook, S. J. (1995). Simulated ASTER data for geologic studies. IEEETransactions on Geoscience and Remote Sensing, 33, 692–699.

Abrams, M., Pieri, D., Realmuto, V., & Wright, R. (2013). Using EO-1 Hyperion data asHyspIRI preparatory data sets for volcanology applied to Mt Etna, Italy. IEEE Journalof Selected Topics in Applied Earth Observations and Remote Sensing, 6, 375–385.

Arribas, A., 1993. Mapa Geologico del Distrito Minero de Rodalquilar, Almeria. InstitutoTecnologico GeoMinero de Espana, Madrid. Scale 1:25.000.

Arribas, A., Cunningham, C., Rytuba, J., Rye, R., Kelly, W., Podwysocki, M., et al. (1995).Geology, geochronology,fluid inclusions, and isotope geochemistry of the Rodalquilargold alunite deposit, Spain. Economic Geology and the Bulletin of the Society of EconomicGeologists, 90, 795–822.

Atzberger, C., & Richter, K. (2012). Spatially constrained inversion of radiative transfermodels for improved LAI mapping from future Sentinel-2 imagery. Remote Sensingof Environment, 120, 208–218.

Bedini, E., van der Meer, F., & van Ruitenbeek, F. (2009). Use of HyMap imaging spectrom-eter data to map mineralogy in the Rodalquilar caldera, southeast Spain. InternationalJournal of Remote Sensing, 30, 327–348.

Berger, M., Moreno, J., Johannessen, J. A., Levelt, P. F., & Hanssen, R. F. (2012). ESA's senti-nel missions in support of Earth system science. Remote Sensing of Environment, 120,84–90.

Boccaletti, M., Bonini, M., Mazzuoli, R., Abebe, B., Piccardi, L., & Tortorici, L. (1998).Quaternary oblique extensional tectonics in the Ethiopian Rift (Horn of Africa).Tectonophysics, 287, 97–116.

Carranza, E. J. M., van Ruitenbeek, F. J. A., Hecker, C., van der Meijde, M., & van derMeer, F. D. (2008). Knowledge-guided data-driven evidential belief modelingof mineral prospectivity in Cabo de Gata, SE Spain. International Journal ofApplied Earth Observation and Geoinformation, 10, 374–387.

Clark, R., Gallagher, A., & Swayze, G. (1990). Material absorption band depth mapping ofimaging spectrometer data using the complete band shape least-squares algorithmsimultaneously fit to multiple spectral features from multiple materials. Proceedingsof the Third Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop. JPLPublication, Vol. 90–54. (pp. 176–186).

Clevers, J. G. P. W., & Gitelson, A. A. (2013). Remote estimation of crop and grass chloro-phyll and nitrogen content using red-edge bands on Sentinel-2 and -3. InternationalJournal of Applied Earth Observation and Geoinformation, 23, 344–351.

Cloutis, E. A. (1996). Hyperspectral geological remote sensing: Evaluation of analyticaltechniques. International Journal of Remote Sensing, 17, 2215–2242.

Cooper, B.L., Salisbury, J. W., Killen, R. M., & Potter, A. E. (2002). Midinfrared spectralfeatures of rocks and their powders. Journal of Geophysical Research-Planets, 107.

Cudahy, T., & Hewson, R. (2002). ASTER geological case histories: Porphyry-skarn-epithermal, iron oxide Cu-Au and Broken Hill Pb-Zn-Ag. Annual General Meeting ofthe Geological Remote Sensing Group: ASTER Unveiled. London, UK: Burlington house(December 5–7).

D'Odorico, P., Gonsamo, A., Damm, A., & Schaepman, M. E. (2013). Experimental evalua-tion of Sentinel-2 spectral response functions for NDVI time-series continuity. IEEETransactions on Geoscience and Remote Sensing, 51, 1336–1348.

Debba, P., Carranza, E. J. M., Stein, A., & van derMeer, F. D. (2009). Deriving optimal explo-ration target zones on mineral prospectivity maps. Mathematical Geosciences, 41,421–446.

Delegido, J., Verrelst, J., Meza, C. M., Rivera, J. P., Alonso, L., & Moreno, J. (2013). A red-edgespectral index for remote sensing estimation of green LAI over agroecosystems.European Journal of Agronomy, 46, 42–52.

Drusch, M., Del Bello, U., Carlier, S., Colin, O., Fernandez, V., Gascon, F., et al. (2012).Sentinel-2: ESA's optical high-resolution mission for GMES operational services.Remote Sensing of Environment, 120, 25–36.

European Space Agency (2010). GMES Sentinel-2 Mission Requirements Document.Technical Report issue 2 revision 1. : European Space Agency (March. URL: http://esamultimedia.esa.int/docs/GMES/Sentinel-2MRD.pdf).

Ferrier, G., White, K., Griffiths, G., Bryant, R., & Stefouli, M. (2002). The mapping of hydro-thermal alteration zones on the island of Lesvos, Greece using an integrated remotesensing dataset. International Journal of Remote Sensing, 23, 341–356.

Frampton, W. J., Dash, J., Watmough, G., & Milton, E. J. (2013). Evaluating the capabilitiesof Sentinel-2 for quantitative estimation of biophysical variables in vegetation. ISPRSJournal of Photogrammetry and Remote Sensing, 82, 83–92.

Fraser, A., Huggins, P., Rees, J., & Cleverly, P. (1997). A satellite remote sensing techniquefor geological structure horizon mapping. International Journal of Remote Sensing, 18,1607–1615.

Gad, S., & Kusky, T. (2006). Lithological mapping in the eastern desert of Egypt, theBarramiya area, using Landsat thematic mapper (TM). Journal of African EarthSciences, 44, 196–202.

Goetz, A. F. H., & Rowan, L. C. (1981). Geologic remote-sensing. Science, 211, 781–791.Gomez, C., Delacourt, C., Allemand, P., Ledru, P., & Wackerle, R. (2005). Using ASTER re-

mote sensing data set for geological mapping, in Namibia. Physics and Chemistry ofthe Earth, 30, 97–108.

Haselwimmer, C. E., Riley, T. R., & Liu, J. G. (2011). Lithologic mapping in the Oscar II Coastarea, Graham Land, Antarctic Peninsula using ASTER data. International Journal ofRemote Sensing, 32, 2013–2035.

Hedley, J., Roelfsema, C., Koetz, B., & Phinn, S. (2012). Capability of the Sentinel 2 missionfor tropical coral reef mapping and coral bleaching detection. Remote Sensing ofEnvironment, 120, 145–155.

Hook, S. J., Myers, J. E. J., Thome, K. J., Fitzgerald, M., & Kahle, A.B. (2001). The MODIS/ASTER airborne simulator (MASTER)—A new instrument for earth science studies.Remote Sensing of Environment, 76, 93–102.

Hubbard, B. E., Crowley, J. K., & Zimbelman, D. R. (2003). Comparative alterationmineral mapping using visible to shortwave infrared (0.4-2.4 μm) Hyperion,ALI, and ASTER imagery. IEEE Transactions on Geoscience and Remote Sensing,41, 1401–1410 (Part 1).

Hunt, G. (1977). Spectral signatures of particulate minerals in the visible and near-infrared. Geophysics, 42, 501–513.

Kalinowski, A., & Oliver, S. (2004). ASTER mineral index processing manual. TechnicalReport. Geoscience Australia (URL: http://www.ga.gov.au/image_cache/GA7833.pdf).

Khan, S. D., Mahmood, K., & Casey, J. F. (2007). Mapping of Muslim Bagh ophiolitecomplex (Pakistan) using new remote sensing, and field data. Journal of Asian EarthSciences, 30, 333–343.

Kruse, F. A., & Perry, S. L. (2013). Mineral mapping using simulated Worldview-3 short-wave-infrared imagery. Remote Sensing, 5, 2688–2703.

Li, P. J., Long, X. Y., & Liu, L. (2007). Ophiolite mapping using ASTER data: a case study ofDerni ophiolite complex. Acta Petrologica Sinica, 23, 1175–1180.

Mars, J. C., & Rowan, L. C. (2006). Regional mapping of phyllic- and argillic-altered rocks inthe Zagros magmatic arc, iran, using Advanced Spaceborne Thermal Emission andReflection Radiometer (ASTER) data and logical operator algorithms. Geosphere, 2,161–186.

Mars, J. C., & Rowan, L. C. (2010). Spectral assessment of new ASTER SWIR surface reflec-tance data products for spectroscopic mapping of rocks and minerals. Remote Sensingof Environment, 114, 2011–2025.

Qiu, F., Abdelsalam,M., & Thakkar, P. (2006). Spectral analysis of ASTER data covering partof the Neoproterozoic Allaqi-Heiani suture, Southern Egypt. Journal of African EarthSciences, 44, 169–180.

Rajesh, H. M. (2004). Application of remote sensing and GIS in mineral resourcemapping - an overview. Journal of Mineralogical and Petrological Sciences, 99, 83–103.

Richter, K., Atzberger, C., Vuolo, F., Weihs, P., & D'Urso, G. (2009). Experimental assess-ment of the Sentinel-2 band setting for RTM-based LAI retrieval of sugar beet andmaize. Canadian Journal of Remote Sensing, 35, 230–247.

Richter, K., Hank, T. B., Vuolo, F., Mauser, W., & D'Urso, G. (2012). Optimal exploitation ofthe Sentinel-2 spectral capabilities for crop leaf area index mapping. Remote Sensing,4, 561–582.

Richter, R., & Schlapfer, D. (2002). Geo-atmospheric processing of airborne imagingspectrometry data. Part 2: Atmospheric/topographic correction. International Journalof Remote Sensing, 23, 2631–2649.

Rodgers, S. E., & Oliver, M.A. (2007). A geostatistical analysis of soil, vegetation, and imagedata characterizing land surface variation. Geographical Analysis, 39, 195–216.

Rowan, L. C., & Mars, J. C. (2003). Lithologic mapping in theMountain Pass, California areausing Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER)data. Remote Sensing of Environment, 84, 350–366.

Rytuba, J. J., Arribas, A., Cunningham, C. G., McKee, E. H., Podwysocki, M. H., Smith, J. G.,et al. (1990). Mineralized and unmineralized calderas in Spain. 2. Evolution of therodalquilar caldera complex and associated gold-alunite deposits. MineraliumDeposita, 25, S29–S35.

Sabins, F. F. (1999). Remote sensing for mineral exploration. Ore Geology Reviews, 14,157–183.

Salama, M. S., Radwan, M., & van der Velde, R. (2012). A hydro-optical model for derivingwater quality variables from satellite images (HydroSat): A case study of the Nile

133F.D. van der Meer et al. / Remote Sensing of Environment 148 (2014) 124–133

River demonstrating the future Sentinel-2 capabilities. Physics and Chemistry of theEarth, 50–52, 224–232.

Salisbury, J. W., Walter, L. S., & Vergo, N. (1989). Availability of a library of infrared(2.1–25.0 μm) mineral spectra. American Mineralogist, 74, 938–939.

Schetselaar, E. M., Chung, C. J. F., & Kim, K. E. (2000). Integration of landsat TM, gamma-ray,magnetic, and field data to discriminate lithological units in vegetated granite-gneissterrain. Remote Sensing of Environment, 71, 89–105.

Schlapfer, D., & Richter, R. (2002). Geo-atmospheric processing of airborne imagingspectrometry data. Part 1: Parametric orthorectification. International Journal ofRemote Sensing, 23, 2609–2630.

Stuffler, T., Kaufmann, C., Hofer, S., Forster, K. P., Schreier, G., Mueller, A., et al. (2007). TheEnMAP hyperspectral imager—An advanced optical payload for future applications inEarth observation programmes. Acta Astronautica, 61, 115–120.

Van der Meer, F. (2006). Indicator kriging applied to absorption band analysis inhyperspectral imagery: A case study from the Rodalquilar epithermal gold miningarea, SE Spain. International Journal of Applied Earth Observation and Geoinformation,8, 61–72.

van der Meer, F. D., van der Werff, H. M.A., van Ruitenbeek, F. J. A., Hecker, C. A., Bakker,W. H., Noomen,M. F., et al. (2012). Multi- and hyperspectral geologic remote sensing:A review. International Journal of Applied Earth Observation and Geoinformation, 14,112–128.

van der Werff, H., van Ruitenbeek, F., van der Meijde, M., van der Meer, F., de Jong, S., &Kalubandara, S. (2007). Rotation-variant template matching for supervised

hyperspectral boundary detection. IEEE Geoscience and Remote Sensing Letters, 4,70–74.

Vaughan, R. G., Hook, S. J., Calvin, W. M., & Taranik, J. V. (2005). Surface mineral mappingat Steamboat Springs, Nevada, USA, with multi-wavelength thermal infrared images.Remote Sensing of Environment, 99, 140–158.

Verrelst, J., Munoz, J., Alonso, L., Delegido, J., Rivera, J. P., Camps-Valls, G., et al.(2012). Machine learning regression algorithms for biophysical parameterretrieval: Opportunities for Sentinel-2 and -3. Remote Sensing of Environment,118, 127–139.

Yamaguchi, Y., Kahle, A.B., Tsu, H., Kawakami, T., & Pniel, M. (1998). Overview of Ad-vanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER). IEEETransactions on Geoscience and Remote Sensing, 36, 1062–1071.

Yamaguchi, Y., & Naito, C. (2003). Spectral indices for lithologic discrimination and map-ping by using the ASTER SWIR bands. International Journal of Remote Sensing, 24,4311–4323.

Yesou, H., Besnus, Y., & Rolet, J. (1993). Extraction of spectral information from Landsat-TM data and merger with SPOT panchromatic imagery—A contribution to the studyof geological structures. ISPRS Journal of Photogrammetry and Remote Sensing, 48,23–36.

Zhang, X., Pamer, M., & Duke, N. (2007). Lithologic and mineral informationextraction for gold exploration using ASTER data in the south ChocolateMountains (California). ISPRS Journal of Photogrammetry and Remote Sensing,62, 271–282.