Hierarchicaldataorganization
People’sperceptionoftherealworldisoftenlimitedtoasmallareaoracertainindividualobject.TheserealworldobjectsarecalledAnentityisanobjectivelyexistingobjectwithcertaincharacteristics.Atypeofentitywithcertaincommoncharacteristicscanbecalledanentitygroup;inaddition,therearemanygeographicalphenomenaonthesurface,whichcanbedescribedbyfields,forexample,Terrainelevation,airtemperature,soilmoisture,etc.;accordingtothedatatypeofthedependentvariable,thefieldcanbedividedintotwotypes:continuousfieldanddiscretefield.Infact,anygeographicalphenomenoncanbeexpressedthroughthefield.However,inmanycases,amethodcanprovideabetterbasisforeffectivecalculations,anditmayalsobeabettermodelforpeople'spsychologicalexpressionofgeographicphenomena.Forexample,thestateorprovinceofacountryisusuallyexpressedinGISasagroupofareaobjectsoragroupoflineobjectsformingitsboundary;ofcourse,itcanalsobeexpressedasadiscrete2-dimensionalfield.Entityandfieldcananswerdifferentquestions(whereandwhat)respectively.Ofcourse,theprocessofansweringquestionsmustbebasedontime(when).ThiswayofcognitionresultsinahierarchicaldataorganizationmethodinGIS,thatis,thestructureofgeographicentitiesintomathematicalpoints,lines,areas,andgridcells(grid).
Thehierarchicalclassificationtheoryisoneofthebasictheoriesforpeopletoperceivetherealworld."Layer"isanimportantbasicconceptinGIS,and"Layer"isoneofthemostbasicandimportantmethodsofGISdataorganization,asshownintheupperdiagraminFigure1.Hierarchyinthevectormodelisaccompaniedbyclassification(mainlyreferstotheclassificationofabstractgeometricelements,suchaspoint,line,surface,andbody,etc.).Therastermodelismainlyhierarchical,andeachlayerofdataafterlayeringhasCorrespondingattributesandspaceandotherinformation,thelogicalorganizationmodelisshowninthefigurebelow.
GISdataisorganizedbyseveralspatialdatalayersandtheirrelatedattributedata.Aspatialdatalayerisstoredintheformofseveralspatialcoordinatesorrasterpixels.Thislogicalorganizationmodelcanbesummarizedascoordinatepair-spatialobject-layer-map.Aspatialobjectanditsattributeinformationbelongtothemostbasiclevelinthismodel,andthemapisthehighestlevelofthismodel.Theinformationinthegeographicdatalogicalorganizationmodelcanbeclassifiedasfollows:
(1)AtlasThisisthetoplevelinformationinthegeographicdataorganization,whichmanagestheentireatlasandprocesseseachmap,Toidentifydifferentmaps,includingthefollowinginformation:mapreference(tablename,maplayernumber,etc.),mapcoordinates(coordinatesystem,registrationinformation,etc.),mapdescription(accessrights,mapdescription,etc.)andotherinformation.
(2)LayersetThisisalayersetcomposedofmultiplespatiallayersthatcanmeetcertainapplicationrequirements,includingthelayerreferences(layerlabel,layerTablename),layerspaceindex(size,label,tablename),layerdisplay,layercoordinaterange(maximumandminimumcoordinates)andotherinformation.
(3)LayerThisisacollectionofmultiplespatialobjectsofthesametypewithcertainsameorsimilarcharacteristics,includingtheidentification(label,name)anddescriptionofthespatialobject(Name,characteristicattribute,type),geometricrepresentationofspatialobjects(binarylargeobject-BLOBformofcoordinates).
TheaboveisthedataorganizationmethodofGISverticallayering.Thislayereddataorganizationismap-oriented.Inthedatabase,thelibrary—mapsheet—layer—geographicalobject—geometricobjectisgenerallyused.Dataorganizationstrategy.Horizontally,GISdataisorganizedbyframingordividinggrids(tiles),andthentheyarespatiallyindexed.Thiskindofdataorganizationisrealizedinthecomputerthroughtheseparationofgraphichierarchicalfilemanagementandattributedatarelationaldatabase(RDBMS)management.Therearetwoways:oneistoembedRDBMSintotheGISsystemInternal;theotheristouseanextendedRDBMStomanageattributedatainadditiontotheinternalRDBMS.TheconnectionbetweenspatialdataandattributedataiscompletedbythekeyID.
Feature-baseddataorganization
Thedescriptionofgeographicphenomenabasedonhierarchicaldataorganizationhasthefollowingshortcomings:
(1)Real-worldspatialgeometrytargetsAbstractionignorestheessentialcharacteristicsofgeographicphenomenaandtheirinternalconnections,andartificiallydividestherealworld,whichresultsinthesimplificationofGISinformationandreducesthecapacityofGISinformation.
(2)PayattentiontothedescriptionofspatiallocationsThevectororrasterdataorganizationmodelofthecompanylosestherichanalysiscapabilitiesprovidedbythestructuredentitiesbasedontheclassificationattributesandinterrelationships;
(3)TheoverlapmethodTherealworldisdividedintoaseriesoflayerswithstrictboundaries,buttheseboundariescannotfullyreflecttheobjectivereality,whichcausesmanyhumanerrors;inaddition,thismethodcannotprovidethespatialanalysiscapabilitiesofmanybasicobjects.
Actually,inthehierarchicaldataorganization,point,line,surface,andgridunitsdonotexist.Inreality,roadsarenotmathematicallines,andcitiesarenotmathematicalpoints.Thisabstractionisnotadescriptionorexpressionofrealgeographicspace.Theperceptualworldaroundusisasetofmaterialentitieswithhighlycorrelatedstructures,thatis,theseentitieshaveasetofcommonattributesthatallowpeopletoclassifyonthebasisofsimilarity.Therecognitionofthecommonalityofgeographicentityattributesandrelationshipsisthestartingpointofpeople'scognition(Usery1993).Itcanbeseenthatpeople'sfirstknowledgeoftheobjectiveworldisbasedongeographicfeatures,andthiswayofcognitioncreatesamethodofdataorganizationbasedongeographicfeatures.ISO/TC211andOGCrespectivelydefinegeographicfeatures:
(1)AccordingtothedefinitionofISO/TC211,featureshavetwolevels:featuretypesandfeatureinstances.Featuretypesaregeographicphenomenawithcommonattributes.Afeatureinstanceisaspecificgeographicphenomenonofafeaturetype;eachfeatureinstancehasauniqueidentifier,whichisencapsulatedwithattributes,functionsandrelationships,andcancomprehensivelydescribethecharacteristicsoftheoccurrenceanddevelopmentofthefeaturetype.Symbols,aseriesofcharacteristicstatesoreventscanbeorganicallylinkedandorganizedtogether,andcanbepositionedinthespace-timecoordinatesystem,whichisconducivetothemanagementandqueryofspace-timedata;
(2)DefinitionofOGC,Featureisthebasicunitofgeospatialinformation;Inaddition,USGISintheUnitedStatesalsogivesitsowndefinitionoffeature:Featureisanentityortarget(digital/orgraphic)expressionoftheobjectiveworld.Itcanbeseenthatafeatureisadescriptionorexpressionofarealgeographicalphenomenon.Thisgeographicalphenomenoncanbearealgeographicalentity,suchasariverorlake,oritcanbeaclassificationresult,suchasdifferenttypesofland,orItisameasurementresultofacertainphenomenon,suchashightemperaturearea,highrainarea,etc.Therefore,ageographicfeatureisageographicentitythatexistsobjectivelyontheearthandhasdescriptiveinformation,andthisgeographicentitycanbedefinedbyitsidentificationanddescriptionofitsattributesandrelationships.
Thebasisoffeature-basedGISdataorganizationisfeatureclassification.Itdirectlyaffectstheeffectivenessofgeographicdataorganization,management,queryandanalysis;itaffectsthecompletenessofgeographicdatamodelsemanticsanddatasharing.Therefore,feature-basedGIScanbeconstructedusingobject-orientedtechnology.Itsdataorganizationframeworkneedstousetherelevantconceptsofcognitiveclassificationtheoryandtherelevantmethodsofcartography.Thismethodofdataorganizationrequiresacorrectandappropriategeographicclassificationsystem.Whilefollowingthegeneraltaxonomyprinciples,thesystemmustalsoconsidertheneedsofGIStechnology(suchasobject-orientedtechnology).Itrequiresthattheclassificationsystembeincorporatedintoanon-spatialattributeInthedeterminedspacesystem.
Comparativeanalysisofthetwomethods
Thehierarchicaldataorganizationandthefeature-baseddataorganizationareatthesamelevelofabstraction.Botharebasedontheentitymodelandthefieldmodel,butFeature-baseddataorganizationusesobject-orientedtechnicalmethodstoorganizedataonthebasisofobject-orienteddatamodels,whilehierarchicaldataorganizationmainlyuseshierarchiesonthebasisofvectordatamodels,rasterdatamodels,andrelationaldatamodels.Themethodtoorganizedata;althoughwiththecontinuousdevelopmentandimprovementoftechnicalmeans,thelayereddataorganizationmethodhasalsoinfiltratedtheobject-orientedtechnology,butthisdoesnotconstituteatrueobject-orienteddatamodel.Itcanbeseenthatthereisafundamentaldifferencebetweenthetwo.
AsshowninFigure2,thesamegeographicphenomenonusesahierarchicaldataorganizationmethodtobedividedintoatleast3layers(points,lines,andareas),whileusingadataorganizationmethodbasedongeographicfeaturesonlyrequires1layer.,Andthespatialrelationshipbetweenthemisobvious.However,whetheritishierarchicalorfeature-baseddataorganization,itisinseparablefromtheanswerstothreebasiccognitivequestions,namelywhere,whatandwhen.
Spatialdataorganization
Intermsofdatamanagement,GISwillstorespatialdataandattributedatainthegeometricsenseseparately,andusefilesanddatabasemanagementrespectively.ThefuturedevelopmenttrendisapplicationObject-orientedmathematicalmodelandspatialdatawarehousetechnologythatusesadatabasetomanagemassiveamountsofdatainarealsense.
Followingtheaboveprinciples,thelarge-scaleGISdataorganizationthatmeetstherequirementsofthe"digitalearth"isasfollows:
(1)Vectordata:Storedseparatelyaccordingtomapframeandscale.
(2)Rasterdata:regulargriddatawithacertaingridspacing(suchasDEM,imagedata,etc.),anditsdataorganizationisdifferentfromvectordata.Duetotheinfluenceofmapprojection,whenthedataisstoredingeographiccoordinates(latitudeandlongitude),thenumberofgridscontainedineachstandardframingmapisthesame,andthedataformsastandardrectangle;andwhenthedataisstoredinrectangularcoordinates(x,y)Whenstoring,thequantitycontainedineachstandardframingmapformsatrapezoid(approximately).Inordertosolvetheproblemofdataconnection,generallythecircumscribedrectangleofthefigureprofileisusedastheboundarytoformastandardrectangle,andthedataoutsidethefigureprofileFillitwithadjacentmapframedata,asshowninFigure3.
Duetotheunequalsizeoftheoriginaldata,dataconnectionproblemsareoftenencounteredduringdataroaming.Therearemanywaystosolvethisproblem.Amorefeasiblemethodistore-splicethedataandcutdatablocksofuniformshapeandsize.Inthisway,indataroaming,thesplicingofadjacentdatablocksisextremelysimple,whichavoidsdataredundancy.Iimprovedtheefficiencyofthesystem.Thedatacuttingrulesare:
①Placealldataunderthespatialcoordinateframeofthe"digitalearth".
②Regardlessofanyscaleandgridspacing,eachdatafile(adatablock)hasauniformsize.
③Breaktheconceptofscaleanduseresolution(gridspacing)asthedatascale.
④Establishfiledescriptioninformation,andrecordtheresolution,projectionzonenumber,filelocationandotherrelatedinformationofeachfileforeasysystemcall.
(3)IntegrationofvectordataandrasterdataDuetothedifferentdatastructure,dataorganizationanddisplaymethodsofvectorandraster,theintegrationofvectordataandrasterdatarequiresthefollowingprocess:
①Datamatchinganddisplay:includingunifiedprojection,scalematchingandcoordinatematching.
②Theconversionofvectordataandrasterdata:Thisconversionincludestheconversionofpoints,lines,andareas.Theconversionfromvectordatatorasterdataisrelativelysimple.Inpracticalapplications,thisconversionisoftenused.