American Society of Photogrammetry and Remote Sensing Imaging & Geospatial Technology Forum Tampa, Florida, USA. May 4th – May 8th 2015.

On the feasibility of detecting Rare Earth Element (REE) deposits using remote sensing

David A. Neavea, Martin Blackb,c, Teal R. Rileyb, Sally A. Gibsona, Graham Ferrierc,

Frances Walld and Sam Broom-Fendleyd

a Department of Earth Sciences, University of Cambridge, United Kingdom, dan27@cam.ac.uk

b British Antarctic Survey, High Cross, Madingley Road, Cambridge, United Kingdom c Department of Geography, Environment and Earth Sciences, University of Hull, United Kingdom

d Camborne School of Mines, University of Exeter, United Kingdom

Rare Earth Elements (REEs) are crucial raw materials in a wide range of technological products and are consequently considered amongst the critical metals, i.e. they are of key economic importance and might be susceptible to future supply restrictions. Carbonatites (igneous rocks containing >50% carbonate) and alkaline silicate rocks are by the far the most significant hosts of REEs. Many REEs have unique spectral properties at visible and near infrared wavelengths (VNIR; 400–1400 nm) that make them directly detectable with reflectance spectroscopy: most lanthanide series elements undergo f-f electronic transitions caused by absorption of photons at specific wavelengths resulting in multiple diagnostic absorptions. Although previous generations of multispectral and hyperspectral instruments have been insufficiently sensitive to resolve the narrow absorption features associated with REEs, improvements in the sensitivity of upcoming instruments, such as HyspIRI and EnMAP, may enable REEs to be detected from satellite-based platforms for the first time. In order to ground truth future remote sensing data, we have acquired spectral reflectance and emissivity data from a range of REE-mineralised samples that contain the following REE-rich phases: monazite-(Ce), bastnäsite-(Ce), synchysite-(Ce), ancylite(-Ce), eudialyte and pyrochlore. VNIR reflectance and TIR (thermal infra-red) emissivity spectra were collected at the NERC Field Spectroscopy Facility (FSF) using an ASD FieldSpec® Pro field spectroradiometer and MIDAC field portable FTIR. Initial findings confirm the presence of strong REE absorbance features (particularly Nd) in the VNIR reflectance spectra of carbonatites and syenites from locations including Bayan Obo, China and Mountain Pass, USA. By resampling our spectra to the spectral responses of various instruments we demonstrate that REE absorption features should be detectable in high quality hyperspectral datasets of high grade deposits.

OnStheSfeasibilitySofSdetectingSrareSearthSelementSLREE8SdepositsSusingSremoteSsensingDavidSA.SNeaveUfafCOartinC_lack†fbfcfCTealCR[CRileybfCSallyC/[C+ibsonafC+rahamCNerriercfCNrancesCWalldCandCSamC_roomHNendleyd

UVorrespondingCauthor)CdanL9–cam[ac[ukEC†®resentingCauthorECaUniversityCofCVambridgeECb_ritishC/ntarcticCSurveyECcUniversityCofCqullECdVamborneCSchoolCofCOines

References[w]CRowanfC=[CV[fCxingstonfCO[C7CVrowleyfC~[C;w68(P[CDconomicC+eologyC8wfC8J9−89w[[L]C_edinifCD[C;L]]6P[CRemoteCSensingCofCDnvironmentCwwKfCwL]8−wLw6[[K]CRowanfC=[CV[C7COarsfC~[CV[C;L]]KP[CRemoteCSensingCofCDnvironmentC8&fCKJ]−K(([[&]COarsfC~[CV[C7CRowanfC=[CV[C;L]wwP[C+eosphereC9fCL9(–L86[[J]Cwww[enmap[org1sites1default1files1pdf1Table_DnO/®_Specs[pdf

[(]CRowanfC=[CV[C7C_owersfCT[C;w66JP[CDconomicC+eologyC6]fCw6((−w68L[[9]CVrowleyfC~[fCRowanfC=[C7C®odwysockifCO[C;w688P[C®roceedingsCofCtheC/irborneCVisible1>nfraredC>magingCSpectrometerC;/V>R>SPC®erformanceCDvaluationCWorkshopfC~®=CpublicationC(−8[[8]C_ioucasHIiasfC~[CO[C7C'ascimentofC~[CO[C®[C;L]]8P[C>DDDCTransactionsConC+eoscienceCandCRemoteCSensingC&(fCL&KJ–L&&J[

RDDCabsorptionCfeaturesCcanCbeC identifiedC inCV'>RCspectraCfromCaCnumberCofCRDDHrichCcarbonatitesCusingCaCfieldHportableCspectroradiometer[

SimulatingC remoteC sensingCdataCbyC resamplingC ;convolvingPC labC spectraC toC aC rangeCofCintrumentC SRNsC indicatesC thatC RDDHspectraC canC theoreticallyC beC detectedC byChyperspectralC sensorsC onC bothC aircraftHC andC satelliteHbasedC platforms[C RDDsC areC notCunambiguoslyCidentifiableCusingCcurrentCmultispectralCsensors[

SensitivityCanalysisCindicatesCthatCRDDCabsorptionCfeaturesfCsuchCasCtheCdiagnosticC'dKvCabsorptionsCcentredCatC9&]CnmCandC8]]CnmfCareCreproducedCinCresampledCspectraCofC ChighCgradeCrocksCatCsignalHtoHnoiseCratiosCasClowCasCw]][

®etrographicC andC geochemicalC analysesC areC currentlyC underwayC inC orderC toC quantifyCrelationshipsCbetweenCspectralCfeaturesfCmineralogyCandCwholeHrockCgeochemistry[

'dKvCabsorptionCfeaturesCareCindependentCofCRDDCmineralogy[

Conclusions

TheCsensitivityCofChyperspectralCremoteCsensingC instrumentsC toC RDDC absorptionCfeaturesCwasCtestedCbyCaddingCrandomC+aussianC noiseC toC aC numberC ofCresampledCspectraCusingCtheCmethodCofC_ioucasHIiasC7C'ascimento[8][

SyntheticC noiseC wasC generatedC atC aCrangeC ofC signalHtoHnoiseC ratiosC ;S'RsPCrelevantC toC remoteC sensingCapplications)C/V>R>SChasCbeenCshownCtoCachieveC S'R3KJ]C atCλ~9]]C nm[C DnO/®CisCexpectedCtoCachieveCS'RsCupCtoC&J]CatCλ~9]]CnmCunderCidealCCconditions[J][

RDDC absorptionC featuresC remainCresolvableC atC S'R0w]]C inC RDDHrichCsamplesC thatC haveC beenC resampledC toCtheC DnO/®C SRNfC suchC asC theC _andedCjreC exmapleC fromC Bayang OboQg ChinaCthatCisCshownCtoCtheCright[C/CS'RCofCw]]CshouldCbeCroutinelyCachieveableCduringChyperspectralCimagingCcampaigns

&3R

efle

ctan

ce3K

offs

et3fo

r3cl

arity

3−3u

nits

3of3F

&3s

how

nd

épp zpp Cpp Sppp S.pp Sépp Szpp SCpp .ppp ..pp .éppWavelength3Knmd

VSD

SNR3=3.F

SNR3=3Fp

SNR3=3Spp

SNR3=3.Fp

SNR3=3Fpp

NdHP Vtmospheric3H.O

SensitivitySanalysis

&3R

efle

ctan

ce3K

offs

et3fo

r3cl

arity

3−3u

nits

3of3F

&3s

how

nd

épp zpp Cpp Sppp S.pp Sépp Szpp SCpp .ppp ..pp .éppWavelength3Knmd

NdHP Vtmospheric3H.O

VSD3showing3SEE3envelopeVSTER Landsat−CVVIRIS EnMVP

Bastnäsite-SVSmonazite-bearingSBandedSOreSfromSBayanqObo–SChina Bastnäsite-bearingScalciocarbonatiteSfromSMountainqPass–SUSA

&3R

efle

ctan

ce3K

offs

et3fo

r3cl

arity

3−3u

nits

3of3F

&3s

how

nd

épp zpp Cpp Sppp S.pp Sépp Szpp SCpp .ppp ..pp .éppWavelength3Knmd

NdHP

Vtmospheric3H.O

VSD3showing3SEE3envelopeVSTER Landsat−CVVIRIS EnMVP

Monazite-bearingSferrocarbonatiteSfromSKangankunde–SMalawi

&3R

efle

ctan

ce3K

offs

et3fo

r3cl

arity

3−3u

nits

3of3F

&3s

how

nd

épp zpp Cpp Sppp S.pp Sépp Szpp SCpp .ppp ..pp .éppWavelength3Knmd

NdHP Vtmospheric3H.O

VSD3showing3SEE3envelopeVSTER Landsat−CVVIRIS EnMVP

Ancylite-bearingScalciocarbonatiteSfromSQeqertaasaq–SGreenland

&3R

efle

ctan

ce3K

offs

et3fo

r3cl

arity

3−3u

nits

3of3F

&3s

how

nd

épp zpp Cpp Sppp S.pp Sépp Szpp SCpp .ppp ..pp .éppWavelength3Knmd

NdHP

Vtmospheric3H.O

VSD3showing3SEE3envelopeVSTER Landsat−CVVIRIS EnMVP

RemoteSsensingSinstruments

ASTERC −C /dvancedC SpaceborneC ThermalC DmissionC andCReflectionCRadiometer

/STDRC isC aC ~apaneseC multispectralC instrumentC onC theCTerraC satelliteC thatC wasC launchedC inC w666[C TheCinstrumentC collectsC imagesC atC w&C visibleC andC infraredCwavelengthCbandsCwithCaCspatialCresolutionCofCwJ−6]Cm[C/lthoughC/STDRChasCbeenCusedCforCmappingCcarbonatitesCwithC someC success[Kf&]fC it@sC spectralC resolutionC isCinsufficentCtoCdetectCnarrowCRDDCabsorptionCfeatures[

AVIRISC−C/irborneCVisible1>nfraRedC>magingCSpectrometer

/V>R>SC isCanCairborneChyperspectralC instrumentCdesignedCbyC theC ~etC ®ropulsionC =aboratoryC ;~®=PC inC theCUS/C thatChasC beenC inC useC sinceC w689[C TheC instrumentC usesC fourCspectrometersC toC imageCL&&CadjacentC spectralCbandsCatC&]]−LJ]]C nm[C /V>R>SC hasC beenC succesfullyC usedC toCundertakeC lithologicalC mappingC ofC carbonatites[(]fCincludingCpossibleCidentificationCofCRDDHrichCmaterial[9][C

Landsat−8

=andsat−8C isC anC /mericanC multispectralC satelliteC thatCwasClaunchedCinCL]wK[CTheCinstrumentCcollectsCimagesCatCnineC visibleC andC shortwaveC infraredfC andC twoC thermalCinfraredCwavelengthCbandsCwithCresolutionsCofCK]CmCandCw]]C mC respectively[C /sC withC /STDRfC theC spectralCresolutionC ofC =andsat−8C isC insufficentC toC detectC narrowC

EnMAPC−CDnvironmentalCOappingCandC/nalysisC®rogram

DnO/®C isC aC +ermanC satelliteHbasedC hyperspectralCinstrumentCplannedCforClaunchCinCL]wJ[CTheCinstrumentCisCplannedC toC imageC L&&C adjacentC spectralC bandsC atC &L]−L&J]C nmC withC aC spatialC resolutionC ofC K]C m[C DnO/®C isClikelyC toC representC aC newC baselineC inC satelliteHbasedChyperspectralCremoteCsensingCimagery[

OultispectralCinstruments

qyperspectralCinstruments

MeasuredSandSresampledSvisibleSVSnearSinfraredSLVNIR8Sspectra

MeasuringSvisibleSVSnearSinfraredSLVNIR8Sspectra

VollectingC V'>RC spectraC fromCtheC Mountaing PassC sampleCusingC anC /SIC NieldSpec®C ®roCspectroradiometerCatCtheC'DRVCNieldCSpectroscopyCNacility[

OeasurementsCwereCmadeC inC aCdarkroomC usingC aC contactCprobefCandCcalibratedCagainstCaCSpectralonCwhiteCpanel[

TheCsamplingCintervalCisCw[&CnmCatCλ0KJ]−w]]]CnmfCandCLCnmCatCλ0w]]]−LJ]]C nmfC whereC λC isCwavelength[

TheCspectralCresolutionCisCCKCnmCatCλ09]]CnmfCw]CnmCatCλ0w&]]CnmCandCwLCnmCatCλ0Lw]]nm[

VSD

VSTER3SRF

&3R

efle

ctan

ce3K

offs

et3fo

r3cl

arity

3−3u

nits

3of3F

&3s

how

nd

épp zpp Cpp Sppp S.pp Sépp Szpp SCpp .ppp ..pp .éppWavelength3Knmd

Resampled3VSTER3−3VSD3×3SRF

SimulatingSremoteSsensingSdataS–SspectralSconvolution

SpectralC dataC wereC spectrallyC resampledC ;convolvedPC inC orderC toC simulateCacquisitionC byC aC numberC ofC airborneC andC spaceborneC multispectralC andChyperspectralCsensors[C>tC isCthusCpossibleCtoCassessCwhetherCvariousCinstrumentsCwouldCtheoreticallyCbeCableCtoCdetectCtheCnarrowCabsorptionCfeaturesCgeneratedCbyCRDDs[CIetailsCofCtheCchosenCsensorsCareCprovidedCinCtheCpinkCboxCtoCtheCright[C

=aboratoryC dataCwereC resampledC toCdifferentC sensorC characteristicsCusingC publishedC spectralC responseCfunctionsC ;SRNsP[C >nC casesC whereCexplicitCspectralCresponsesChaveCnotCbeenCpublishedfCSRNsCwereCapproxiHmatedC byC curveC fittingC ;e[g[C+aussianPC usingC theC expectedCspectralC characteristicsC ofC sensorsC;wavelengthC centreC andC fullHwidthChalfHmaxP[J][C /nC exampleC showingCtheCresamplingCofCaCRDDHrichCsampleCfromCBayangOboCtoCtheC/STDRCSRNCisCshownCtoCtheCright[

Methods

OapC showingC theCgeographicalCdistributionCofC samplesCusedC inC thisC studyfC asC wellC asC theC collectionsC fromCwhichC theyC wereC sourced[C WholeHrockC analysesCindicateCthatCsamplesCcontainC~w]]–w]]]]]CppmC'd[

SamplesC wereC selectedC toC coverC aC diversityC ofCcarbonateCmineralogiesfCwholeHrockCRDDCcontentsfCandCstylesCofCCRDDCmineralisation[

PyrochloreHCandCeudialyteHbearingCalkalineCrocksC fromCtheC MotzfeldtC andC Ilímaussaqg centresC inC GreenlandCthatC representC 'bfC TaC andC RDDC prospectsC wereC alsoCinvestigated[

SamplesC includeC calciocarbonatitesfC suchC asC OkaQg

CanadafC ferrocarbonatitesfC suchC asC KangankundeQg

MalawifCRDDCoresfCsuchCasCtheC_andedCjreCfromCBayang

ObofCChinaQCandCextrusiveCcarbonatitesCfromgUyaynahfgUAECandCFortgPortalfCUganda[

−SCp˚

−SCp˚

−S.p˚

−S.p˚

−zp˚

−zp˚

p˚

p˚

zp˚

zp˚

S.p˚

S.p˚

SCp˚

SCp˚

−zp˚ −zp˚

−Hp˚ −Hp˚

p˚ p˚

Hp˚ Hp˚

zp˚ zp˚Ilímausaaq

Qeqertaasaq3T3Tikiusaaq

Kangankunde

Mountain3Pass×ayan3Obo

Mount3WeldPhalaborwaJacupiranga

Oka

Stb3Honoré

FenMotzfeldt

Sokli

Uyaynah

Kaiserstuhl

Tororo3T3Sukulu

Panda3Hill

Chilwa3IslandB3Songwe3T3Tundulu

Harker3CollectionB3Cambridge Natural3History3MuseumB3LondonSRK3Consulting3KUKd3Ltdb Camborne3School3of3Mines

University3of3Stb3Vndrews

SampleSsourcesSVScharacteristics

VarbonatitesC andC theirC alkalineC associatesC areC theC primaryC hostsC forC globalC reservesC ofC rareC earthCelementsC;RDDsP[COanyCRDDsChaveCuniqueCspectralCpropertiesCatCvisibleCandCnearCinfraredCwavelengthsC;V'>REC &]]−w&]]C nmPC thatC makeC themC directlyC detectableC withC reflectanceC spectroscopy[C ®reviousCstudiesC indicatedC thatC neodymiumC ;'dPC hasC theC mostC diagnosticC spectralC featuresC ofC theC RDDs[w][C/lthoughCvariousCstudiesChaveCdemonstratedCthatCcarbonatiteCcomplexesCcanCbeCmappedCusingCremoteCsensingC techniques[LfK]fC currentC satelliteHbasedC multispectralC instrumentsC haveC rarelyC beenC ableC toCresolveC theC narrowC absorptionC featuresC associatedC withC RDDsC becauseC ofC theirC coarseC spectralCresolutions[&][CqoweverfCtheCimprovedCspectralCresolutionCinCplannedChyperspectralCinstrumentsfCsuchCasCDnO/®[J]fCmayCenableCRDDsCtoCbeCdetectedCdirectlyCfromCsatelliteHbasedCplatformsCinCfuture[

>nCorderCtoCgroundCtruthCfutureCremoteCsensingCdatasetsfCweChaveCacquiredCspectralCreflectanceCdataCfromC aC rangeC ofC petrologicallyHcharacterisedC RDDHrichC samplesC thatC containC theC followingC RDDHrichCphases)C monazite-)CeäfC bastnäsite-)CeäfC parisite-)CeäfC synchysite-)CeäQg ancylite-)CeäQg eudialyteC andCpyrochlore[C jurC findingsC confirmC theC presenceC ofC strongC 'dKvC absorbanceC featuresC inC theC V'>RCreflectanceCspectraCofCRDDHrichCrocksCfromClocationsCincludingCBayangOboQgChinaECMountaingPassQgUSAECKangankundeQg MalawiEC andg QeqertaasaqQg Greenland[C ResamplingC ;i[eC convolvingPC ourC laboratoryCspectraCtoCtheCspectralCresponseCfunctionsC;SRNsPCofCaCrangeCofCremoteCsensingCinstrumentsCshowsCthatCwhileCRDDsCcannotCbeCdetectedCbyCmultispectralCinstrumentsfCRDDsCabsoptionsCcanCbeCresolvedCbyCbothCaircraftHC andC satelliteHbasedC hyperspectralC platformsC atC realisticlyC acheivableC signalHtoHnoiseC ratios[CVhallengesC facingC RDDC detectionC byC remoteC sensingC areC thusC dominantlyC geologicalC ratherC thanCinstrumentalCinCnaturefCi[eCrelateCtoCdepositCsizefClithologicalCheterogeneityCandCstyleCofCoutcropping[

Abstract