UnderstandingRasterData

Introduction
Thefollowingdocumentisintendedtoprovideabasicunderstandingofrasterdata.
Rasterdatalayers(commonlyreferredtoasgrids)aretheessentialdatalayersusedin
alltoolsdevelopedbytheNationalInteragencyFuelsTechnologyTeam(NIFTT).Ifyou
areanexperiencedArcGISuserandwellfamiliarwithrasterdata,youmayskipthis
primerorreviewitasarefresher.

Whatisrasterdata?
Initssimplestform,arasterconsistsofamatrixofcells(orpixels)organizedintorows
andcolumns(oragrid),asshowninthegraphicbelow,whereeachcellcontainsa
valuerepresentinginformation,suchastemperature.Rastersaredigitalaerial
photographs,imageryfromsatellites,digitalpictures,orevenscannedmaps.

Datastoredinarasterformatrepresentrealworldphenomenasuchas:
Thematicdata(alsoknownasdiscretedata),representingfeaturessuchasland
useorsoilsdata.LANDFIREdatalayersdepictingfuels,vegetation,fire
regimes,andotherfeaturesarealsoexamplesofthis.
Continuousdatarepresentingphenomenasuchastemperatureandelevation
dataorspectraldata,including,forexample,satelliteimagesandaerial
photographs.
Pictures;examplesincludescannedmapsordrawingsandbuilding
photographs.

Thematicandcontinuousrastersmaybedisplayedasdatalayersalongwithother
geographicdataonamap,buttheyareoftenusedasthesourcedataforspatialanalysis
withtheArcGISSpatialAnalystextension.Picturerastersareoftenusedasattributesin
tablestheycanbedisplayedwithgeographicdataandareusedtoconveyadditional

1
 UnderstandingRasterData

informationaboutmapfeatures.

Whilethestructureofrasterdataissimple,itisexceptionallyusefulforawiderangeof
applications.WithinaGIS,theusesofrasterdatafallunderfourmaincategories:

1)Rastersasbasemaps
AcommonuseofrasterdatainaGISisasabackgrounddisplayforotherfeature
layers.Forexample,orthophotographsdisplayedunderneathotherlayersprovide
themapuserwithconfidencethatmaplayersarespatiallyalignedandrepresentreal
objects,asshownintheimagebelow.Threemainsourcesofrasterbasemapsare
orthophotosfromaerialphotography,satelliteimagery,andscannedmaps.

2)Rastersassurfacemaps
Rastersarewellsuitedforrepresentingdatathatchangescontinuouslyacrossa
landscape(surface).Theyprovideaneffectivemethodforstoringthecontinuityasa
surface.Theyalsoprovidearegularlyspacedrepresentationofsurfaces.Elevation
valuesmeasuredfromtheearthssurfacearethemostcommonapplicationofsurface
maps,asdepictedinthegraphicbelow.Othervalues,suchasrainfall,temperature,
concentration,andpopulationdensity,canalsodefinesurfacesthatcanbespatially
analyzed.

2
 UnderstandingRasterData

3)Rastersasthematicmaps
Rastersrepresentingthematicdatacanbederivedfromanalyzingotherdata.A
commonanalysisapplicationistheclassificationofasatelliteimagebylandcover
categories.Basically,thisactivitygroupsthevaluesofmultispectraldataintoclasses
(suchasbiophysicalsettingsshowninthegraphicbelow)andassignsacategorical
value.Thematicmapscanalsoresultfromgeoprocessingoperationsthatcombine
datafromvarioussources,suchasvector,raster,andterraindata.Forexample,auser
canprocessdatathroughageoprocessingmodeltocreatearasterdatasetthatmaps
suitabilityforaspecificactivity.MostoftheLANDFIREdatalayersarederivedin
thismanner.

4)Rastersasattributesofafeature
Rastersusedasattributesofafeaturemaybedigitalphotographs(seeimagebelow),
scanneddocuments,orscanneddrawingsrelatedtoageographicobjectorlocation.A
parcellayermayhavescannedlegaldocumentsidentifyingthelatesttransactionfor
thatparcel,oralayerrepresentingcaveopeningsmayhavepicturesoftheactualcave

3
 UnderstandingRasterData

openingsassociatedwiththepointfeatures.

Whystoredataasaraster?
Sometimesthereisnochoiceastohowdataarestored;forexample,imagerymayonly
beavailableasaraster.However,therearemanyotherfeatures(suchaspoints)and
measurements(suchasrainfall)thatcouldbestoredaseitherarasterorafeature
(vector)datatype.

Followingisalistoftheadvantagesofstoringdataasaraster:
AsimpledatastructureAmatrixofcellswithvaluesrepresentingacoordinate
andsometimeslinkedtoanattributetable
Apowerfulformatforadvancedspatialandstatisticalanalysis
Theabilitytorepresentcontinuoussurfacesandperformsurfaceanalysis
Theabilitytouniformlystorepoints,lines,polygons,andsurfaces
Theabilitytoperformfastoverlayswithcomplexdatasets

Storagespacemustbeaconsiderationwhenworkingwithrasters,astheycanbe
potentiallyverylargedatasets.Resolutionincreasesasthesizeofthecelldecreases;
however,costnormallyalsoincreasesinbothdiskspaceandprocessingspeeds.Fora
givenarea,changingcellstoonehalfthecurrentsizerequiresasmuchasfourtimesthe
storagespace,dependingonthetypeofdataandstoragetechniquesused.Thereisalso
alossofprecisionthataccompaniesrestructuringdatatoaregularlyspacedrastercell
boundary.

Generalcharacteristicsofrasterdata
Inrasterdatasets,eachcell(whichisalsoknownasapixel)hasavalue.Thecellvalues
representthephenomenonportrayedbytherasterdatasetsuchasacategory,
magnitude,height,orspectralvalue.Thecategorycouldbealanduseclasssuchas

4
 UnderstandingRasterData

grassland,forest,orroad.Amagnitudemightrepresentgravity,noisepollution,or
percentrainfall.Height(distance)couldrepresentsurfaceelevationabovemeansea
level,whichcanbeusedtoderiveslope,aspect,andwatershedproperties.Spectral
valuesareusedinsatelliteimageryandaerialphotographytorepresentlight
reflectanceandcolor.

Cellvaluescanbeeitherpositiveornegative,integer,orfloatingpoint.Integervalues
arebestusedtorepresentcategorical(discrete)data,andfloatingpointvaluesto
representcontinuoussurfaces.AllLANDFIREdatalayersarestoredasintegervalue
rasters.CellscanalsohaveaNoDatavaluetorepresenttheabsenceofdata.

Acellvalueappliestothecenterpointofthecellandtotheentireareaofthecell,as
depictedbelow,dependingontherasterapplication.

Thearea(orsurface)representedbyeachcellconsistsofthesamewidthandheightand
isanequalportionoftheentiresurfacerepresentedbytheraster(seegraphicbelow).
Forexample,arasterrepresentingelevation(thatis,digitalelevationmodel)maycover
anareaof100squarekilometers.Iftherewere100cellsinthisraster,eachcellwould
representonesquarekilometerofequalwidthandheight(thatis,1kmx1km).
LANDFIREdatacellsare30metersby30meters,witheachcellorpixelrepresentingan
areaof900squaremeters,or.2224acres.

5
 UnderstandingRasterData

Thedimensionofthecellscanbeaslargeorassmallasneededtorepresentthesurface
conveyedbytherasterdatasetandthefeatureswithinthesurface,suchasasquare
kilometer,squarefoot,orevenasquarecentimeter.Thecellsizedetermineshowcoarse
orfinethepatternsorfeaturesintherasterwillappear.Thesmallerthecellsize,the
smootherormoredetailedtherasterwillbe.However,thegreaterthenumberofcells,
thelongeritwilltaketoprocess,anditwillincreasethedemandforstoragespace.Ifa
cellsizeistoolarge,informationmaybelostorsubtlepatternsmaybeobscured.For
example,ifthecellsizeislargerthanthewidthofaroad,theroadmaynotexistwithin
therasterdataset.Inthediagrambelow,youcanseehowthissimplepolygonfeature
willberepresentedbyarasterdatasetatvariouscellsizes.

Rasterattributetables
Rasterdatasetsthatcontainattributetablestypicallyhavecellvaluesthatrepresentor
defineaclass,group,category,ormembership.Forexample,asatelliteimagemay
haveundergoneaclassificationanalysistocreatearasterdatasetthatdefineslanduses.
Someoftheclassesinthelanduseclassificationmaybeforestland,wetland,cropland,
andurban.Thenumbersbelowcouldrepresentwhichcellvalueintherasterdataset
woulddefinethelanduse:

6
 UnderstandingRasterData

1Forestland
2Wetland
3Cropland
4Urban

Bybuildingarasterattributetable,youcanmaintainthistablesattributeinformation
withthisclassifiedrasterdataset,aswellasdefineadditionalfieldstobestoredinit.
Forexample,theremaybespecificcodesassociatedwiththoseclassesorfurther
descriptionsofwhatthoseclassesrepresent.Youmayalsowanttoperform
calculationsontheinformationinthetable.Forexample,youmaywanttokeep
recordsofthetotalarearepresentedbythoseclassesbycalculatingthenumberofcells
multipliedbytheareaeachcellrepresents.Youcanalsojointherasterattributetableto
othertables.

Thegraphicbelowillustratesarasterdatasetwithattributetable.TheNoDatavalues
arenotcalculatedintherasterattributetable.Therearealsothreecolumnsthatare
calculatedbydefault;theothercolumnscanbeaddedindividuallyorbyusingajoin
operation.

Whenarasterattributetableisgenerated,therearethreedefaultfieldscreatedinthe
table:OID,VALUE,andCOUNT.Itisnotpossibletoeditthecontentinthesefields.
TheObjectID(OID)isauniquesystemdefinedobjectidentifiernumberforeachrowin
thetable.VALUEisalistofeachuniquecellvalueintherasterdatasets.COUNT
representsthenumberofcellsintherasterdatasetwiththecellvalueintheVALUE
column.CellvaluesrepresentedbyNoDataarenotcalculatedintherasterattribute
table.

Insummary,understandingthestructureandfunctionofrasterdataisanimportant
foundationforworkingwithLANDFIREdataandthesuiteofNIFTTtools.For

7
 UnderstandingRasterData

additionalinformationregardingrasterdata,pleaseconsultArcGISDesktopHelpor
visittheESRIsupportwebsiteatsupport.esri.com.

PortionsofthisworkincludetheintellectualpropertyofESRIandareusedhereinunderlicense.Aspects
oftheArcGISDesktopHelptextandgraphicshavebeenusedandeditedfortrainingpurposesby
AcadiaWestLLC.Copyright2009ESRI.Allrightsreserved.