47

47

GeoSciML – A GML Application forGeoscience Information Interchange

ByStephenM.RichardandCG��nteroperabilityWorkingGroup

U.S.GeologicalSurveyandArizonaGeologicalSurvey416W.Congress#100

Tucson,AZ85701Telephone:(520)770-3500

Fax:(520)770-3505e-mail:steve.richard@azgs.az.gov

CG� �nteroperability Working Group is (in alphabetic order): Eric Boisvert,BoyanBrodaric,SimonCox,TimDuffy,JonasHolmberg,BruceJohnson,JohnLaxton,TomasLindberg,StephenRichard,AlistairRitchie,FrancoisRobida,MarcusSen,JeanJacquesSerrano,BruceSimons,andLesleyWyborn.

INTRODUCTION

TheGeoSciMLapplicationisastandards-baseddataformatthatprovidesaframeworkforapplication-neutralencodingofgeosciencethematicdataandrelatedspatialdata.GeoSciMLisbasedonGeographyMarkupLan-guage(GML,Coxetal.,2004)forrepresentationoffea-turesandgeometry,andtheOpenGeospatialConsortium(OGC)ObservationsandMeasurementsBestPractices(Cox, 2006) for observational data. Geoscience-specific aspectsoftheschemaarebasedonaconceptualmodelforgeoscienceconceptsandincludegeologicunit,geologicstructure,andEarthmaterialfromtheNorthAmericaDataModel(NADMC1,NorthAmericanGeologic-MapDataModelSteeringCommittee,2004),andboreholeinforma-tionfromtheeXplorationandMiningMarkupLanguage(XMML,https://www.seegrid.csiro.au/twiki/bin/view/Xmml/WebHome).Developmentofcontrolledvocabularyresourcesforspecifyingcontenttorealizesemanticdatainteroperabilityisunderway.

TheintendedscopeforinitialversionsofGeoSciMLincludesinformationtypicallyfoundongeologicmapsaswellasinformationtypicallyrecordedwithboreholes.ThepossibleusesforGeoSciMLincludetransporting,storing,andarchivinginformation.Amongstthese,themost significant is transport—or information exchange—whichenablesinformationtobevisualized,queried,anddownloadedinspatialdatainfrastructures.ThisroleforGeoSciMLisparticularlyimportant,asgeoscienceinfor-mationconsumersarebecomingmoredigitallysophisti-cated and are no longer satisfied with images and portray-alsofdata,butwantdigitaldatainstandardizedformatsthatcanbeusedimmediatelyinapplications.Hours,days,orweeksspentmergingdatasetsobtainedseparatelyfrommultipleagenciesistimewasted.Useofastandardized

markupforserializinggeoscienceinformationsupportsacommitmentbydataproviderstopublishdatatousersinastandardizedformat.Thus,GeoSciMLallowsapplica-tionstoutilizegloballydistributedgeosciencedataandinformation.

TheGeoSciML(https://www.seegrid.csiro.au/twiki/bin/view/CG�Model/GeoSciML)projectwasinitiatedin2003undertheauspicesoftheCommissionfortheManagementandApplicationofGeoscience�nforma-tion(CG�)workinggrouponDataModelCollaboration(https://www.seegrid.csiro.au/twiki/bin/view/CG�Model/WebHome).TheCG�isacommissionofthe�nternationalUnionofGeologicalSciencesandhastheobjectivetoenabletheglobalexchangeofgeoscienceinformationforlegal, social, environmental, and geoscientific reasons. TheprojectispartofwhatisnowknownastheCG��n-teroperabilityWorkingGroup(https://www.seegrid.csiro.au/twiki/bin/view/CG�Model/�nteroperabilityWG),whichhas the specific objectives to:

• develop a conceptual model of geoscientific infor-mationthatdrawsonexistingdatamodels,

• implementanagreedsubsetofthismodelinanagreedschemalanguage,

• implementanXML/GMLencodingofthemodelsubset,

• developatestbedtoillustratethepotentialofthedatamodelforinterchange,and

• identify areas that require standardized classifica-tionstoenableinterchange.

GeoSciMLdrawsfrommanygeosciencedatamodeleffortsandfromthemestablishesacommonsuiteoffea-turetypesbasedongeologicalcriteria(units,structures,fossils)orartifactsofgeologicalinvestigations(speci-

48 D�G�TALMAPP�NGTECHN�QUES‘06

mens,sections,measurements).Supportingobjectsarealsoconsidered(timescale,lexicons,etc),sothattheycanbe used as classifiers for the primary objects. Predecessor projects that have had a strong influence on the develop-mentofGeoSciMLincludeactivitiesundertakenwithinnationalstatutorybodies(e.g.,theUSGS/AASGNationalGeologicMapDatabase,BritishGeologicalSurvey,andJapaneseGeologicalSurvey)inmulti-jurisdictionalcontexts(theNorthAmericanDataModel,http://nadm-geo.org/,forgeologicalmaps),andactivitiesorientedtoanindustrysector(eXplorationandMiningMarkupLanguage–XMML,https://www.seegrid.csiro.au/twiki/bin/view/Xmml/WebHome).Currently,severalexter-nal projects are leveraging GeoSciML for more specific applications,includingWaterResourcesmonitoringandmanagement,Soils,GeotechnicalandEngineering,AssayData,andGeochemistry.

Thisreportsummarizestheschemaandinstancedocumentsasimplementedinatestbeddemonstratedatthe�AMGmeetinginLiege,BelgiuminSeptember,2006.Theworkinggroupmetsubsequenttothetestbeddemonstration and has identified a number of aspects of themodelandschemainneedofupdate,aswellasmodelelementsthatneedtobeadded.Anticipatedchangesarediscussedhereaswell.Version1.1isthecurrentversionofthemarkuplanguage,withschemaavailableathttps://www.seegrid.csiro.au/subversion/xmml/GeoSciML/tags/1.1.0/schema/.Planningisunderwayforevolutionoftheschematoversion2toexpandthescopeandclarifysomeofthetoplevelmodelissues.Workinggroupactiv-ityiscurrentlyfocusedinseveraltask-groups(pendingformalization):

• Use-casesandrequirementstaskgroup,responsibleforsettingtechnicalgoals.

• Designtaskgroup,responsibleforthestructural

andsyntacticaspectsofthe“�nformationModel”ofaGeoSciML-basedservicearchitecture.

• Servicearchitecturetaskgroup,responsibleforthe“ComputationalModel”ofGeoSciML-basedservicearchitecture.

• Concepts definition task group, responsible for the“SemanticModel,”whichwillbeastandardsetofconcepts(ontology)forthecontentusedtopopulateGeoSciML,andwillfacilitatesemanticinteroperabilitywithGeoSciML.

• �mplementationtestbedtaskgroup,responsibleforliaisonwithGeoSciMLDesignandServiceAr-chitecturetaskgroupstoensurethatrequirementsare satisfied and coordinate and deliver TestBed3 demonstratingtheGeoSciMLv.2use-cases.

• Outreachandtechnicalassistancetaskgroup,responsibleforprovidingadviceandassistancetodirectcollaborators,assistingthemtodeployconformantGeoSciMLservices.

GEOLOGIC MAP DATA SCHEMATIC INTEROPERABILITY

Thedevelopmentofstandardizedmarkuplanguagesisacriticalstepnecessarytoachieveinteroperability,which is defined by ISO/IEC 2382-01 (SC36 Secretariat, 2003)tomean:“Thecapabilitytocommunicate,executeprograms,ortransferdataamongvariousfunctionalunitsinamannerthatrequirestheusertohavelittleornoknowledgeoftheuniquecharacteristicsofthoseunits.”Technicalrequirementstomeetthisgoalincludesys-tem-levelsharedprotocolsfornetworkcommunication,resourcediscovery,andserviceinvocation(Figure1).Applicationsthatusetheseprotocolsmustcommunicateby way of a shared data language that defines how infor-mationwillbeencoded.GeographyMarkupLanguage

Figure 1.Multiplelevelsofinteroperability(BrodaricandGahegan,2006).

49GEOSC�ML–AGMLAPPL�CAT�ONFORGEOSC�ENCE�NFORMAT�ON�NTERCHANGE

(http://www.opengeospatial.org/standards/gml)isthedatalanguageadoptedforGeoSciMLdevelopment.GMLprovides a framework for encoding geometry, defining featuresandassociatingthemwithproperties(includinggeometry),andconstructingdictionariesinwhichcon-trolled vocabularies can be defined.

GeoSciMLisaGMLapplicationscheme,whichisdefined by a collection of XML schema that utilize and extendelementsfromGMLtorepresentstandardgeologicobservationsanddescriptionsinageospatialcontext.GeoSciML is not a database structure. GeoSciML defines aformatfordatainterchange(Figure2).AgenciescanprovideaGeoSciMLinterfaceontotheirexistingdatabasesystems,withnorestructuringofinternaldatabasesrequired.

Thesemanticlevelofinteroperability(Figure1)requiresagreementonthemeaningofwordsusedtoexpresspropertyvaluescontainedinGeoSciMLelements.DevelopingcommonmeaningsforGeoSciMLcontentsthatcanbeappliedtovariousmulti-lingualvocabular-iesisaplannedfutureactivity.Atpresent,weanticipatethatimplementationofschematicinteroperabilitywilldemonstratetheneedfordatacontentstandardstoenablesemanticinteroperability.

IMPLEMENTATION

GeoSciMLwasdevelopedbyrepresentativesfromaninternationalgroupofgeologicmapdataprovidersinaseriesofface-to-facemeetingsandonlinediscussion(seeTwikiathttps://www.seegrid.csiro.au/twiki/bin/view/CG�-Model/GeoSciML).Onedesignobjectivewastore-use,revise,andextendexistingstandardswhereverpossible.Thedesignphilosophyofthisinterchangeformathasfocusedonanaccuraterepresentationofgeosciencein-formationinageneralway.Thisresultsingreatrepresen-tational flexibility at the price of complexity and verbose encoding.Fortunately,text-basedXMLcompressesveryefficiently, and the markup is designed for machine input andoutput,nothumanreadability.

ModeldevelopmenthasutilizedUMLnotationwithaUML profile to enable systematic mapping from UML to XMLschema.ThemappingfromUMLmodelstoGMLisdescribedinhttps://www.seegrid.csiro.au/twiki/bin/view/AppSchemas/UmlGmlandhttps://www.seegrid.csiro.au/twiki/bin/view/AppSchemas/UmL2GMLAS.AdetailedprocedureforgeneratingaGML-compliantXMLschemaissummarizedinhttps://www.seegrid.csiro.au/twiki/bin/view/AppSchemas/HollowWorldand

Figure 2.CommunicationbetweendataprovidersandconsumersutilizesstandardGeoSciMLschema.ClientsthatcaninterpretGeoSciMLcanoperatewithanyGeoSciML-enableddatasource.

50 D�G�TALMAPP�NGTECHN�QUES‘06

http://www.seegrid.csiro.au/twiki/bin/view/AppSchemas/OandMCookbook.SeealsoBoisvertetal.(2004)fromtheUSGSDMT2004workshop.Useofastandardgraph-icalnotationformodelrepresentationduringdevelopmentmakesgroupanalysisandreviewoftheevolvingmodelmucheasier.

Major Entities

OnlyasmallpartoftheGeoSciMLmodelisdis-cussedhere.SeetheGeoSciMLTwiki(https://www.seegrid.csiro.au/twiki/bin/view/CG�Model/GeoSciML)formoreinformationaboutthefullmodel.Figure3presentsthelogicalframeworkthatunderliesthedraftGeoSciMLversion2GeologicFeatureimplementation,acoreaspectofGeoSciML.Startingfromthecenterleft,aMappedFeatureassociatesaGeologicFeaturewithaGML_geometry that specifies a location on or within theEarth.ThemappedfeaturemaybetheresultofanObservationifobservation-relatedmetadataconcerningidentification of the mapped feature are recorded. Each GeologicFeatureisassociatedwithaControlledConceptclassifier that specifies the intention of what the Geolog-icFeaturerepresents.AGeologicFeaturemayhaveoneormoreassociatedGeologicFeatureDescriptionsthatspecifypropertiesassignedtothefeature.EachdescriptionmayalsoberepresentedastheresultofanObservation.Table1summarizesthepackagesincludedintheGeoSciMLUMLmodel.EachpackageisimplementedasaseparateXMLschema.

Geologic Feature

�nthedraftGeoSciMLversion2model,Geolog-icFeatureisanassociationclassthatbindsmapped

feature(s) and description(s) with one or more classifica-tionconcepts.Geologicfeatureisanentitythatrepre-sentssomeparticularphenomenonthatmaybeobservedin the Earth. It has a primary classification in terms of acontrolledconcept,andthisassociationestablishesacontentmodelorconceptspacewithinwhichthefeatureis located/given identity by specification of a collection ofpropertiesinadescription.AMappedFeatureinstancespecifies a particular located occurrence of a geologic featurebyassociatingitwithalocation(GML_geometry).GeologicFeatures may be classified by geologic unit or geologicstructureControlledConceptsterms.�nadditionto its primary classification (e.g. a lithostratigraphic desig-nation), a feature may carry alternative classifications (e.g. geotechnical classification). GeologicFeature corresponds witha“legenditem”fromatraditionalgeologicmapandwith“occurrence”inconceptualmodelspresentedbyBrodaricandGahegan(2006)orRichard(2006).Geo-logicFeaturesmayhaveoneormoreassociatedGeolog-icFeatureDescriptions.Multipledescriptionsassociatedwithafeaturemaybetheresultofdifferentobservations(differentobserver,differenttime,differentobservationprocedure…),ormayspecifydifferentproperties

Mapped Feature

A MappedFeature is a specific bounded occurrence, suchasanoutcropormappolygonthatcarriesageometryorshape(throughitssamplingFrameassociation).�thasan associated GeologicFeature instance that specifies what kindofthingisrepresentedbythemappedgeometry,bothby classification with a vocabulary term (ControlledCon-cept)andthroughassociationwithoneormoredescrip-tionobjects(GeologicUnitDescription)thatspecifypropertyvalues.

Figure 3.CoreGeoSciML2.0logicalmodel.

51GEOSC�ML–AGMLAPPL�CAT�ONFORGEOSC�ENCE�NFORMAT�ON�NTERCHANGE

Table 1.PackagesinGeoSciMLmodel.

Package Name Contents

TopLevel Thecoremodelformappedentitiesdistinguishesbetweengeologicfeatures,mappedfeatures,andcontrolledconcepts

BasicTypes Simple package, extends gml:MeasureType to represent quantification of measurements using relativecomparisons,e.g.greaterthan,lessthan.

LiteralValue TheGeoSciML“value”modelprovidesagenericwayofencoding“literal”values,bothtextualandnumeric,whichhaveuncertaintyandmaybearange.Thesevaluesareusuallyobtainedastheresultofanobservation.Thedescriptionoftheassociatedobservationeventwillprovidemoredetailabouttheobservationmethod,resultquality,etc.

RootDoc GenericcollectionelementforpackagingobjectsfromtheGeoSciMLschema.GeologicMetadata �nterimmodelforrepresentationofdataset,feature,andattribute-levelmetadata.�SO19115

metadatawouldbepreferred,buttheXMLimplementation(�SO19139)isnotyetsupportedbycommonsoftwareTheinterimmodelisintendedtohavesomeforwardcompatibilitywith�SO.Forexample,thescope-codesareasubsetofMD_ScopeCodefrom�SO19115.

GeologicVocabulary Modelforcontrolledvocabulariesoftermslinkedtonormativedescriptions,linktoontology.AGeologicVocabularyisacollectionofterms(ControlledConcept)andrelationships(VocabRelation).

BoreHole SupportforboreholedatainGeoSciMLisprovidedbyXMMLcomponents.Boreholeismod-eled as a kind of sampling profile that may have various sorts of associated ‘logs’, modeled as kindsofcoverages.

EarthMaterial EarthMaterialisaclassthatholdsamaterialdescription.AnaturallyoccurringsubstanceintheEarth.EarthMaterialrepresentssubstance,andisthusindependentofquantityorlocation.Ideally, Earth Materials are defined strictly based on physical properties, but because of standard geologicalusage,geneticinterpretationsenterintothedescriptionaswell.

GeologicAge Theageofaparticulargeologicaleventorfeatureexpressedintermsofyearsbeforepresent(absoluteage),referredtothegeologicaltimescale,orbycomparisonwithothergeologicaleventsorfeatures(relativeage).AGeologicAgecanrepresentaninstantintime,anintervaloftime, or any combination of multiple instants or intervals. Specifications of age in years before presentarebasedondeterminationoftimedurationsbasedoninterpretationofisotopicanalysesofEarthMaterial(someothermethodsareusedforgeologicallyyoungmaterials).Agesreferredtogeologicaltimescalesareessentiallybasedoncorrelationofageologicalunitwithastandardchronostratigraphicunitthatservesasareference.Relativeagesarebasedonrelationshipsbe-tweengeologicalunitssuchassuperposition,intrudedby,cross-cuts,or‘containsinclusionsof’.

GeologicRelation GeologicRelationsaretyped,directedassociationsbetweengeologicobjects.RepresentsanyofawidevarietyofrelationshipsthatcanexistbetweentwoormoreGeologicFeatures.Forexample,theGeologicRelation‘intrudes’isarelationshipbetweenanintrusiveigneousrockandsomehostrock.�ncludesspatial,temporal,sequence,correlation,andparent/childrelations.

TwoormoreGeologicFeaturesareassociatedinaGeologicRelation;eachhasaroleintherelationship.Examplesofgeologicalrolesinclude“overlies”,“isoverlainby”,“isyounger”,“isolder”,“intrudes”,“isintrudedby”,andsoforth.�narelationshipwhereanigneousunitintrudesasedimentaryunit,thegeologicalrelationshipis‘intrudes’,theintrudedsedimentaryunithastherole‘host’,andtheigneousunithastherole‘intrusion’.Manyothertypesofrela-tionshipscanalsobeaccommodatedviaGeologicRelation,forexample,topologicalrelationsbetween spatial objects could be described where they are scientifically significant.

GeologicTime TheGeoSciMLGeologicTimescalemodelandencodingisdescribedindetailinthepaper‘Aformalmodelforthegeologictimescaleandglobalstratotypesectionandpoint,compatiblewithgeospatialinformationtransferstandards’(CoxandRichard,2005).

Theclassic“geologicaltimescale”isahierarchicalordinalsystem,inwhichtheerasareranked:“stages”nestwithin“series”within“systems”within“eras”within“eons”(inthemostcommonversionoftherankingsystem).

GeologicUnit Packagecontainingcontentmodelforgeologicunit.Geologicunitisanotionalunit,whosecom-pleteandpreciseextentisinferredtoexist.Practically,spatialpropertiesareonlyavailablethroughassociationwithaMappedFeature.�ncludesbothformalunits(i.e.formallyadoptedandnamedinthe official lexicon) and informal units (i.e. named but not promoted to the lexicon) and unnamed units (i.e. recognizable and described and delineable in the field but not otherwise formalized).

StructureObject Packagecontainingcontentmodelforgeologicstructure.Version1includesfaultsystem,fault,contact,andfaultdisplacement.

52 D�G�TALMAPP�NGTECHN�QUES‘06

Controlled Concept

ControlledConceptsrepresenthumanconceptsincomputerform,typicallyaswords(lexicalobjects)withan associated definition. Because GeoSciML extends GML,eachcontrolledconceptinstancemayhaveoneormoregml:names,buttheGeoSciMLmodeladdsapre-ferred name element that specifies one term that is used to identifytheconcept.Practicallyspeaking,eachpreferred-Nameshouldbeassociatedwithauniqueconcept,butinadistributedsystem,thiscardinalitycannotbeenforced.ControlledConceptsareaggregatedintoGeologicVocabu-larycollections,whicharederivedfromaGMLdiction-ary.DataproducersshouldensurethatpreferredNamesareuniquewithinaparticularvocabulary.AControlledConceptmayhaveanassociatedprototypeentity(notshowninFigure3)thatcanbeaGeologicFeature,Earth-Material,orSpecimen.Theprototypeentityprovidesamechanismtoassociatemachine-analyzablepropertieswithControlledConceptterms.SimilarfunctionalitymightbeprovidedbylinksfromtheControlledConcepttosomeotherformalontologyrepresentation.

Geologic Feature Description

Descriptionsarecollectionsofpropertieswithas-signedvalues(e.g.attributes)thatcharacterizesomefeature.Differentkindsofdescriptionsspecifydifferentproperties.DescriptionsmaybeassociatedwithObserva-tionelementsthatsupplyinformationontheoriginofthepropertyvalueassignments.

Observation

Observationdescribesthe“metadata”associatedwithaninformationcaptureevent,togetherwithavaluefortheresultoftheobservation.Observationsarethebasisforclassified features, interpretations, and models. GeoSciML usestheObservationandMeasurementmodelfromtheOpenGeospatialConsortium(Cox,2006),whichmodelsobservationasakindofevent,inwhicharesultvalueisassignedtosomepropertyofafeatureofinterest,usingsomeprocedure.

SOME SCHEMA DESIGN AND USAGE PATTERNS ISSUES

Names and Identifiers

AnyGMLObjectorFeaturemayhaveanunlimitednumber of gml:name properties, which reflects the fact that the same object often has different identifiers assigned bydifferentauthorities.Toassert“thisisthenameoridentifier assigned by authority XYZ corporation,” use the codeSpace attribute on gml:name (i.e. the scope identifier).

If the codespace for the gml:name is not specified, then the valueisimplicitlyundertheauthorityoftheorganizationorservicethatsuppliesthedocument,whichshouldbeindicatedbyassociateddocument-levelmetadata.

NotethatGMLdocumentelementsalsoincludeagml:idattribute,whichplaysadifferentrolefromthegml:nameelement.Thevalueofthegml:idhastype=”xsd:�D”,soitmustbeuniquewithinthe(XML)document.�tis a document fragment identifier that acts as a handle for anXMLelementinthescopeofitsappearancewithinaparticulardocument,andisusuallyassignedbytheinfor-mation management system since it is primarily signifi-cantinthatcontext.Thegml:idsupportscross-referenceswithinadocumentandreferencesthatinvolveindividualnodes(elements)withinasystemofdocuments.Thevalueofagml:namehastype=”gml:CodeType”,whichisastringwitha“codeSpace”attribute.�nthecontextofaGMLobject,thevalueofagml:nameisalabeloridenti-fier for the object described by the containing element, andistypicallyassignedbythedataprovideragency.Thegml:name should be used for identifiers that are required tobepersistentandaresubjecttoconstraints(e.g.unique-ness)applicabletoacontextwiderthanjustthedocumentscope.Differentauthoritiesmayhavedifferentauthorita-tive identifiers for the same item.

Namespace and Packaging

ThenamespaceforGeoSciMLversion2.0schemaishttp://www.cgi-iugs.org/xml/GeoSciML/2.Versioningstrategyfornamespaceevolutionwillfollowpracticede-scribedinOGC05-062r3.Forfutureupgrades,eachmi-norversionofanysuchschemathatretainsthenamespaceofthepredecessorshallnotintroduceanynewXMLtypesorelementsthatcouldnotbesafelyignoredbyexistingapplicationbasedonthepreviousminorversion,whichensuresastrongformofbackwardcompatibility.Componentsfromothernamespaces(e.g.,http://www.opengis.net/om)mayalsoconstitutea“canonical”partofGeoSciMLbutwillbeincorporatedusingtheWXSimportmechanismand,thus,retaintheirownnamespacenames.

Thephysicaldocumentlocation(path)forGeoSciMLschemawillincludethecompleteversionnumber—ini-tially 1.0.0, moving to 1.0.x for bug-fix releases, and 1.1.x (etc.)forextensionsthatdonotchangethescopeoftheschema.SchemadocumentsarehostedintheGeoSciMLpublish/buildrepository,whichisathttps://www.seegrid.csiro.au/subversion/xmml/GeoSciML/tags/.

Use of Scoped Names

Useofscopednames,i.e.,atermorwordwithanidentifier for the source of the term, provides a method for linkagetoformalcontrolledvocabularies(e.g.anontol-ogy)thatmaythenbeusedforsemanticmediation.For

53GEOSC�ML–AGMLAPPL�CAT�ONFORGEOSC�ENCE�NFORMAT�ON�NTERCHANGE

example, a GeoSciML file might have a property value specified by the following element:

<CGI_TermValue> <qualifier>common</qualifier> <valuecodeSpace=”http://www.iugs-

cgi.org/outcropCharacterVocabulary”>ledgeforming</value>

</CGI_TermValue>

The<value>elementcontainsascopedname“ledgeforming” from the vocabulary specified by the codeSpace attribute.�fthedatainterpreterisfamiliarwiththe“http://www.iugs-cgi.org/outcropCharacterVocabulary”codeSpace(vocabulary),thentheymayusethescopednamedirectlyorbycorrelationwithapreferredterminadifferent vocabulary. On the other hand, if the identified codeSpace (vocabulary) is not familiar and its identifier is aresolvableURLthatpointstoservicethatcanprovidea definition of the term in a known format (e.g. free text, OWL,K�F...),itispossibletointerprettheterm.Thismaybe as simple as someone studying a free text definition anddeterminingtheclosestcorrespondingtermintheirvocabulary.Anautomatedsemanticmediatormightbeable to use a formal definition (e.g., OWL) to match with theclosestsubsumingterminadifferentformalvocabu-larythatispreferredbythedatainterpreter.

Value specification

The GeoSciML data model includes a flexible value specification scheme that is designed to capture value descriptionsconventionallyrecordedbygeologists.All

values may carry a qualifier. Numeric values include units of measure. Values may be specified in several manners:

• byasinglenumericvaluewithoptionaluncertain-ty,e.g.,5.24+/-0.12

• byanumericrange,e.g.,5.7-13.6• by a term with an identifier for the source vocabu-

lary,e.g.,“thick-bedded(NADMSLTTs)”• asarangewithboundsassignedbytermsorby

a term and a numeric value, e.g. “fine- to me-dium-grained(Folk1968)”or“Miocene(�UGS2004)”–1.7Ma.

Instance Document Example

Exampleinstancedocumentsassociatedwitheachversionoftheschemainthesubversionreposi-tory(https://www.seegrid.csiro.au/subversion/xmml/GeoSciML/tags/)arestoredinan“�nstances”subdirec-toryinthedirectoryforthatversion.Thefollowinglistingprovidesanexampleusageofmanyoftheelementsforgeologicunitdescription.Thebaseelementinthedocu-mentisaGeoSciMLcollection(gsml:);eachmemberofthecollectionstartswitha<member>element.GeoSciMLcollectionmembersmaybe:

1.Geologicfeatures(akindofGMLfeature)2.GMLgeometryelements3.MappedFeatures(outcrops,samplelocations,

traverses/sections) 4. Controlled concepts (vocabulary definitions)5.Geologicrelationships6.Dictionaries(collectionsofcontrolledconcepts)

Commentsinthefollowinglistingaredelimitedby‘<!--’and‘-->’.

<?xmlversion=”1.0”encoding=”UTF-8”?><Gsmlxmlns=”http://www.cgi-iugs.org/xml/GeoSciML/1”...othernamespacedeclara-tions> <!--Thelexiconwouldprobablybeinaseparatefile.TheStratigraphicLexi-conelementextendsGMLdictionary(throughGeologicVocabularyGeoSciMLelement)--><member> <StratigraphicLexicongml:id=”AZGSGeologicUnits”> <!--Thisisalexiconelementthatincludesthreeunits--> <gml:description>CollectionofgeologicunitsdefinedbyStateofArizona</

gml:description> <gml:name>Arizonastratigraphicunitlexicon</gml:name> <member> <ControlledConceptgml:id=”MartinFormationConcept”> <gml:description>lithostratigraphicformationdefinedby...</gml:

description> <gml:name>urn:x-cgi:def:lithostratigraphy:USGS:2006:Geolex:Martin-Formation</gml:name> <preferredName>MartinFormation</preferredName> <prototypexlink:href=”#Feature2524”/>

54 D�G�TALMAPP�NGTECHN�QUES‘06

<vocabularyxlink:href=”#AZGSGeologicUnits”/> <metadata/> </ControlledConcept> </member> <member> <ControlledConceptgml:id=”LS2”>...</ControlledConcept> </member> <member> <ControlledConceptgml:id=”LS3”>...</ControlledConcept> </member> </StratigraphicLexicon></member>

<member> <GeologicFeatureRelationgml:id=”rel-100”> <!--Thisisageologicrelationshipelement--> <gml:name>urn:x-cgi:def:lithostratigraphy:USGS:2006:featureRelation:

Stratigraphicposition</gml:name> <rolecodeSpace=”http://www.iugs-cgi.org/featureRelationVocabulary”>overli

es</role> <sourcexlink:href=”#BeckersButteMemberPrototype”/> <targetxlink:href=”#JeromeMemberPrototype”/> </GeologicFeatureRelation></member>

<member><!--GeologicFeatureisderivedfromGMLAbstractFeature,itassociatesade-scription,aclassifier(whatisdescribed)andanextent(whereitwasde-scribed,ifdefined).TheClassifierelementdefinesthetypeofafeature.Mul-tipledescriptionsmaybeassociatedwithaGeologicFeature--><GeologicFeaturegml:id=”Feature2524”><!--ThisisageologicunitGMLfeature,

whichisthebasiccontainerforgeologicunitdescriptionsinGeoSciMLv.1-->

<gml:description>ThetypesectionoftheMartinFormationatMt.MartinnearBisbeeconsistsalmostentirelyofmedium-graytomediumdark-grayaphantiictofine-grainelimestone.dolostoneisentirelysubordi-nate,...

</gml:description> <gml:name>urn:x-cgi:def:lithostratigraphy:USGS:2006:Geolex:TypeMartinForma-tion</gml:name> <gml:boundedBy> <gml:Envelope> <gml:lowerCorner/><!--cornersofaboundingboxfortypeareaofthe

MartinFormation;geometryspecificationelementsnotincludedhere-->

<gml:upperCorner/> </gml:Envelope> </gml:boundedBy> <purpose>typicalNorm</purpose> <age><!--Geologicageelementincludesadatevaluespecification(seebelow),andaneventspecificationthatexplicitlyidentifiestheeventtowhichtheageisas-signed(e.g.deposition,coolingthroughbiotiteclosuretemperature...)--> <GeologicAge> <value> <CGI_TermValue> <valuecodeSpace=”http://www.iugs-cgi.org/geologicAgeVocabulary”>Middle

55GEOSC�ML–AGMLAPPL�CAT�ONFORGEOSC�ENCE�NFORMAT�ON�NTERCHANGE

Devonian</value> </CGI_TermValue> </value> <event> <CGI_TermValue> <valuecodeSpace=”http://www.iugs-cgi.org/EventVocabulary”>deposition</

value> </CGI_TermValue> </event> </GeologicAge> </age> <classifierxlink:href=”#MartinFormationConcept”/><!--here’sthelinktothe

controlledconceptthatdefinestheintentionoftheMartinFormation.Linkisreferencetocontrolledconceptinstanceinthisdocument-->

<description> <LithostratigraphicUnitDescription> <metadata/><!--xlinktometadataforthisdescription;thisprovidestieto

Observationmodel-->

<partOf> <GeologicUnitDescriptionPart><!--310-340thinbedded,nonfossiliferous

dolostone--> <unit> <LithostratigraphicUnitDescriptiongml:id=”GeoUnitPart0235”><!--partisalsoalithostratigraphicunit,usessamedescriptionschemaascon-tainingunit;itcouldhavepartsitself;partonomyisrecursive.--> <descriptionSourcexlink:href=”referencetodescriptionsourceobserva-

tion”/><!—Sourceobservationelementnotincludedhere--> <bodyMorphologyxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <outcropCharacterxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <grossGenesisTermxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <exposureColorxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <grossChemistry> <CGI_TermValue> <qualifier>always</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/grossChemistryList”>carb

onate</value> </CGI_TermValue> </grossChemistry> <rankcodeSpace=”http://www.iugs-cgi.org/Vocabulary”>DescriptionPart</

rank> <weatheringCharacterxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <metamorphicGrade/> <!--notspecifiedsoimpliessameascontaining

unit--> <unitThickness> <CGI_NumericValue> <principalValueuom=”meter”>30</principalValue> <plusDeltauom=”meter”>20</plusDelta> <minusDeltauom=”meter”>10</minusDelta> </CGI_NumericValue> </unitThickness> <beddingStylexlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <beddingPatternxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <beddingThickness> <CGI_TermValue> <valuecodeSpace=”http://www.iugs-cgi.org/ThicknessVocabulary”>Thin-

56 D�G�TALMAPP�NGTECHN�QUES‘06

bedded</value> </CGI_TermValue> </beddingThickness> </LithostratigraphicUnitDescription> </unit> <rolecodeSpace=”http://www.iugs-cgi.org/unitPartRoleVocabulary”>Stratigrap

hicpart</role> <type>codeSpace=”http://www.iugs-cgi.org/unitPartTypeVocabulary”>Descriptiv

ePart</role> <proportion> <CGI_NumericValue> <qualifier>approximate</qualifier> <principalValueuom=”percent”>12</principalValue> <plusDeltauom=”percent”>0</plusDelta> <minusDeltauom=”percent”>0</minusDelta> </CGI_NumericValue> </proportion></GeologicUnitDescriptionPart> </partOf><!--endofpartdescriptions.Followingpropertiesapplytoentiredescribedunit-->

<descriptionSourcexlink:href=”referencetodescriptionsourceobservation”/> <bodyMorphologyxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <outcropCharacter> <CGI_TermValue> <qualifier>common</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/outcropCharacterVocabulary”>le

dgeforming</value> </CGI_TermValue> </outcropCharacter> <grossGenesisTerm <CGI_TermValue> <qualifier>always</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/GenesisVocabulary”>Sedimentary

,marine</value> </CGI_TermValue> </grossGenesisTerm> <exposureColor> <CGI_TermValue> <qualifier>common</qualifier> <valuecodeSpace=”http://www.color.org/ColorVocabulary”>Lightgray</val-

ue> </CGI_TermValue> <CGI_TermValue> <qualifier>common</qualifier> <valuecodeSpace=”http://www.color.org/ColorVocabulary”>Mediumgray</

value> </CGI_TermValue> <CGI_TermValue> <qualifier>rare</qualifier> <valuecodeSpace=”http://www.color.org/ColorVocabulary”>Pink</value> </CGI_TermValue> <exposureColor/> <grossChemistry> <CGI_TermValue>

57GEOSC�ML–AGMLAPPL�CAT�ONFORGEOSC�ENCE�NFORMAT�ON�NTERCHANGE

<qualifier>common</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/grossChemistryList”>carbonate</

value> </CGI_TermValue> <CGI_TermValue> <qualifier>occasional</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/grossChemistryList”>siliceous</

value> </CGI_TermValue> </grossChemistry> <rankcodeSpace=”http://www.iugs-cgi.org/Vocabulary”>Formation</rank> <weatheringCharacterxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <metamorphicGrade> <CGI_TermValue> <qualifier>always</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/metamorphicGradeVocabulary”>not

metamorphosed</value> </CGI_TermValue> </metamorphicGrade> <unitThickness> <CGI_NumericValue> <principalValueuom=”meter”>340</principalValue> <plusDeltauom=”meter”>10</plusDelta> <minusDeltauom=”meter”>10</minusDelta> </CGI_NumericValue> </unitThickness> <beddingStyle> <CGI_TermValue> <qualifier>common</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/Vocabulary”>Planarbedding</

value> </CGI_TermValue> </beddingStyle> <beddingPatternxlink:href=”urn:x-ogc:def:nil:OGC:unknown”/> <beddingThickness> <CGI_TermRange> <lower> <CGI_TermValue> <qualifier>common</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/Vocabulary”>thinbedded</

value> </CGI_TermValue> </lower> <upper> <CGI_TermValue> <qualifier>rare</qualifier> <valuecodeSpace=”http://www.iugs-cgi.org/Vocabulary”>thickbedded</

value> </CGI_TermValue> </upper> </CGI_TermRange> </beddingThickness> </LithostratigraphicUnitDescription> </description></GeologicFeature></member>

</Gsml>

58 D�G�TALMAPP�NGTECHN�QUES‘06

TEST BED DEMONSTRATION

Sixnationalandtwostategeologicalsurveyagencies,inAustralia,Europe,andNorthAmerica,participatedinaproof-of-conceptdemonstrationofGeoSciMLatthe�nter-nationalAssociationofMathematicalGeologists(�AMG)meetinginLiege,Belgium,inSeptember2006.Thedem-onstrationshowedthatitispossibletoaccessinformationinrealtimefromgloballydistributeddatasources.Geo-logicalmappolygonsandattributeinformation,andbore-holedata,weredisplayed,queried,andre-portrayedusingwebapplicationshostedbytheGeologicalSurveyofCanadaandtheFrenchBureaudeRechercheGéologiquesetMinières(BRGM).Functionsdemonstratedincludedcontinuous map portrayal with attribute query, reclassifi-cationaccordingtoattributes,anddownloadofcomplexdatastructuresencodedinGeoSciML.

�nformationdeliveryfromdifferentcomplexdatastoresusingacommunitystandardschemademonstratedthatGeoSciMLprovidesadatamodelandformatcapableofsupportingtransferofgeologydatafrommultiplejurisdictions.Thisalsodemonstratedthatadistributeddatadeliverysystemcanbeconstructedbyspecifyingstandardinterfaces,notlimitedtosinglevendorsoftware.Newservicescanbeaddedeasily,providingtheycon-formtotheinterface.Alloftheservicesinthetestbeduseddifferentdatastores,wrappedbyavarietyofserversoftware applications. Deployment requires configuration ofserver-andclient-sidesoftwaretoconformtothedatamodel,butdoesnotrequiredevelopmentofnewsoftware“fromscratch.”

Threeusecasesweredemonstratedatthe�AMG2006meetinginBelgium.UseCase1demonstrateddis-playofmapdataandqueryforthedescriptionofasingle

mapobject.Whentheclientasksforthemap,theserverreturnsamapwithdefaultsymbolization.Ausercanthenclickonanygraphicfeaturefromalayertoretrieveinformationforthefeature,whichcanbepresentedtotheuserasrawGeoSciMLorasamoreclearly-renderedHTMLversion.PresentationformatsotherthanHTMLcanberequestedbytheclientiftheserversupportsthem.Thetypesoffeaturesusedmustincludeatleastoneofthefollowing:geologicunits,faults,contactsorboreholes.

Usecase2demonstratedselectionanddownloadof features; a geographic bounding box is specified and thecontentsdownloadedasaGeoSciMLdocument.TheGeoSciMLdocumentcanbereformatted(e.g.byXSLTfordisplayinabrowser)orserveasinputforanotherprocess in a workflow. The GeoSciML document contains acollectionofGeologicFeaturesorBoreholes.

Usecase3demonstrateddynamicqueryandre-sym-bolizationofmappedfeaturesonthebasisofage,usingthe�UGSstandardgeologicagecolorscheme,oronthebasisof lithology, using a CGI defined lithology color scheme. TheresultsofsymbolizationbylithologyfordatafromCanada,theU.S.,andScandinaviancountriesisshowninFigure 4. A very simple lithologic classification and sym-bolizationwasused,withfourclassesandrelatedcolors:igneous(pink),sedimentary(green),metamorphic(purple),andunconsolidated(yellow).Eachparticipanthadtoimple-mentamechanismtomapfrompropertiesassociatedwiththemappedfeaturestothestandardizedlithologyclasses.�tistheserviceprovider’sprerogativetodeterminethemap-ping from the data source to the classification.

SUMMARY

Astandardizedschemaandsyntaxforinformationen-codingisafundamentalrequirementforinteroperableinfor-

Figure 4.UseCase3fromTestbed2,re-symbolizationofgeologicunitsbylithologyforCanada,U.S.andScandina-viancountries:igneous(mediumgray),sedimentary(lightgrey),metamorphic(darkgray),andunconsolidated(nearlywhite).

59GEOSC�ML–AGMLAPPL�CAT�ONFORGEOSC�ENCE�NFORMAT�ON�NTERCHANGE

mationsystems.The�UGSCG�DataModelcollaborationworkinggrouphasdevelopedGeoSciML,anXML-basedGML(geographymarkuplanguage)application,tomeetthisrequirementfortheinterchangeofgeoscienceinforma-tion.Theschemaforthisapplicationreusesexistingmarkuplanguages where possible. Newly developed markup specifi-cationsarebasedonexistingconceptualmodelsinmostcas-es.Thisstandards-baseddataformatprovidesaframeworkforapplication-neutralencodingofgeosciencethematicdataandrelatedspatialdata.�tisintendedforuseinpublish-ingorinterchangingdatabetweenorganizationsthatusedifferentdatabaseimplementationsandsoftware/systemsenvironments.Fullrealizationofdatainteroperabilityatthesemanticlevelwillrequiredevelopmentofcontrolledvo-cabularyresourcesforspecifyingactualcontent.ATestbeddemonstratedsimpleinteroperabilityusingwebmapandfeatureservices(WMS,WFS)betweengeologicalsurveysinseveraldifferentcountries.GeoSciMLisbeingconsid-eredasanationalstandardforgeosciencedataexchangebyfederalandstategeologicalsurveysinAustraliaandtheEuropeanUnionSpatialData�nfrastructure(�NSP�RE),andwill be submitted in 2007 as an IUGS-CGI specification.

DevelopmentofGeoSciMLisanopenprocesswiththeintenttoinvolveasmanyparticipantsaspossible.Thiswillensuredevelopmentofaschemaandservicesthatwillmeettheneedsofawidevarietyofgeosciencedataproducersandusers.Threetypesofparticipationareavailable:1)directparticipationinGeoSciMLdevelop-ment,2)monitoringGeoSciMLdevelopmentviatheweb-collaborationtoolsand3)deployinganinternetservertoprovidedatainGeoSciMLformat.

REFERENCES

Boisvert,E.,Johnson,B.R.,Cox,S.J.,andBrodaric,B.M.,2004,GMLEncodingofNADMC1,inSoller,D.R.,ed.,Digital

MappingTechniques‘04—WorkshopProceedings:U.S.GeologicalSurveyOpen-FileReport2004–1451,p.95-103.Availableathttp://pubs.usgs.gov/of/2004/1451/bois-vert/index.html(accessed12/12/2006).

Brodaric,B.M.,andGahegan,Mark,2006,Representinggeosci-entific knowledge in cyberinfrastructure: some challenges, approachesandimplementations,inSinha,Krishna,ed.,Geoinformatics-Datatoknowledge:GeologicalSocietyofAmericaSpecialPaper397,p.1-20.

Cox,S.J.D.,ed.,2006,Observationsandmeasurements:OpenGeospatialConsortium,�nc.,documentOGC05-087r4,version0.14.7,168pages.Availableathttp://portal.opengeospatial.org/files/?artifact_id=17038(accessed12/13/2006).

Cox,S.J.D.,Daisey,P.,Lake,R.,Portele,Clemens,andWh-iteside,Arliss,eds.,2004,OpenG�SGeographyMarkupLanguage (GML) Implementation specification: Open GIS Consortium,�nc,documentOGC03-105r1,version3.1.0,580p.

Cox,S.J.D.,andRichard,S.M.,2005,Aformalmodelforthegeologictimescaleandglobalstratotypesectionandpoint,compatiblewithgeospatialinformationtransferstandards:Geosphere,v.1,p.119-137,DO�:10.1130/GES00022.1.Availableathttp://geosphere.geoscienceworld.org/cgi/con-tent/abstract/1/3/119(accessed12/12/2006).

NorthAmericanGeologic-MapDataModelSteeringCommit-tee,2004,NADMConceptualModel1.0,AConceptualModelForGeologicMap�nformation:U.S.GeologicalSurveyOpen-FileReport2004-1334,61p.,availableathttp://pubs.usgs.gov/of/2004/1334/.

Richard,S.M.,2006,Geoscienceconceptmodels,inSinha,Krishna,ed.,Geoinformatics-Datatoknowledge:Geologi-calSocietyofAmericaSpecialPaper397,p.81-108.

SC36Secretariat,2003,ProposedDraftTechnicalReportfor:�SO/�EC2382,�nformationtechnology—Learning,educa-tion, and training—Management and delivery—Specifica-tion and use of extensions and profiles: ISO/IEC 2382-01, �SO/�ECJTC1SC36N0646,availableathttp://jtc1sc36.org/doc/36N0646.pdf(accessed12/12/2206).