0

MODISCollection6.1(C6.1)LAI/FPAR

ProductUser’sGuide

TheC61MOD/MYD15productisidenticalinformattotheC6product.

This(C61)reprocessingdoesnotcontainanychangetothesciencealgorithmused tomake this product. Any improvement or change in the C61 productcomparedtotheproductfromthepriormajorcollectionreprocessing(C6)isfromchangesandenhancementstothecalibrationapproachusedingenerationof the Terra and AquaMODIS L1B products and changes to the polarizationcorrection used in this collection reprocessing. For further details on C61calibration changes and other changes user is encouraged to refer to theCollection 6.1 specific changes that have been summarized here:https://landweb.modaps.eosdis.nasa.gov/QA_WWW/forPage/MODIS_C61_Land_Proposed_Changes.pdf

(Updated:April21,2020)

1

Contents

1.Definitions Page22.SummaryofChangesinC6 Page23.AlgorithmDescription Page24.StandardMODISProducts Page45.HowtoObtaintheData Page66.Contentoftheproductfile Page67.Policies Page118.ContactInformation Page119.RelatedPapers Page11

2

1.DefinitionsLeafareaindex(LAI;dimensionless)isdefinedastheone−sidedgreenleafareaper

unitgroundareainbroadleafcanopiesandasone−halfthetotalneedlesurfaceareaperunitgroundareainconiferouscanopies.STD LAI is the estimated retrieval uncertainty, i.e., “true LAI” can differ from its

retrievalcounterpartby±STDLAI(SeeFigure1).Fraction of Photosynthetically Active Radiation absorbed by vegetation (FPAR;

dimensionless)isdefinedasthefractionofincidentphotosyntheticallyactiveradiation(400−700nm)absorbedbythegreenelementsofavegetationcanopy.STDFPARis theestimatedretrievaluncertainty, i.e., “trueFPAR”candiffer fromits

retrievalcounterpartby±STDFPAR(SeeFigure1).2.SummaryofChangesinC6• UsesL2G–litesurfacereflectanceat500mresolutionas(MOD09GA1)inputinplace

ofreflectanceat1kmresolution(MODAGAGG2)inCollection5.Anintermediatedailysurface reflectance product (MOD15IP 3 ) at 500 m resolution is created fromMOD09GAbeforebeingusedforLAI/FPARretrieval.

• Productsaregeneratedataspatialresolutionof500m.• Usesimprovedmulti-yearlandcoverproduct.3.AlgorithmDescriptionThe MODIS LAI/FPAR algorithm consists of a main Look-up-Table (LUT) based

procedurethatexploitsthespectralinformationcontentoftheMODISred(648nm)andnear-infrared (NIR,858nm)surface reflectances,and theback-upalgorithmthatusesempirical relationships between Normalized Difference Vegetation Index (NDVI) andcanopy LAI and FPAR. The LUT was generated using 3D radiative transfer equation[Knyazikhinetal.,1998]. Inputstothealgorithmare(i)vegetationstructuraltype,(ii)sun-sensorgeometry,(iii)BRFsatred(648nm)andnear-infrared(NIR,858nm)spectralbandsand(vi)theiruncertainties(Table1).Figure1illustratesthemainalgorithm:for

1MOD09GA is aMODIS daily surface reflectance product, which provides daily atmosphericallycorrectedsurfacereflectanceat500mresolutioninsevenspectralbands.MOD09GAcanbeaccessedviaReverbtool(PleaserefertotheSection5.HowtoObtaintheData)2 MODAGAGG is a MODIS daily aggregated surface reflectance product, which provides dailyatmospherically corrected surface reflectance at 1 km resolution in seven spectral bands.MODAGAGGisnotanarchivedproduct.3MOD15IPistheintermediateMODISdailysurfacereflectanceproductat500mresolution,whichispreprocessedfromthedailyMOD09GAsurfacereflectanceproduct,forLAI/FPARproduction.ThisproductisanequivalentofMODAGAGGinC5andnotarchived.

3

each pixel it compares observed and modeled spectral BRFs for a suite of canopystructuresandsoilpatternsthatrepresentanexpectedrangeoftypicalconditionsforagiven biome type. All canopy/soil patterns and corresponding FPAR values forwhichmodeledandobservedBRFsdifferwithinaspecifieduncertaintylevelareconsideredasacceptablesolutions.ThemeanvaluesofLAI,FPAR,theirdispersions,STDLAIandSTDFPAR,arereportedasretrievalsandtheiruncertainties[Knyazikhinetal.,1998].Inthecaseofdensecanopies,thereflectancessaturate,andarethereforeweaklysensitivetochangesincanopyproperties.Thereliabilityofparametersretrievedundertheconditionof saturation is low, that is, the dispersion of the solution distribution is large. Suchretrievals are flagged inQA layers (Table5).When theLUTmethod fails to localize asolution,theback-upmethodisutilized.Thealgorithmpath(mainorbackup)isarchivedinQAlayers(Table5).Analysesofthealgorithmperformanceindicatethatbestquality,highprecisionretrievalsareobtainedfromthemainalgorithm[Yangetal.2006b;Yangetal.2006c].Thealgorithmpathisthereforeakeyqualityindicator.ThealgorithmhasinterfaceswiththeMODISSurfaceReflectanceProduct(MOD09GA)

andtheMODISLandCoverProduct(MCD12Q1).TechnicaldetailsofthealgorithmcanbefoundintheAlgorithmTheoreticalBasisDocument(ATBD)4.

A

B

Figure 1. Schematic illustration of the main algorithm. Panel A: Distribution ofvegetated pixels with respect to their reflectances at red and near-infrared (NIR)spectralbandsfromTerraMODIStileh12v04.Apointonthered-NIRplaneandanareaaboutit(yellowellipsedefinedbya𝜒!distribution)aretreatedasthemeasuredBRFatagivensun-sensorgeometryanditsuncertainty.Eachcombinationofcanopy/soilparametersandcorrespondingFPARvaluesforwhichmodeledreflectancesbelongtotheellipseisanacceptablesolution.PanelB:Densitydistributionfunctionofacceptable

4ATBDforMODISLAI/FPARproductcanbedirectlydownloadedfrombelowlink:http://modis.gsfc.nasa.gov/data/atbd/atbd_mod15.pdf

Red$

Near(Infrared

$

Soil$line$(LAI=0)$

Prob

abili

ty d

ensi

ty

4

solutions.ShownissolutiondensitydistributionfunctionofLAIforfivedifferentpixels.The mean LAI and its dispersion (STD LAI) are taken as the LAI retrieval and itsuncertainty.This technique isused toestimatemeanFPARand itsdispersions(STDFPAR).From[Knyazikhinatal,1998].Table1.Theoreticalestimatesofuncertainties(%)intheBRFsusedintheC6.1LAI/FPARalgorithm

BiomeTypeUncertainty

Red(648nm) NIR(858nm)Biome1(Grasses/Cerealcrops) 20% 5%Biome2(Shrubs) 20% 5%Biome3(Broadleafcrops) 20% 5%Biome4(Savanna) 20% 5%Biome5(EvergreenBroadleafforest) 30% 15%Biome6(DeciduousBroadleafforest) 30% 15%Biome7(EvergreenNeedleleafforest) 30% 15%Biome8(DeciduousNeedleleafforest) 30% 15%4.StandardMODISProductsThestandardMODISC6.1LAI/FPARproducts(M*D15A*H)areat500−meterspatial

resolutionandincludeLAI/FPARretrievals fromTerraMODIS,AquaMODISandTerraMODIS+Aqua MODIS Combined. The temporal compositing periods are 8 and 4 days(Table2).

Table2.DescriptionoftheStandardMODISLAI/FPARproducts

OfficialName Platform RasterTypeSpatialResolution

TemporalGranularity

MOD15A2H Terra Tile 500m 8Day

MYD15A2H Aqua Tile 500m 8Day

MCD15A2H Terra+AquaCombined

Tile 500m 8Day

MCD15A3H Terra+AquaCombined

Tile 500m 4Day

TheMODISLAI/FPARproductsusetheSinusoidalgridtillingsystem(Figure2).Tiles

are10degreesby10degreesattheequator(Table3).Thetilecoordinatesystemstarts

5

at (0, 0) (horizontal tile number, vertical tile number) in the upper left corner andproceedsright(horizontal)anddownward(vertical).Thetileinthebottomrightcorneris(35,17).Table3.DatasetcharacteristicsoftheMODISLAI/FPARproducts

Characteristics C6.1Product

TemporalCoverageMOD15:February18,2000–MYD15&MCD15:July4,2002–

Area ~10x10lat/longFileSize ~0.8MBcompressedProjection SinusoidalDataFormat HDF−EOSDimensions 2400x2400rows/columnsResolution 500meterScienceDataSets(SDSHDFLayers) 6

Figure2.MODISSinusoidalTilingSystem

MODISproductfilenames(i.e.,thelocalgranuleID)followanamingconventionthat

gives useful information regarding the specific product. For example, the filenameMOD15A2H.A2006001.h08v05.006.2006012234657.hdfindicates:

ü MOD15A2H–ProductShortName

6

ü .A2006001–JulianDateofAcquisition(A−YYYYDDD)ü .h08v05–TileIdentifier(horizontalXX,verticalYY)ü .006–CollectionVersionü .2006012234657–JulianDateofProduction(YYYYDDDHHMMSS)ü .hdf–DataFormat(HDF−EOS)TheMODIS LAI/FPAR products have two sources ofmetadata: the embeddedHDF

metadata, and the external ECS metadata. The HDF metadata contains valuableinformationincludingglobalattributesanddataset−specificattributespertainingtothegranule.TheECS(generatedbytheEOSDISCoreSystem).metfileistheexternalmetadatafile in XML format, which is delivered to the user along with the MODIS product. Itprovidesasubsetof theHDFmetadata.Somekey featuresofcertainMODISmetadataattributesincludethefollowing:

ü TheXdimandYdimrepresenttherowsandcolumnsofthedata,respectively.ü TheProjection andProjParams identify the projection and its corresponding

projectionparameters.ü TheSinusoidalProjectionisusedformostofthegriddedMODISlandproducts,

andhasauniquespheremeasuring6371007.181meters.ü The UpperLeftPoinitMtrs is in projection coordinates, and identifies the very

upperleftcorneroftheupperleftpixeloftheimagedata.ü TheLowerRightMtrsidentifiestheverylowerrightcornerofthelowerrightpixel

of the image data. These projection coordinates are the only metadata thataccuratelyreflecttheextremecornersofthegriddedimage.

ü ThereareadditionalBOUNDINGRECTANGLEandGRINGPOINT fieldswithinthemetadata, which represent the latitude and longitude coordinates of thegeographictilecorrespondingtothedata.

5.HowtoObtaintheDataNASA EARTHDATA (https://earthdata.nasa.gov/): This tool provides access to a

completedatarecordofallMODISproductsavailablefromtheLPDAAC.

6.ContentoftheproductfileTheMODISLAI/FPARproductisat500−meterresolutioninaSinusoidalgrid.Science

Data Sets provided in the product include LAI, FPAR, quality ratings, and standarddeviationforeachvariable,STDLAIandSTDFPAR(Table4).

7

Table4.ScientificDataSetsincludedintheMODISLAI/FPARproduct

ScientificDataSets(HDFLayers)(6)

Units BitTypeFillValue

ValidRange

MultiplyByScaleFactor

Fpar_500m Dimensionless8−bitunsignedinteger

249−255 0−100 0.01

Lai_500m Dimensionless8−bitunsignedinteger

249−255 0−100 0.1

FparLai_QC Classflag8−bitunsignedinteger

255 0−254 N/A

FparExtra_QC Classflag8−bitunsignedinteger

255 0−254 N/A

FparStdDev_500m5 Dimensionless8−bitunsignedinteger

248−255 0−100 0.01

LaiStdDev__500m5 Dimensionless8−bitunsignedinteger

248−255 0−100 0.1

6.1.DescriptionofQCSDSQualitycontrol (QC)measuresareproducedatboth the file (containingoneMODIS

tile)andatthepixellevelsfortheM*D15A*Hproduct.Atthetilelevel,theseappearasaset of EOSDIS core system (ECS) metadata fields. At the pixel level, quality controlinformation is representedby2data layers (FparLai_QCandFparExtra_QC) in the filewithM*D15A*Hproduct.NotethattheLAI/FPARalgorithmisexecutedirrespectiveofinputquality.ThereforeusershouldconsulttheQClayersoftheLAI/FPARproducttoselectreliableretrievals.

5ThemainalgorithmemploysaLUTmethodsimulatedfroma3-Dradiativetransfermodel.TheLUTmethodessentiallysearches forLAI/FPARs foraspecificsetof solarandviewzenithangles,observedBRFsatcertainspectralbandsandbiometypes.TheoutputsaretheLAI/FPARmeanvalues(i.e.,Lai_500m/Fpar_500mscientificdata)averagedoverallacceptablesolutions,andthestandarddeviation(i.e.,LaiStdDev/FparStdDevscientificdata)servingasameasureofthesolutionaccuracy.

8

Table5.ValuesofFparLAI_QC(8−bit)

BitNo. ParameterName BitComb.

FparLai_QC

0 MODLAND_QCbits0

Goodquality(mainalgorithmwithorwithoutsaturation)

1 Otherquality(back−upalgorithmorfillvalues)

1 Sensor0 Terra1 Aqua

2 DeadDetector0

Detectorsapparentlyfineforupto50%ofchannels1,2

1 Deaddetectorscaused>50%adjacentdetectorretrieval

3−4

CloudState(inheritedfromAggregate_QCbits{0,1}cloudstate)

00 0SignificantcloudsNOTpresent(clear)01 1SignificantcloudsWEREpresent

10 2Mixedcloudpresentinpixel

11 3Cloudstatenotdefined,assumedclear

5−7 SCF_QC(five−levelconfidencescore)

0000Main(RT)methodused,bestresultpossible(nosaturation)

001 1Main(RT)methodusedwithsaturation.Good,veryusable

0102Main(RT)methodfailedduetobadgeometry,empiricalalgorithmused

0113Main(RT)methodfailedduetoproblemsotherthangeometry,empiricalalgorithmused

1004Pixelnotproducedatall,valuecouldn’tberetrieved(possiblereasons:badL1Bdata,unusableMOD09GAdata)

Note, in theFparLai_QC, the fieldMODLAND is thestandardonecommon to theall

MODLANDproductsandspecifiestheoverallqualityoftheproduct.Also,severalbitfieldsintheM*D15A*HQAarepassed-thrufromthecorrespondingbitfieldsoftheMOD09GAsurface reflectancesproduct (CloudState, LandSea, etc.). The key indicator of retrievalqualityoftheLAI/FPARproductisSCF_QCbitfielddthatrepresentsalgorithmpath.M*D15A*Hbitpatternsareparsedfromrighttoleft.Individualbitswithinabitword

are read from left to right. The following example illustrates the interpretation ofFparLai_QC. Let assume a single pixel’s value from FparLai_QC layer is 64, thus this

9

decimal value can be converted to a binary value of 1000000 as shown in Figure 3.Interpretationofbit-stringsisalsoshowninFigure3basedonTable5.

Figure3.ExampleofFparLai_QCbit-stringanditsinterpretation

Table5.ValuesofFparExtra_QC(8−bit)

BitNo. ParameterName BitComb. FparExtra_QC

0−1 LandSeaPass−Thru

00 0LANDAggrQC(3,5)values{001}

011SHOREAggrQC(3,5)values{000,010,100}

102FRESHWATERAggrQC(3,5)values{011,101}

11 3OCEANAggrQC(3,5)values{110,111}

2Snow_Ice(fromAggregate_QCbits)

0 Nosnow/icedetected1 Snow/icedetected

3 Aerosol0 Noorlowatmosphericaerosollevels

detected1 Averageorhighaerosollevelsdetected

4Cirrus(fromAggregate_QCbits{8,9})

0 Nocirrusdetected

1 Cirruswasdetected

5 Internal_CloudMask0 Noclouds1 Cloudsweredetected

6 Cloud_Shadow0 Nocloudshadowdetected1 Cloudshadowdetected

7 SCF_Biome_Mask0 Biomeoutsideinterval<1,4>1 Biomeininterval<1,4>

10

ExampleforinterpretationofFparExtra_QCbit-stringsisshowninFigure4.

Figure4.ExampleofFparExtra_QCbit-stringanditsinterpretation

6.2.DescriptionofFillvalueforSDSsUsing the MODIS land cover product (MCD12Q1), each 500m pixel is classified

accordingtoitsstatusasalandornon-landpixel.Anumberofnon-terrestrialpixelclassesarenowcarriedthroughintheproductdatapixels(notQA/QCpixels)whenthealgorithmcouldnotretrieveabiophysicalestimate(Table6and7).Table6.LAIandFPARFillvalueLegends

Value Description

255Fillvalue,assignedwhen:theMOD09GAsurfacereflectanceforchannelVIS,NIRwasassignedits_Fillvalue,orlandcoverpixelitselfwasassignedFillvalue255or254

254 landcoverassignedasperennialsaltorinlandfreshwater253 landcoverassignedasbarren,sparsevegetation(rock,tundra,desert)252 landcoverassignedasperennialsnow,ice251 landcoverassignedas“permanent”wetlands/inundatedmarshlands250 landcoverassignedasurban/built−up249 landcoverassignedas“unclassified”ornotabletodetermine

11

Table7.STDLAIandSTDFPARFillValueLegends

Value Description

255Fillvalue,assignedwhen:theMOD09GAsurfacereflectanceforchannelVIS,NIRwasassignedits_Fillvalue,orlandcoverpixelitselfwasassigned_Fillvalue255or254

254 landcoverassignedasperennialsaltorinlandfreshwater253 landcoverassignedasbarren,sparsevegetation(rock,tundra,desert)252 landcoverassignedasperennialsnow,ice251 landcoverassignedas“permanent”wetlands/inundatedmarshlands250 landcoverassignedasurban/built−up249 landcoverassignedas“unclassified”ornotabletodetermine248 Nostandarddeviationavailable,pixelproducedusingbackupmethod

7.PoliciesPlease find the currentMODIS−relatedDatapolicies on theMODISPolicies page at

https://lpdaac.usgs.gov/lpdaac/products/modis_policies.For informationonhowtociteLPDAACdata,pleaseseeourDataCitationspageat

https://lpdaac.usgs.gov/about/citing_lp_daac_and_data.

8.ContactInformationRangaMyneniDepartmentofGeographyandEnvironment,BostonUniversityEmail:ranga.myneni@gmail.comWeb:http://cliveg.bu.edu9.RelatedPapersAhletal.,2006.MonitoringSpringCanopyPhenologyofaDeciduousBroadleafForest

UsingMODIS,RemoteSens.Environ.,104:88–95.Baret et al., 2006. Evaluation of the representativeness of networks of sites for the

validationandinter–comparisonofgloballandbiophysicalproducts.Propositionofthe CEOS–BELMANIP. IEEE Trans. Geosci. Remote Sens., 44: 1794–1803. DOI:10.1126/science.1199048,2011.

Gangulyetal.,2008.Generatingvegetationleafareaindexearthsystemdatarecordsfrommultiplesensors.Part1:Theory.RemoteSens.Environ.,Vol.112(2008)4333–4343,doi:10.1016/j.rse.2008.07.014

12

Gangulyetal.,2008.Generatingvegetationleafareaindexearthsystemdatarecordsfrommultiple sensors. Part 2: Implementation, Analysis and Validation. Remote Sens.Environ.,112(2008)4318–4332,doi:10.1016/j.rse.2008.07.013

Gaoetal.,2008.AnAlgorithmtoProduceTemporallyandSpatiallyContinuousMODIS–LAITimeSeries.Geophys.Res.Lett.,doi:10.1109/LGRS.2007.907971.

Garrigues et al., 2008. Intercomparison and sensitivity analysis of leaf area indexretrievals fromLAI–2000, AccuPAR, and digital hemispherical photography overcroplands,Agric.For.Meteorol.,doi:10.1016/j.agrformet.2008.02.014.

Garriguesetal.,2008.ValidationandIntercomparisonofGlobalLeafAreaIndexProductsDerived from Remote Sensing Data, J. Geophys. Res., VOL. 113, G02028,doi:10.1029/2007JG000635,2008.

Hashimoto et al., 2012 Exploring Simple Algorithms for Estimating Gross PrimaryProduction in Forested Areas from Satellite Data, Remote Sens., 4, 303–326;doi:10.3390/rs4010303

Huang et al., 2006. The Importance of Measurement Error for Deriving AccurateReference Leaf Area IndexMaps for Validation of theMODIS LAI Product. IEEETrans.Geosci.RemoteSens.,44:1866–1871.

Justice,etal.,1998.Themoderateresolutionimagingspectroradiometer(MODIS):Landremote sensing for global change research. IEEE Trans. Geosc. Remote Sens.,36:1228–1249.

Knyazikhinetal.,1998.SynergisticalgorithmforestimatingvegetationcanopyleafareaindexandfractionofabsorbedphotosyntheticallyactiveradiationfromMODISandMISRdata.J.Geophys.Res.,103:32,257–32,276.

Morisetteetal.,2006.ValidationofglobalmoderateresolutionLAIProducts:aframeworkproposedwithintheCEOSLandProductValidationsubgroup. IEEETrans.Geosci.RemoteSens.44:1804–1817.

Mynenietal.,2002.Globalproductsofvegetation leafareaandfractionabsorbedPARfromyearoneofMODISdata.RemoteSens.Environ.,83:214–231.

Mynenietal.,2007.Largeseasonalchangesinleafareaofamazonrainforests.Proc.Natl.Acad.Sci.,104:4820–4823,doi:10.1073/pnas.0611338104.

Privette et al., 1998. Global validation of EOS LAI and FPARproducts.EarthObserver,10(6):39–42.

Privetteetal.,2002.EarlyspatialandtemporalvalidationofMODISLAIproductinAfrica.RemoteSens.Environ.,83:232–243.

Samantaetal.,2011.Commenton"Drought–InducedReductioninGlobalTerrestrialNetPrimaryProductionfrom2000Through2009",Science,Vol.333,p.1093,

Samantaetal.,2012SeasonalchangesinleafareaofAmazonforestsfromleafflushingand abscission, J. Geophys. Res. VOL. 117, G01015, doi:10.1029/2011JG001818,2012

13

Shabanovetal.,2003.TheeffectofspatialheterogeneityinvalidationoftheMODISLAIandFPARalgorithmoverbroadleafforests,RemoteSens.Environ.,85:410–423.

Tanetal.,2006.TheimpactofgeolocationoffsetsonthelocalspatialpropertiesofMODISdata: Implications for validation, compositing, and band–to–band registration,RemoteSens.Environ.,105:98–114.

Tian et al., 2000. Prototyping of MODIS LAI and FPAR algorithm with LASUR andLANDSATdata.IEEETrans.Geosci.RemoteSens.,38(5):2387–2401.

Tian et al., 2002a. Multiscale Analysis and Validation of the MODIS LAI Product. I.UncertaintyAssessment.RemoteSens.Environ.,83:414–430.

Tian et al., 2002b. Multiscale Analysis and Validation of the MODIS LAI Product. II.SamplingStrategy.RemoteSens.Environ.,83:431–441.

Tianetal.,2002c.RadiativetransferbasedscalingofLAI/FPARretrievalsfromreflectancedataofdifferentresolutions.RemoteSens.Environ.,84:143–159.

Wang et al., 2001. Investigation of product accuracy as a function of input andmodeluncertainities:CasestudywithSeaWiFSandMODISLAI/FPARAlgorithm.RemoteSens.Environ.,78:296–311.

Xu et al., 2018. Analysis of global lai/fpar products from viirs andmodis sensors forspatio-temporalconsistencyanduncertaintyfrom2012–2016.Forests,9(2),p.73.

Xuetal.,2020ImprovingleafareaindexretrievaloverheterogeneoussurfacemixedwithwaterRemoteSens.Environ.,240:111700.

Yanetal.,2016a.EvaluationofMODISLAI/FPARproductcollection6.Part1:Consistencyandimprovements.RemoteSens.,8(5),p.359.

Yanetal.,2016b.EvaluationofMODISLAI/FPARproductcollection6.Part2:Validationandintercomparison.RemoteSens.,8(6),p.460.

Yangetal.,2006a.AnalysisofLeafAreaIndexandFractionofPARAbsorbedbyVegetationProducts from the TerraMODIS Sensor: 2000–2005. IEEE Trans. Geosci. RemoteSens.,44:1829–1842.

Yanget al., 2006b.Analysisofprototype collection5productsof leaf area index fromTerraandAquaMODISsensors,RemoteSens.Environ.,104,297–312.

Yang et al., 2006c. MODIS Leaf Area Index Products: From Validation to AlgorithmImprovement.IEEETrans.Geosci.RemoteSens.,44:1885–1898.