您好,欢迎访问三七文档
AssistingAustralianindigenousresourcemanagementandsustainableutilizationofspeciesthroughtheuseofGISandenvironmentalmodelingtechniques.测绘0911第八组2012.5.3Firstofall,letusgraspagoodoverviewofAustraliaIntroductionBackgroundandstudysiteMethodsResultsconclusionLANDOFAUSTRLIATHESPECIESNORTHAUSTRALIAPROBLEMINTRODUCTIONTHEMETHODIntroductionThecontinentofAustraliawithaCoastline----30000kilometers,alandarea----7682300km,itisthelargestislandintheworld.ThecountryalsoincludesTasmania,theTorresStraitsIslandsandasmallnumberofislandsinthePacificandtheIndianOceans.So,Therefore,Australia'sgeographicalsituationisverycomplexAcomplexcontinentWesternAustraliaNorthernTerritorySouthAustraliaNewSouthWalesQueenslandVictoriaTasmaniaAustraliaCapitalTheindigenousresoucesareusedbythelocalpeopleforalongtime,but,howcantheymanagethemproberlyandkeepitsustainable?THESPECIESOFAUSSomeofthemarethreatenedwithextinctionnow!kangaroocamelcrocodilesharkkoalaUnusualanimalsinAustraliaOTHERRESOUCESLivestockForestryHOUDOESLOCALSUESTHEMINTHEPASTHuntingLoggingMiningTheproblemToday,alotofresourcescannotberationalexploitation.Alsotherearemanyspeciesthreatenedwithextinction,ahugewasteofresourcesisexisted!Meanwhile,tomanagethemissuchadifficultthingforthelocalpeople!THEBAKCGROUNDANDSTUDYSITEThereisanexampleofakindofspecies:CycadsStudysite:ThestudyareaistheManingridaregionofcentralArnhemLand,500KMlandscapeisdominatedbytheArnhenmLandsandstoneescarpmenttothesouthEnviomentalmodelandGISapplicationMethodThetypeofsurveydatathatisavailableplaysamajorroleindeterminingthetypeofmodelingmethodthatcanbeapplied.Presence-absencemodelshavebeennotedasbeingsuperiortopresence-onlymodels.ColletdateDataClassificationModelingAustralia'sgeographicalsituationInthissituation,thereshouldbeaenviromentalmodelEnviomentalmodelThecomplexityoftheenvironmentofthesurveyistheneedforinnovativetools.InnovationistheIDRISIsoftwaredevelopmentgoals.TheIDRISIinnovationincludingtheestablishmentofadvancedmodelingtools.TheapplicationoftheintegratedEarthtrendmodelingtools(ETM)providesobservation,detectionandtimeseriesanalysistoolstodiscoverthetrendofenvironmentalchange,especiallythestudyofglobalclimatechange.Integratedlandchangemodule(LCM)toleadtheoperatorthroughthecomplicatedstepstocreatechangemodels.Usesadvancedalgorithmsandtools,andpastlandcoverchangetopredictfuturechangesoftheland.ThisfeatureasREDD(Reducingemissionsfromdeforestationandforestdegradation)project.LCMprojectalsoincludestheevaluationoftheimpactoflandusechangeonbiodiversity.Modelsystemforotherapplicationsbythemacromodeling.FieldObvervationTotryandminimizetheinfluenceoftheroad/huntingtracksthequadratswerepositioned50mntothebushinarandomdirectionwithoutregardforthepresenceofthetargetsecies.ThecoordinatesofthemiddleofeachquadratwastakenusingGarminGPS12XL.Enviomental/predictordatasetsDatawereselectedfromavailablemappedenvironmentalthemestoreflectthefactoesthoughtmostlikelytoinfluencespeciesdistribution,basedonbothaknowledgeofthespeciesecologyfromconsultationwithexperts,andindigenousknowledgeofthespecies.Amixtureofrasterandbroadscalevectordatawasused.ASdataisonlycapturedatonerecordingstationinthevicinity.Afaindataforthisareaisnotavailableatascalewheredifferencesinelevation(anditsderivationofslopeandaspect)wouldhavemuchinfluenceontheoutput.sothisdatawasnotincludedinthemodel.ValidationValidationofthepredictivemapsinvolvedindependentadditionalgroundsampling.Thisinvolvedsurveying271and235quadrats(50*50m)withineachoftheC.arnhemandB.diversifoliusstudysites.respectivelyandrecordedpresenceorabsenceofthespecies.WeusedArcView3.3toaplace1200mbufferaroundtheroadandhuntingtracksandgeneraterandompointsIneachofthethreeprobabilityclasses.Ateachstudysiteanumberofrandompointswerechosenforeachofthehigh,mediumorlowprobabilitycategories(intotil22forC.arnhemand16B.diversifolius)andthesewereusedasstartpointsfromwhichafurther1250*50mquardratsweresampledevery100m,the100mmarkbeingthecenterofthesmallerquardrat.Thedirectionofthequadratsfromthesestaetingpointswasrandomlychosen(fromdegreesonpaperandchoseninahat)withpossibledirectionsbeingbetween0and180fromtheroadtoensuretransectsdidnotcrossovertheroad..Subsequntly,theindividual50*50mquadratsfellinthesamedirectionasthelargerquadrat.Thismethodwasusedtosurveyawayfromtheroad/trackwhilestillminimizingthetimespentcollectingthevalidationdata.TestingthemodelCohen’skappastatisticsweretotestoverallmodelperformationandprovideasimples,effective,standardizedandappropriatestatisticforevaluatingpresence-abencesmodels.adescripitionofthismethodologyisavailableinfieldingandbell(1997)sowithnotbedescribledinthispaper.IthasbeensuggestedthatwhencalculatingKappastatistic,insteadofusinganarbitrarythresholdof50%todistinguishsimulatedpresencefromsimulatedabsence,aprobabilitythresgoldthatmaximizesthemodel’sperformanceshouldbedeterminedbyevaluatingkatsuccessiveprobabilityincrements.InusingKappastatisticstotesttheaccuracyofthemodelsgeneratedinthisstudywehaveadoptedasimilarapproach.ResultMapshowingpredictedprobabilityofoccurre
本文标题:GISAUS1
链接地址:https://www.777doc.com/doc-1085021 .html