IRfilezillaclient

MODISCollection5.
1BurnedAreaProduct-MCD45User'sGuideVersion3.
0.
1,May2013LuigiBoschetti,UniversityofIdahoDavidRoy,SouthDakotaStateUniversityAnjaA.
Hoffmann,LMUniversityofMunichMichaelHumber,UniversityofMarylandCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May2013iCONTACTSOFTECHNICALRESOURCEPERSONSTopicContactDataformatLuigiBoschetti-luigi@uidaho.
eduAlgorithmdescriptionDavidRoy-David.
Roy@sdstate.
eduGeotiffinArcGISMichaelHumber-mhumber@umd.
eduCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May2013iiABBREVIATIONSANDACRONYMSBABurnedAreaBRDFBidirectionalReflectanceDistributionFunctionCMGClimateModelGridEOS(NASA)EarthObservingSystemEOSDISEOSDataInformationSystemGeotiffGeoreferencedTaggedImageFileFormatHDFHierarchicalDataFormatLP-DAACLandProcessesDistributedActiveArchiveCenterMCD45MODISglobalburnedareaproductMODISModerateResolutionImagingSpectrometerNDVINormalizedDifferenceVegetationIndexSDSScienceDataSetSzSunZenithAngleQAQualityAssessmentVzViewZenithAngleCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May2013iiiTableofContentsCONTACTSOFTECHNICALRESOURCEPERSONS.
IABBREVIATIONSANDACRONYMSII1INTRODUCTION.
12TERMINOLOGYOFTHEMODISBURNEDAREAPRODUCT.
12.
1ProcessingLevels(fromtheEOSdataproductshandbook)12.
2Tiles.
22.
3Collections.
32.
3.
1Collection5.
1GlobalBurnedAreaProduct(MCD45A)ImprovementsoverthepreviousCollection5.
0Product.
32.
4MCD45Nomenclature.
33MCD45PRODUCTSUITE33.
1Level3MonthlyTiledProduct:MCD45A1.
43.
2GeotiffsubsetforGISvisualizationandanalysis:MCD45monthly.
43.
3ShapefilesubsetforGISvisualizationandanalysis:MCD45monthly.
63.
4ContentsoftheMODISBurnedAreaProduct63.
4.
1MCD45A1ScienceDataSets(SDS)63.
4.
2MCD45A1Metadata.
83.
4.
3Geotiffs.
93.
4.
4Shapefiles.
93.
5Forthcomingplannedformats94OBTAININGTHEPRODUCTS104.
1DownloadingtheproductsviaFTP104.
2DatastructureonFTPserver114.
2.
1HDFfiles.
114.
2.
2GeotifffilesandShapefiles115LOADINGANDPROCESSINGTHEPRODUCTINTOENVI4.
8.
115.
1MCD45A1(hdf)115.
2MCD45monthly(Geotiff)125.
3MCD45monthly(Shapefile)126LOADING,DISPLAYINGANDPROCESSINGTHEPRODUCTINARCGIS126.
1MCD45monthly(Geotiff)126.
1.
1AreaofInterest(AoI)136.
2MCD45montly(Shapefile)147CAVEATSANDKNOWNPROBLEMS158USEFULLINKSANDTOOLS158.
1LP-DAACTools159FREQUENTLYASKEDQUESTIONS15Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May2013iv10CITATION.
1611REFERENCES1612ALGORITHMBACKGROUND.
1912.
1TheBidirectionalReflectanceModel-basedExpectationApproach.
1912.
2TemporalImplementation.
2112.
2.
1IterativeProcedureforIdentificationofBurnedCandidates2213COORDINATECONVERSIONFORTHEMODISSINUSOIDALPROJECTION.
2513.
1WritingtheWorldfileforaMODIStile26ListofAppendicesAppendix1AlgorithmBackground18Appendix2CoordinateconversionfortheMODISsinusoidalprojection.
24Appendix3CoverageoftheGeotiffsubsets.
28ListofTablesTable1:RegionsandboundingcoordinatesoftheGeotiffsubsets.
29ListofFiguresFigure1:MODIStilingscheme.
2Figure2:CoverageoftheGeotiffsubsets.
5Figure3:DisplayofGeotifffromAugust2010,Window20.
13Figure4:ExportRasterdata.
14Figure5:DisplayofShapefilefromOctober2002,Window20.
15Figure6:CoverageofGeotiffsubsets30Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201311IntroductionMappingthetimingandtheextentoffiresisimportantasfireisaprominentchangeagentaffectingecosystemstructureandthecyclingofcarbonandnutrientsaswellasisaglobally-significantcauseofgreenhousegasemission.
Satellitedatahavebeenusedtomonitorbiomassburningatregionalandglobalscaleformorethantwodecadesusingalgorithmsthatdetectthelocationofactivefiresatthetimeofsatelliteoverpassandinthelastdecadeusingburnedareaalgorithmsthatmapdirectlythespatialextentoftheareaaffectedbyfires.
Previouslyandintheabsenceofaccurateburnedareaproducts,burnedareaassessmentshavebeencreatedbasedoncalibratingtheavailableactivefiredatafromregionalAVHRRandglobalMODISdata.
Howeverseveralremotesensing,environmental,andfirebehaviorfactorslimittheaccuracyofsuchderivedfireaffectedareadatasets.
Theavailabilityofrobustlycalibrated,atmosphericallycorrected,cloud-screened,geolocateddataprovidedbythelatestgenerationofmoderateresolutionremotesensingsystemsallowsformajoradvancesinsatellitemappingoffireaffectedarea.
AcomplementaryMODISalgorithmdefinedtomapfireaffectedareahasbeendevelopedandthefirstglobalburnedareaproductfrom2000onwardsisnowbeinggeneratedaspartoftheMODISLandcollection5productsuite.
ThisdocumentprovidespracticalinformationabouttheMODISBurnedAreaProduct(MCD45),includingproductandalgorithmdescription,orderinganddownloadprocedures,datalayerdescriptions,andknownproblems.
Theguidealsoprovidesafewexamplesonhowtoperformbasicoperationswiththeproductusingsomewidelyusedsoftwarepackages.
2TerminologyoftheMODISBurnedAreaProductThissectionexplainssomeofthestandardtechnicaltermsusedinthedocumentationofthestandardMODISproducts.
2.
1ProcessingLevels(fromtheEOSdataproductshandbook):Level0-Reconstructedunprocessedinstrument/payloaddataatfullresolution;anyandallcommunicationsartifacts(e.
g.
,synchronizationframes,communicationsheaders)removed.
Level1A-Reconstructedunprocessedinstrumentdataatfullresolution,time-referenced,andannotatedwithancillaryinformation,includingradiometricandCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May20132geometriccalibrationcoefficientsandgeoreferencingparameters(e.
g.
,platformephemeris)computedandappended,butnotapplied,totheLevel0data.
Level1B-Level1Adatathathavebeenprocessedtosensorunits(notallinstrumentshaveaLevel1Bequivalent).
Level2-DerivedgeophysicalvariablesatthesameresolutionandlocationastheLevel1sourcedata.
Level3-Variablesmappedonuniformspace-timegridscales,usuallywithsomecompletenessandconsistency.
Level4-Modeloutputorresultsfromanalysesoflowerleveldata(e.
g.
,variablesderivedfrommultiplemeasurements).
2.
2TilesMODISLevel2andhigherproductsaredefinedonaglobal250m,500m,or1kmsinusoidalgridwherespatialresolutionisdefinedbytheindividualproducts(Level0and1aregranule-based,asimpleunprojectedsegmentoftheMODISorbitalswath).
Gridsaredividedintofixed-areatilesofapproximately10degreesx10degreesinsize.
Eachtileisassignedahorizontal(h)andvertical(v)coordinaterangingfrom0to35and0to17respectively(Figure1).
Thetileintheupperleftcornerisdesignatedash0,v0.
Adetaileddescriptionofthesinusoidalprojection,andalltheformulasneededfortransformationsbetweentileandgeographiccoordinatesisprovidedinAppendixII.
Figure1:MODIStilingschemeCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201332.
3CollectionsReprocessingoftheentireMODISdataarchiveisperformedperiodicallysoastoincorporatebettercalibration,algorithmrefinements,andimprovedupstreamproductsintoallMODISproducts.
TheupdatedMODISdataarchiveresultingfromeachreprocessingisreferredtoasacollection.
Latercollectionssupersedeallearliercollections.
TheMCD45MODISBurnedAreaProductwasnotproducedinCollections1through4.
ItwasproducedforthefirsttimeaspartofCollection5,anditisavailableforthewholeMODISrecord(2000topresent).
2.
3.
1Collection5.
1GlobalBurnedAreaProduct(MCD45A)ImprovementsoverthepreviousCollection5.
0ProductTheCollection5.
1productgenerationalgorithmimprovesontheCollection5.
0algorithmby:-correctionofaknownissuerelatedtothehandlingoftheinputMODISthermalbanddata,whichwascausingomissionerrorsonsurfaceswithtemperaturesgreaterthan327.
7K.
-removalofallMODISband6(mid-IR,1550-1750nm)multi-temporaltests,whichwerecausingomissionerrorsovercertainforestandagriculturalareas.
-introductionofmono-temporalspectralteststoreduceburnedcommissionerrorsprimarilyassociatedwithagriculture.
-ingestofthelatestCollection5.
1MODISlandcoverproductthathasimprovedaccuracyandisgeneratedat500mratherthan1kmresolution-refinementofthe"BurnedAreaPixelQA"sciencedatasetstoincludenewsetting:QA=5toindicateburnedareasdetectedinagriculture,whichshouldbetreatedaslowconfidencebecauseoftheinherentissuesinmappingagriculturalburnedareasreliably.
2.
4MCD45NomenclatureAllMODISproductsareidentifiedbyanalphanumericcodeofthreelettersandtwonumbers.
Thethreelettersdefinethedatasource,whilethetwonumbersidentityuniquelytheproduct.
TheburnedareaproductisnamedMCD45:MCDindicatesthatbothMODISAquaandTerradataareusedasinput,and45istheidentifieroftheburnedareaproduct.
3MCD45ProductSuiteThreedifferentversionsoftheMODISburnedareaproductareavailable:Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201341.
TheofficialMCD45A1inHDF-EOSformat,whichisavailableaspartoftheMODISsuiteofgloballandproducts.
2.
There-projectedmonthlyGeotiffversionavailablefromtheUniversityofMaryland.
3.
There-projectedmonthlyShapefileversionavailablefromtheUniversityofMaryland.
3.
1Level3MonthlyTiledProduct:MCD45A1TheMCD45A1BurnedAreaProductisamonthlyLevel3gridded500mproductcontainingper-pixelburningandqualityinformation,andtile-levelmetadata.
Productusersareprovidedwithavarietyofqualityassessmentinformationandasinglesummaryqualityassessmentscoreforeachpixel.
Adetaileddescriptionoftheproductisprovidedinsection4.
AllthestandardMODISproductsadoptanomenclaturethatprovidesinformationaboutthespatialandtemporalcoverageoftheproduct,aswellasabouttheversionoftheproduct.
Anexamplefilenameoftheproductis:MCD45A1.
A2000306.
h16v07.
051.
2006360205845.
hdfwhichcanbereadas:a)MCD45A1=MODISlevel3Monthlytiledburnedareaproductb)A2000306=yearandJuliandateofthestartingdayofthemonthcoveredbytheproduct:306istheJuliandateofNov1,hence2000306meansthattheproductcoversNovember2000.
c)h16v07=spatialextent:thefilecoverstileh16v07d)051=versionidentifier.
051indicatesCollection5.
1e)2006360205845=processingdayandtime(year2006,Julianday360,20h58'45'')Thisnamingconventionensuresthatfilesalwayshaveauniquename:ifatileisreprocessedthelastnumber,indicatingthedayandtimeinwhichthefilewasprocessed,willbedifferent,thusavoidinganyconfusionwithobsoletedata.
TheJuliandayisdefinedasadatingsystemthatdividestheyearinto12monthsand365days(leapyear366days)andatablecanbefoundforexamplehere:regularyears:http://www.
quadax.
com/Support/Julian2009.
pdfleapyears:http://www.
quadax.
com/Support/Julian2008.
pdf3.
2GeotiffsubsetforGISvisualizationandanalysis:MCD45monthlyAuser-friendlyGeotiffversionoftheMCD45productisderivedfromthestandardMCD45A1hdfversionbyUniversityofMaryland.
TheGeotiffsarereprojectedinPlate-Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May20135Carréeprojectionandcoverasetofsub-continentalwindows(Figure2).
Atablecontainingtheregionscoveredandboundingcoordinatesofthe24windows(plusonesub-window)isavailableinAppendixIII.
Figure2:CoverageoftheGeotiffsubsets.
Ahighresolutionversionofthemap,aswellasatablewiththeboundingcoordinates,isavailableinappendixIII.
TheGeotiffsfollowanamingconventionsimilartotheofficialMCD45A1product.
However,astheGeotiffsareobtainedbymosaicing,resamplingandreprojectingseveraltilesoftheoriginalproduct,theprocessingtimeisnotavailable.
Anexampleoffilenameoftheproductis:MCD45monthly.
A2000306.
Win01.
051.
burndate.
tifMCD45monthly.
A2000306.
Win01.
051.
ba_qa.
tifwhichcanbereadas:a)MCD45monthly=monthlyGeotiffversionofMCD45A1b)A2000306=yearandJuliandateofthestartingdayofthemonthcoveredbytheproduct:306istheJuliandateofNov1,hence2000306meansthattheproductcoversNovember2000.
c)Win01=spatialextent:thefilecoverswindow1(Alaska)Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May20136d)051=versionidentifier.
051indicatesCollection5.
1e)burndate/ba_qa=contentofthefile:unlikehdf,Geotifffilescontainasinglelayer.
Atthemoment,twolayersoftheoriginalproductareavailableasGeotiffs:"burndate"and"QA".
See3.
4.
3fordetails.
3.
3ShapefilesubsetforGISvisualizationandanalysis:MCD45monthlyAshapefileversionoftheMCD45productisderivedfromthemonthlyGeotiffversionbytheUniversityofMaryland.
Theshapefilesareavailablewiththesameprojection(Plate-Carrée)andgeographicextentastheGeotiffsub-continentalwindows(Figure2).
TheshapefilenamingconventionisidenticaltotheGeotiffnamingconvention.
Eachshapefileconsistsofmultiplefileswhichmustremaininthesamelocationrelativetooneanother.
Anexampleofashapefile,asavailablefromtheUniversityofMarylandFTP,includesthefollowing:MCD45monthly.
A2000306.
Win01.
051.
burndate.
shpMCD45monthly.
A2000306.
Win01.
051.
burndate.
shxMCD45monthly.
A2000306.
Win01.
051.
burndate.
prjMCD45monthly.
A2000306.
Win01.
051.
burndate.
dbfwhichcanbereadas:a)MCD45monthly=monthlyshapefileversionofMCD45A1b)A2000306=yearandJuliandateofthestartingdayofthemonthcoveredbytheproduct:306istheJuliandateofNov1,hence2000306meansthattheproductcoversNovember2000.
c)Win01=spatialextent:thefilecoverswindow1(Alaska)d)051=versionidentifier.
051indicatesCollection5.
1e)burndate=contentofthefile.
Onlythe"burndate"layerisavailableintheshapefileformat.
See3.
4.
4fordetails.
3.
4ContentsoftheMODISBurnedAreaProductThefollowingtextdescribesthecontentofthefiles;someofthelayersoftheproductrequireknowledgeofhowthealgorithmusedfortheproductionworks;thedetaileddescriptionofthealgorithmcanbefoundinAppendixI.
3.
4.
1MCD45A1ScienceDataSets(SDS)TheproductisdistributedinthestandardMODISlandformat(HDF),andincludesthefollowingtendataScienceDataSets(i.
e.
layers),definingforeach500mpixel:Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201371.
Burndate(2bytes):ApproximateJuliandayofburningfromeightdaysbeforethebeginningofthemonthtoeightdaysaftertheendofthemonth,oracodeindicatingunburnedareas,snow,water,orlackofdata.
0-unburned1-366-approximateJuliandayofburning900–snoworhighaerosol9998-waterbodies(internal)9999-waterbodies(seasandoceans)10000-notenoughdatatoperforminversionthroughouttheperiod2.
BApixelQA(1byte):Confidenceofthedetection(1(mostconfident)to4(leastconfident)).
1-mostconfidentlydetectedpixels,regardlessofdirectionintime(forward,backwardorboth),passingtest(4)describedinappendix1.
2-pixelswherebackwardandforwarddirectionintimepredictthesamechange,passingtest(5)describedinappendixI.
3-pixelsselectedinthefirststageofthecontextualanalysis.
4-pixelsselectedinthesecondstageofthecontextualanalysis.
5-detectionsoveragriculturalareas,accordingtotheMCD12landcovermask.
Usersareadvisednottousethosedetectionsinanyquantitativeanalysis,duetothelowaccuracyofburnedareadetectionwhenfireisduetoagriculturalpractices.
3.
NumberofPasses(1byte):Numberofobservationswherethetemporalconsistencytestispassed.
4.
NumberUsed(1byte):Numberofobservationsusedinthetemporalconsistencytest.
5.
Direction(1byte):Directionintimeinwhichburningwasdetected(forward,backwardorboth).
1-forward2-backward3-both6.
SurfaceType(1byte):Informationdescribingthelandcovertypeandproperties.
Theinformationisstoredintheindividualbitsofthelayer.
bit0(1=yes,0=no):-water(NDVI65thresholdorSz>65)bit6(1=yes,0=no)-highviewandsolarzenithangle(Vz>50andSz>55)bit7(1=yes,0=no)-snowOR[highaerosol(fromMOD09QA)ANDhighview/solarzenith(Vz>55andSz>55)]Example:iftheSurfaceTypevalueofapixelis18,thecorrespondingbinarynumberis01001000;asbits1and4aresetto1,itmeanslowNDVIandcloudshadowdetected.
7.
GapRange1(2bytes):Informationdescribingthelargestnumberofconsecutivemissing/cloudydays(ifany)inthetimeseriesandthestartdayofthemissing/cloudyperiod.
bits0-8-Juliandayofthestartofthegapbits9-13-numberofmissingdaysincludingthestartdayExample:ifthepixelvalueis3372,thecorrespondingbinarynumberis001101001011000.
Thefirstninebits(001101001)representthenumber300,thefourfollowingbits(0110)representthenumber6.
HenceasixdaygapstartingatJulianday300.
8.
GapRange2(2bytes):Informationdescribingthesecondlargestnumberofconsecutivemissing/cloudydays(ifany)inthetimeseriesandthestartdayofthemissingperiod.
bits0-8-Juliandayofthestartofthegapbits9-13-numberofmissingdaysincludingthestartday3.
4.
2MCD45A1MetadataInadditiontothemandatorymetadatarequiredbytheEOSDataInformationSystem(EOSDIS)CoreSystem(ECS),asetofproductspecific,tile-levelmetadataareincludedtoenabletheburnedareaproducttobearchivedandorderedviaECSDAACorderingsystems.
Themetadatareportforeachtileincludes:Thepercentageoflandpixelsdetectedasburned.
Thepercentageofpixelsnotprocessedduetoinsufficientcloud-freedata.
Thepercentageofpixelsineachofthe"BApixelQA"categories.
Thenumberofpixelsdetectedineachdirectionintime(forward,backwardorboth).
Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201393.
4.
3GeotiffsOnlytwoSDSs(layers)areavailableinGeotiffformat.
Asexplainedin3.
2,foreachmonthtwofilesareavailable,oneforeachSDS:1.
Burndate(2bytes):ApproximateJuliandayofburningwithinthemonth,oracodeindicatingunburnedareas,snow,water,orlackofdata.
UnliketheHDFversionoftheproduct,theGEOTIFFSdonotincludeanyoverlapbetweenconsecutivemonths.
ThetemporalfilteringisdescribedinRoyetal.
,2008.
0-unburned1-366-approximateJuliandayofburningwithinthemonth900–snoworhighaerosol9998-waterbodies(internal)9999-waterbodies(seasandoceans)10000-notenoughdatatoperforminversionthroughouttheperiod-32768-pixelnotcoveredbyanyMODIStile2.
BApixelQA(1byte):Confidenceofthedetection(1(mostconfident)to4(leastconfident)).
1-mostconfidentlydetectedpixels,regardlessofdirectionintime(forward,backwardorboth),passingtest(4)describedinAppendix1.
2-pixelswherebackwardandforwarddirectionintimepredictthesamechange,passingtest(5)describedinsection4.
13-pixelsselectedinthefirststageofthecontextualanalysis4-pixelsselectedinthesecondstageofthecontextualanalysis3.
4.
4ShapefilesOnlythe"Burndate"SDSisavailableinshapefileformat.
Burndatevaluesarestoredinthe"burndate"fieldoftheshapefile.
UnliketheburndatevalueintheGeotiffversion,theshapefileburndatevaluesincludeonlytheapproximateJuliandayofburningwithinthemonth,withvaluesrangingfrom1-366.
AllothervaluesarediscardedintheshapefileversionoftheMCD45monthlyproduct.
Forthediscriminationbetweenunburnedandunmappedareas,usersarerecommendedtodownloadtheHDForGEOTIFFversionoftheproduct3.
5Forthcomingplannedformats1.
MCD45ClimateModelGrid(CMG)AggregatedBurnedAreaProductCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201310Thisproductiscurrentlyunderdevelopment.
Dataformatinformationwillbeincludedatthetimeofproductrelease.
2.
MCD45YearlySynthesisBurnedAreaProductThisproductiscurrentlyunderdevelopment.
Dataformatinformationwillbeincludedatthetimeofproductrelease.
4ObtainingtheproductsAllMODISproductsareavailablefreeofcharge.
TheMODISBurnedAreaProductisavailablefororderingfromtheLandProcessesDistributedActiveArchiveCenter(LP-DAAC)usingtheEOSDataGatewaywebinterfacelocatedat:http://reverb.
echo.
nasa.
govAdditionally,anftpserverismaintainedbytheUniversityofMaryland,mostlytoprovidesupporttothescienceuserswhoneedtodownloadsystematicallylargevolumesofdata.
4.
1DownloadingtheproductsviaFTPTheMODISburnedareaproductisavailablefordownloadviaftpfromthewebsitehttp://modis-fire.
umd.
edu/Werequestuserstofillinauseronlineformforstatisticalpurposesandinordertoobtainusernameandpasswordfortheserver.
Theformisavailableunderhttp://modis-fire.
umd.
edu/form.
aspYouareaskedtoenteryourname,affiliationandashortdescriptionoftheintendeduseoftheproduct.
Ausernameandpasswordissenttoyourgivene-mailaddress.
OnceyouhavereceivedyourusernameandpasswordyoucanstartretrievingthedataeitherinHDForGeotiffformat.
FordownloadingthedataviaFTPyoucanuseyourcurrentwebbrowsersuchastheFTPextensionofFirefoxorInternetExplorer.
HoweverwerecommendusingspecialFTPsoftwarefordownloadinglargeamountsofdata.
YoucanusefreelyavailablesoftwaresuchasFileZillaClientorSmartFTP,whichhastheadvantagethatyoucanscheduleyourdownloadtostartatalatertimeandifyoudownloadmultiplesources,itwilltrytodownloaddifferentportionsfromthemtospeedup.
http://filezilla-project.
org/download.
phphttp://www.
smartftp.
com/Whicheverprogramyouwillusefordatadownloadyouwillneedtoconnecttothegivenftpsitethatis:ftp://ba1.
geog.
umd.
eduCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May2013114.
2DatastructureonFTPserverTheHDF,Geotiff,andshapefiledatasetsareavailableonthesameFTPserver.
However,thesubdirectorystructureisdifferentandisdescribedinthefollowingparagraphs.
4.
2.
1HDFfilesThefilesystemontheftpserverisstructuredtoorganizethedatahierarchicallybyyearandbymonth.
Allthedatafromthesamemonthislocatedinadirectoryidentifiedbytheyearandmonthas:/Collection51/HDF/YYYY/DDD/where:YYYYistheyearDDDisthejuliandayofthebeginningofthemonthForexample,thedirectory/Collection51/HDF/2001/182containsalltheCollection5.
1tiles(namedwiththeconventionexplainedin3.
1)oftheproductforJuly2001.
4.
2.
2GeotifffilesandShapefilesThefilesystemontheftpserverisstructuredtoorganizethedatahierarchicallybywindow,andthenbyyear.
Allthedataforthesamewindowfromthesameyearislocated(forGeotifffilesandshapefiles,respectively)inadirectoryidentifiedas:/Collection51/TIF/WinXX/YYYY//Collection51/SHP/WinXX/YYYY/where:XXisthenumberofthewindow(figure2)YYYYistheyearForexample,thedirectory/Collection51/TIF/Win01/2001containsalloftheCollection5.
1Geotifffilesformonthsof2001forwindow01(Alaska).
5LoadingandprocessingtheproductintoENVI4.
85.
1MCD45A1(hdf)HDFMODISproductsareonlypartiallysupportedinENVI.
Toopenthem,insteadoftheFile->OpenImageFilemenu,itisnecessarytoselectFile->OpenExternalFiles->GenericFormats->HDFCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201312OpenedasgenericHDF,thefilesloseanygeographicinformation.
Torestoreit,itisnecessarytoentermanuallytheprojectionparameters.
Alternatively,itispossibletodownloadandinstalltheMODISconversiontoolkitdevelopedbyEXELIS:http://www.
exelisvis.
com/language/en-US/UserCommunity/CodeLibrary.
aspx5.
2MCD45monthly(Geotiff)TheGeotifffilesarefullycompatiblewithENVI.
Toopenthem,itispossibletogosimplythroughtheFile->OpenImageFilemenu.
5.
3MCD45monthly(Shapefile)ShapefilesarenotdirectlysupportedinENVI,rathertheyareconvertedtotheENVIVectorFile(.
evf)format,thendisplayedonthescreen.
Toloadashapefile,selectVector->OpenVectorFile->ChangefileformatstoShapefileorAll->SelectShapefileWhenpromptedtoset"ImportVectorFilesParameters",setthedesiredlayernameandoutputfilelocationandselect"OK".
Donotadjusttheprojectioninformation,thedefaultvaluescorrespondtothePlate-CarréeprojectionusedbytheGeotifffiles.
Usethe"AvailableVectorsList"(Vector->AvailableVectorsList)tooverlaythevectoronanexistingdisplayordisplayitinitsownwindow.
6Loading,displayingandprocessingtheproductinArcGISHandlingEOS-hdffilesisnotstraightforwardinArcGIS.
FortheusersoftheofficialEOS-HDFproduct,werecommendtoperformanyscientificanalysisinothersoftwarepackages(e.
g.
ENVI)andthenexporttoArcGISinadifferentformatsuchasGeotiff.
6.
1MCD45monthly(Geotiff)TheGeotiffscandirectlybeloadedintoArcGISthrough->AddLayerInordertodisplaytheburnedareasonly,under>LayerProperties>UniqueValues>SymbologySetthecolorto"NoColor"forthefollowingValues:0-unburned900–snoworhighaerosol9998-waterbodies(internal)Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May2013139999-waterbodies(seasandoceans)10000-notenoughdatatoperforminversionthroughouttheperiod-32,768–nodataArcGISwilldisplaytheburnedareainJuliandaysofthegivenmonthineitherindividualcolorsorinthesamecolordependingonyoursettings.
Figure3showsanexampleofMCD45monthly.
A2010213.
Win20.
005.
burndate.
tifFigure3:DisplayofGeotifffromAugust2010,Window20.
Eachburndayofthemonthisshownindifferentcolors6.
1.
1AreaofInterest(AoI)InordertoreducefilesizeoftheregionalGeotiffandfocusonaspecificregionan"AreaofInterest(AoI)"canbesubset.
->DisplaydesiredAoI->Rightmouseclickdatalayer->Data->ExportDataCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201314The"ExportRasterData"windowwillopen(figure4).
CheckbothdataframesoptionstocurrentandchooseFormat(TIFF).
Figure4:ExportRasterdata6.
2MCD45montly(Shapefile)TheShapefilescandirectlybeloadedintoArcGISthrough->AddLayerTochangetheappearanceofthefile,selectLayerProperties->Symbology->GraduatedColorsOptionally,toremovetheoutlinesofthedata,rightclickasymbolintheGraduatedColorspanelandselect"PropertiesForAllSymbols",thenset"OutlineColor"to"NoColor".
Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201315Figure5:DisplayofShapefilefromOctober2002,Window20.
Left:True-color(R,G,B)Landsat7sceneacquiredOctober30,2002atpath98,row71.
Right:MCD45monthlyOctoberShapefilesuperimposedovertheburnedarea.
Olderburnedareasareshowninbluetones,newerburnsareshowninredtones.
7CaveatsandKnownProblemsTheMODISinstrumenthadaprolongedoutageinJune2001.
Asaconsequence,notenoughinputdataisavailablefortheproductionofMCD45A1forthatmonth.
8UsefulLinksandTools8.
1LP-DAACToolsAcomprehensivesetoftoolsforhandling,convertingandreprojectingMODISproductsinhdfformatisavailableat:https://lpdaac.
usgs.
gov/tools9FrequentlyAskedQuestionsIcannotordertheMCD45A1productforJune2001.
WhyisitnotonlineTheMODISinstrumenthadaprolongedoutageinJune2001.
Asaconsequence,notenoughinputdataisavailablefortheproductionofMCD45A1forthatmonth.
IdownloadedsomeMCD45A1datainhdfformat.
WhenIopenit,Iseeseverallayers.
WhichoneshallIstartlookingatCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201316Formostoftheusers,therelevantinformationisthelocationoftheburnedpixels,andthedayofburning.
Thisdataiscontainedinthefirstlayer(SDSintheMODISjargon),called'burndate'.
WhenItrytoopenafileinENVI,Igetanerrormessage"unsupportedMODISproduct"YoucaneitheropenthefileasgenericHDFwith:File->OpenExternalFiles->GenericFormats->HDForinstalltheMODISconversiontoolkitfromtheEXELISwebsite:http://www.
exelisvis.
com/language/en-US/UserCommunity/CodeLibrary.
aspxIreceivedtheusernameandpasswordfortheftpserver,Iclickedonthelinkandnothinghappens.
WhatshallIdoFTPserversdonotworklikenormalwebpages.
Youneedanftpclient(alistisgiveninparagraph4.
1)andyouneedtoreadcarefulthedocumentationoftheftpclienttounderstandhowtodownloaddata.
10CitationTheMODISburnedareaproduct,andthedetailsofthealgorithm,aredescribedinthefollowingpapers.
TheyshouldbereferencedwhenusingtheMODISburnedareaproduct:D.
P.
Roy,L.
Boschetti,C.
O.
Justice,J.
Ju.
2008.
TheCollection5MODISBurnedAreaProduct-GlobalEvaluationbyComparisonwiththeMODISActiveFireProduct.
RemoteSensingofEnvironment,112,3690-3707.
(PDFfile,4.
5MB)Roy,D.
P.
,Jin,Y.
,Lewis,P.
E.
,Justice,C.
O.
2005.
Prototypingaglobalalgorithmforsystematicfire-affectedareamappingusingMODIStimeseriesdata.
RemoteSensingofEnvironment,97:137-162.
(PDFfile,4MB)RoyD.
P.
,LewisP.
E.
,JusticeC.
O.
2002.
Burnedareamappingusingmulti-temporalmoderatespatialresolutiondata-abi-directionalreflectancemodel-basedexpectationapproach.
RemoteSensingofEnvironment,83:263-286.
(PDFfile,2.
3MB)11ReferencesJournalArticlesCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201317W.
Lucht,ViewingtheEarthfrommultipleangles:Globalchangeandthescienceofmultiangularreflectance,ReflectionPropertiesofVegetationandSoilwithaBRDFDatabase,Eds.
M.
vonSchonermark,B.
Geiger,andH.
P.
Roser,WissenschaftundTechnikVerlag,Berlin,2004.
W.
LuchtandP.
E.
Lewis,"TheoreticalnoisesensitivityofBRDFandalbedoretrievalfromtheEOS-MODISandMISRsensorswithrespecttoangularsampling",InternationalJournalofRemoteSensing,vol.
21,pp.
81-98,2000.
W.
LuchtandJ.
L.
Roujean,"ConsiderationinparametricmodelingofBRDFandalbedofrommulti-angularsatellitesensorobservations",RemoteSensingReviews,vol.
18,pp.
343-379,2000.
D.
P.
Roy,L.
Giglio,J.
KendallandC.
O.
Justice,"Multitemporalactive-firebasedburnscardetectionalgorithm,"InternationalJournalofRemoteSensing,vol20,pp.
1031-1038,1999.
D.
P.
Roy,P.
E.
LewisandC.
O.
Justice,"Burnedareamappingusingmulti-temporalmoderatespatialresolutiondata–abi-directionalreflectancemodel-basedexpectationapproach,"RemoteSensingofEnvironment,vol.
83,pp.
263-286,2002.
D.
P.
Roy,Y.
Jin,P.
E.
LewisandC.
O.
Justice,"PrototypingaglobalalgorithmforsystematicfireaffectedareamappingusingMODIStimeseriesdata,"RemoteSensingofEnvironment,vol.
97,pp.
137-162,2005.
C.
B.
Schaaf,F.
Gao,A.
H.
Strahler,W.
Lucht,X.
Li,T.
Tsang,N.
Strugnell,X.
Zhang,Y.
Jin,J.
-P.
Muller,P.
E.
Lewis,M.
Barnsley,P.
Hobson,M.
Disney,G.
Roberts,M.
Dunderdale,R.
P.
d'Entremont,B.
Hu,S.
Liang,J.
Privette,andD.
P.
Roy,"FirstOperationalBRDF,AlbedoandNadirReflectanceProductsfromMODIS,"RemoteSensingofEnvironment,vol.
83,pp.
135-148,2002.
E.
F.
Vermote,N.
Z.
ElSaleousandC.
O.
Justice,"OperationalatmosphericcorrectionoftheMODISdatainthevisibletomiddleinfrared:Firstresults"RemoteSensingofEnvironment,vol.
83,pp.
97–111,2002.
P.
J.
Zarco-Tejada,C.
A.
RuedaandS.
L.
Ustin,"WatercontentestimationinvegetationwithMODISreflectancedataandmodelinversionmethods"RemoteSensingofEnvironment,vol.
85,pp.
109–124,2003.
Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201318Appendix1AlgorithmBackgroundCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May20131912AlgorithmBackgroundBurnedareasarecharacterizedbydepositsofcharcoalandash,removalofvegetation,andalterationofthevegetationstructure(Royetal.
1999).
TheMODISalgorithmtomapburnedareastakesadvantageofthesespectral,temporal,andstructuralchanges.
Thealgorithmdetectstheapproximatedateofburningat500mbylocatingtheoccurrenceofrapidchangesindailysurfacereflectancetimeseriesdata.
Itisanimprovementonpreviousmethodsthroughtheuseofabidirectionalreflectancemodeltodealwithangularvariationsfoundinsatellitedataandtheuseofastatisticalmeasuretodetectchangeprobabilityfromapreviouslyobservedstate(Royetal.
2005).
Thealgorithmmapsthespatialextentofrecentfiresonlyandexcludesfiresthatoccurredinpreviousseasonsoryears.
Thebidirectionalreflectancemodel-basedchangedetectionalgorithmdevelopedfortheMCD45productisagenericchangedetectionmethodthatisappliedindependentlytogeolocatedpixelsoveralongtimeseries(weekstomonths)ofreflectanceobservations(Royetal.
2002,Royetal.
2005).
Reflectancesensedwithinatemporalwindowofafixednumberofdaysareusedtopredictthereflectanceonasubsequentday.
Astatisticalmeasureisusedtodetermineifthedifferencebetweenthepredictedandobservedreflectanceisasignificantchangeofinterest.
Ratherthanattemptingtominimizethedirectionalinformationpresentinwidefield-of-viewsatellitedatabycompositing,orbytheuseofspectralindices,thisinformationisusedtomodelthedirectionaldependenceofreflectance.
Thisprovidesasemi-physicallybasedmethodtopredictchangeinreflectancefromthepreviousstate.
12.
1TheBidirectionalReflectanceModel-basedExpectationApproachMethodshavebeendevelopedtomodeltheBRDFwithalimitednumberofparametersandthentoestimatethemodelparametersfromafinitesetofremotelysensedobservations(Lucht2004).
Thesemi-empiricalRossThick-LiSparsereciprocalBRDFmodelisusedfortheMODISglobalburnedareaproductasitperformsrobustlyintheglobalMODISBRDF/albedoproduct(Schaafetal.
2002).
Likeotherlinearkernel-drivenmodelsitallowsanalyticalmodelinversionwithanestimateofuncertaintyinthemodelparametersandlinearcombinationsthereof(LuchtandRoujean2000,LuchtandLewis2000).
AteachgeolocatedpixelthethreeparameterRossThick-LiSparsereciprocalBRDFmodelisinvertedagainstm7reflectanceobservationssensedinatemporalwindowofn(16)daysduration.
TheBRDFmodelparametersareusedtocomputepredictedreflectanceanduncertaintiesfortheviewingandilluminationanglesofasubsequentobservation.
AZ-scoreisusedasanormalizedmeasurerelatedtotheprobabilityofthenewobservationbelongingtothesamesetasthatusedintheBRDFmodelinversion:Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201320(1)where:istheZ-scorevalue,isthenewreflectanceobservation,isthemodelpredictedreflectanceatwavelengthcomputedbyanalyticalinversionoftheBRDFmodelagainstpreviousreflectanceobservations,and'aretheviewingandilluminationvectorsrespectivelyofthenewreflectanceobservation,isafixedpre-assignedestimateofthenoiseindefinedbyVermoteetal.
(2002),eistherootmeansquaredoftheresidualsoftheBRDFinversion(usedasanestimateofnoiseintheobservationsandthelackofabilityofthemodeltofitthemeasurements),andwisthe'weightofdetermination'of(LuchtandLewis2000).
isadaptivetotheviewingandilluminationanglesofthenewobservation,aswellastheangulardistribution,amountofnoise,andnumberofobservationsusedintheBRDFinversion.
TheZ-scoreiscomputedforMODISbands2and5asthesebandsarebothsensitivetoburningandexperienceadecreaseinreflectancepost-fire.
Anewobservationisconsideredasaburncandidateif:(-(3)ANDCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201321->-whereisthenewreflectanceobservationandisthemodelpredictedreflectancecomputedbyanalyticalinversionoftheBRDFmodelagainstm7previousreflectanceobservations.
Thejustificationforequation(3)isthatburningcausesareductioninband2and5reflectancebutlesschangeinband7reflectance,whereaspersistentcloud,shadow,orsoilmoisturechangeswouldhaveasimilareffectinbothbands.
Band2helpstoremovechangesassociatedwithincreasingplantwatercontentwhichisnegativelyrelatedtoband5and7reflectancebutnotband2reflectance(Zarco-Tejadaetal.
2003).
InthisworkZthresh=3.
0todetectonlythosereflectancechangesthatfalloutsideoftheexpectedreflectancevariationmodeledfrompreviousvalues(theprobabilitythatZ<-3.
0is~0.
0013).
12.
2TemporalImplementationThecomputation(equations1-3)isrepeatedindependentlyforeachgeolocatedpixel,movingthroughthereflectancetimeseriesindailystepstodetectchange.
Atemporalconstraintisusedtodifferentiatebetweentemporarychanges,suchasshadows,undetectedresidualclouds,soilmoisturechangesanddataartifacts,thatpass(1)–(3)fromfire-affectedareasthathavepersistentlylowerpost-firereflectance.
Gapsinthereflectancetimeseries,forexampleduetocloudcoverorbadqualityinputdata,reducethetemporalfrequencyofZ-scorecalculationsastheyreducethenumberofobservationsavailableforpredictionandthenumberofwindowsthathavesufficientobservationsforBRDFinversion.
Toreducetheimpactofgaps,thedurationoftheBRDFinversionwindowisallowedtoincreaseandtheZ-scoreiscomputednotjustforthesubsequentdaybutforseveralsubsequentdays.
ThedurationoftheBRDFinversionwindowisallowedtoincrease,fromaminimumofn=16daysuptoamaximumof(n+nextra)days,untilthereareatleast7observations.
Whentherearefewerthan7observationsnoinversionisperformed.
Inthisway,moreBRDFinversionsmaybeperformedinthepresenceofmissingdata,providingmoreopportunitiesfordetectingburningevents.
Ateachwindowcontaining7ormoreobservationstheBRDFparametersareusedtocomputeZ-scoresforthenon-missingobservationssensedonthefollowingSsearchdays.
IfwithinthefollowingSsearchdaysaburncandidateisfound,i.
e.
criteria(1)-(3)aremet,thentheZ-scorescontinuetobecomputedforStestdaysafterthefirstburncandidate.
Foreachinversionwindow,thedaythatthefirstburncandidatewasdetected(Dayfirst),themaximumofitsband2and5Z-scores(Zfirst),andthetotalnumberofobservationsoverthesubsequentStestdaysthatwereconsidered(Nused)anddetectedasburned(Npass),arederived.
DifferentDayfirstcandidatesmaybedetectedduetosensitivityoftheadaptivewindowdurationandmulti-datepredictiontogapsinthetimeseries.
Inaddition,thesamegeolocatedpixelmayburnonseparatedates.
TheresultsfromthedifferentinversionwindowsarerankedwithrespecttoNpassandCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201322thenNusedtoprovideresultsinorderofthemostevidenceofpersistentburning.
IfthereareresultswithequalNpassandNusedvaluesthentheonewiththegreatestZfirstisrankedasmorepersistent.
SearchingbothforwardandbackwardintimeallowsburncandidatestobedetectedintheSsearchdaysprecedingorfollowingperiodsofpersistentlymissingdata.
ThisalsoallowsburncandidatestobedetectedinthefirstandlastSsearchdaysofthetimeseries.
Resultsfortheforwardandbackwarddirectionsarederivedindependently.
Whensearchingbackwardintime,anincreaseinreflectanceintheappropriateMODISbandsissearchedforratherthanadecreaseinreflectance.
12.
2.
1IterativeProcedureforIdentificationofBurnedCandidatesTheglobalalgorithmattemptstoreduceerrorsofcommissionbyselectingonlyburnedpixelswherethereareburncandidatesthatprovidepersistentevidenceoffireoccurrence.
Asthemeasuredpersistencevariesdependingongapsinthereflectancetimeseriesandthetimingofthefirerelativetonon-missingdata,aniterativeratherthansimplethresholdingapproachisused.
Burncandidatesfoundinboththeforwardandbackwarddirectionsareconsidered.
First,burnedpixelsareselectedasoccurringonDayfirstif:Npass3AND(Npass/Nused)0.
5ANDNinv3(4)Inthiswayonlycandidatesareselected,regardlessofthedirectionofthedetection,whereatleast50%oftheobservationsconsideredoverthesubsequentStestdaysaredetectedasburnedandatleast3inversions(Ninv)areusedfortheconsistencytest.
Ifseveralburncandidatesarefoundatagivenpixel,thentheyareconsideredinorderofdecreasingevidenceofpersistentburningandthefirstonethatpassescondition(4)isselected.
Ifforwardandbackwardsearchresultshaveequalpersistencethentheforwarddirectionresultsaregivenprecedence.
Incloudyregions,evenconfidentlydetectedburncandidatesmighthaveinsufficientdatafor3inversionswithinthetimeframeoftheconsistencytest.
Asaconsequence,if–andonlyif–backwardandforwardpredictthesamechange,burnedpixelsareselected,regardlessofNinvusingthelessrestrictivetest:Npass3AND(Npass/Nused)0.
5(5)Second,ratherthandiscardburncandidatesthatarelikelyburnedbutdonotpassconditions(4)and(5)duetoinsufficientobservations,theyareconsideredusinglessrestrictivecriteriathan(4)or(5)inaniterativesearchmethod.
Thismethodisbasedontheprinciplethatthereisincreasingexpectationofaburnoccurringinpixelsneighboringconfidentlydetectedburns(Royetal.
2002,Graetzetal.
2003).
Inthissearchprocedure,theburncandidatesselectedby(4)and(5)areconsideredseedpixels.
Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201323Inthefirstsetofiterations,non-seedpixelswhereburncandidatesweredetectedthatdidnotpassconditions(4)or(5)areacceptedasburnediftheyhavetwoormoreadjacentseedneighborsandif:Dayfirst-Dayfirst_seed25(6)whereDayfirst,Npass,andNusedarethevaluesfortheburncandidatethatdidnotpassconditions(4)or(5)andDayfirst_seedisthemeanDayfirstvalueofthetwotoeightadjacentseedpixels.
TheNgapconstraintensuresthatonlyburncandidatesthatoccurtemporallyaswellasspatiallyclosetotheneighboringseedpixelsareconsidered.
Thisprocedureisrepeatedinanexhaustiveiterativemannerwiththepixelsthatpassedcondition(6)beingconsideredasseedsforthenextiterationuntilnomoreburncandidatesthatpass(6)canbeincluded.
Aswithcondition(4),ifseveralburncandidatesarefoundatagivenpixelthentheyareconsideredinorderofdecreasingevidenceofpersistentburninguntil(6)ismet.
Again,ifforwardandbackwardsearchresultshaveequalpersistencethentheforwarddirectionresultsaregivenprecedence.
Inthesecondpartoftheprocedure,theresidualburncandidatesnotselectedinthepreviousstepsareconsideredifatleastthreeneighborshavebeenselected.
Theaveragedayofburningoftheneighborsiscomputed,andthepixelisacceptedifthebackwardCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201324Appendix2CoordinateconversionfortheMODISsinusoidalprojectionCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May20132513CoordinateconversionfortheMODISsinusoidalprojectionMostsoftwarepackagesdonotsupportfullytheMODISprojecteddata;thisappendixprovidessomeusefulformulasfor:definingtheMODISmapprojectionandassociatingaWorldfiletoaMODIStilecalculatingtheimagecoordinatesofageographicpoint(directproblem)calculatingthegeographiccoordinatesofaspecificpointofanimage(inverseproblem).
TheMODISdataarere-projectedusinganequiarealsinusoidalprojection,definedonasphereofradius=6371007.
181m,andwiththeGreenwichmeridianasthecentralmeridianoftheprojection.
Defining(x,y)astheEastandNorthcoordinateinmetersofapointinthemapspace,and(,)itslatitudeandlongitudeindegrees,thedirectformulasis:(1)(2)andtheinverseformulasderivefrom(1)and(2)throughsimplemathematicalpassages:(3)()(4)Therangeofxandycanbedeterminedeasily:xmin=x(=0,=-180)=-20015109m(5)xmax=x(=0,=180)=20015109m(6)ymin=y(=-90,=0)=-10007554m(7)ymax=y(=90,=0)=10007554m(8)ForthedistributionofMODISdata,thesinusoidalmapspaceispartitionedinto36x18tilesoffixedlocation,eachofthemcorrespondingto1200x1200"1km"pixels,or2400x2400"500m"pixelsor4800x4800"250m"pixels.
Equations1and2allowtheconversionfromthegeographicspace,tothemapspace.
However,topassfromthemapspacetotheimagecoordinatesweneedafewmoresteps.
Eachpixelisidentifiednotbytwo,butbyfourcoordinates:(i,j)beingthecolumn,rowcoordinatesofthepixelwithinthetile,and(h,v)beingthehorizontalandverticalcoordinatesofthetile.
As5though8providetherangeofxandy,thesizetofthegriddefiningthetilescanbecalculatedas:Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201326(9)Dividingbythenumberofpixelsineachtile,weobtainthattheexactpixelsizeforthelevel3MODISdata:(10)wherenpixels={1200;2400;4800}isthesizeinpixelsofatile.
ForthevariousMODISresolutionsweobtain:926.
625mforthe"1km"pixels463.
313mforthe"500m"pixels234.
656mforthe"250m"pixelsTheformulasthatlink(h,v,i,j)to(x,y),foratileofare:()(11)()(12)[()](13)[()](14)Theinverseformulasaretrivial:(15)(16)Byusing(1)through(4)and(11)through(16)it'spossibletoperformanyconversionbetweengeographic,mapandimagecoordinatesforMODISlevel3data.
13.
1WritingtheWorldfileforaMODIStileHDFdataarenotfullysupportedbymostsoftwarepackages,andthegeographicinformationiseasilylost.
InordertoassociatethegeographicinformationtoaMODIStile,ortoanyrasterderivedthroughfurtherprocess,itisnecessarytoknowtheparametersoftheprojection,theexactpixelsizeandthecoordinatesofthecorneroftheimage.
Itisoftennecessarytoknowthecoordinatesofthecornerinthemapprojection(i.
e.
x,yinmetres)ratherthanthegeographiccoordinates.
Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201327AwidelyusedformatinGISapplicationstostorethemapinformationistheWorldFile.
Aworldfileisatextfilecontainingthepixelsize,therotationinformationandthecoordinatesoftheupperleftcorner.
Theworldfilehasthesamenameasthefilecontainingthedata,andanextensionmadebythefirstandthirdletteroftheoriginalextension,plustheletter"w".
Forinstance,'filename.
tif'willbeaccompaniedby'filename.
tfw'and'filename.
gif'willbeaccompaniedby'filename.
gfw'ForagenericMODIS(h,v)tile,thesixlinesoftheworldfileare:line1:pixelsize,from(10)line2:0line3:0line4:negativeofthepixelsize,from(10)line5:upperleftxcoordinate,from(15)withi=0line6:upperleftycoordinate,from(16)withj=0Forexample,theworldfilefortileh08v05is:463.
312716600-463.
3127166-11119273.
5414447570.
423Collection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201328Appendix3CoverageoftheGeotiffsubsetsCollection5MODISBurnedAreaproductUserGuideVersion3.
0.
1,May201329Table1:RegionsandboundingcoordinatesoftheGeotiffsubsets.
Window#CoverageMinLonMaxLonMinLatMaxLat1Alaska-180-140.
550702Canada-141.
5-5040703USA(Conterminous)-125-6523504CentralAmerica-118-587335SouthAmerica(North)-82-34-10136SouthAmerica(Central)-79-34-35-107SouthAmerica(South)-77-54-56-358Europe-113533709WestandNorthAfrica-195037.
510CentralandNorthAfrica525037.
511EastAfricaandArabianPenisula2565037.
512SouthernAfrica(North)8.
548-155.
512BTanzania29.
540.
6-12-0.
613SouthernAfrica(South)1041-35-1514Madagascar4259-27-1015RussiaandCentralAsia13590337016RussiaandCentralAsia290145337017Russia(Kamachatka)145180407018SouthAsia609353619SouthEastAsia90155-103320Australia112155-45-1021NewZealand165179-48-3322Azores-31.
6-24.
836.
84023CapeVerdeIsland-25.
5-22.
514.
617.
524Hawaii-161-1541824Collection5MODISBurnedAreaproductUserGuideVersion3.
0October201230Figure6:CoverageofGeotiffsubsets