矿产资源定量预测思路与方法流程(成秋明PPT)

••““””•••••Agterberg769492199620081988“GIS”200GIS89Agterberg&Bonham-Carter1994GraemeBonham-Carter“GIS”GIS1990-1997SPANS,ArcView,ArcGISGIS982005NRR“”2007JCUG2008SPRINGER“”IAMG40Bonham-Carter&ChengMGNRR“”“”DonSinger,GaryRainse,LarryDraw,FabbriandChung-JoChungetc.,“”GIS,“”:ABC334-1334-2334-3ABC334-1334-2334-311234562122.2.22.2.31123425002.5P(D)=0.00258=20/7779.9P(D|A)=15.5/3065=0.0050P(D|notA)=4.5/4714=0.000963)(log/)|(log)()(log/)|(log)(APDAPAAPDAPA==αα(Cheng2007HtN-λ2006CuPbZnSnSnAuAsSbS-ASbAuAsWSnCuPbZnSn+Au+As+W-Cu–Pb-ZnS-A312(A)(B)(D)(C)45612GIS“1(A)(C)(K)(Ca)”“AandCandKnotCa”7123SpatialDecisionSupportSystem(SDSS)SpatialDecisionSupportSystem(SDSS)GISDataIntegrationforPredictionGISDataIntegrationforPredictionGISDataSourcesRemoteSensingGeologicalGeochemicalGeophysical..PotentialEvidentialLayers(X)Modeling(F)OutputData(S)ProcessingDataPreprocessingInterpretingInformationExtractionIntegrationDBMSDBMSDBMSDBMSGeographicalAA+AW−AW+++++=AA)|(−−+++=AA)|(--5.0)|(=DABPBB+BW−BW+++++=BB)|(−−+++=BB)|(5.0)|(=DABPBABAABBA+++++++++=BABA)|(+−+−+++++=BABA)|(−−−−+++++=BABA)|(−+−++++++=BABA)|(LogisticModelforSDSSLogisticModelforSDSS...)|(ABDPPosteriorProbabilityPosteriorProbability......}|{0+++=BA=F(x1,x2,…,xn)SS––FavorabilityFavorabilityProbabilityProbabilityFuzziness)Fuzziness)SuitabilitySuitabilityxxii--mapsorevidencesmapsorevidenceswwii--weightsweightsSimpleLinearModelSimpleLinearModelnnxwxwxwS+++=...2211ModelConstraintsModelConstraintsNormalization:Normalization:1.1.ConvertmapsintocomparableunitConvertmapsintocomparableunit⎩⎨⎧=noyesxi,0,12.2.WeightsshowingrelativeWeightsshowingrelativeimportanceofmapsimportanceofmaps101...21≤≤=+++in⎪⎩⎪⎨⎧=nounknownyesxi,0,5.0,110...,,2,1=ixMethodsforCalculatingWeightsforDataMethodsforCalculatingWeightsforDataIntegrationIntegrationDataDrivenMethods:DataDrivenMethods:WeightsofevidenceWeightsofevidenceLogisticregressionLogisticregressionArtificialNeuralnetworkArtificialNeuralnetworkKnowledgedrivenMethods:KnowledgedrivenMethods:FuzzylogicFuzzylogicHybridMethods:HybridMethods:FuzzyweightsofevidenceFuzzyweightsofevidenceModelTypesModelTypes1.Probabilistic1.Probabilistic2.Deterministic2.DeterministicSS––randomvariableshowingrandomvariableshowingprobability0probability0≤≤SS≤≤1withuncertainty1withuncertaintySS––Score0Score0≤≤SS≤≤11RelationshipsBetweenDifferentModelsSimpleOverlayModel(Union,Intersect,Identity)LinearModel(addingweights)LogisticModel(WeightsofEvidence,LogisticRegression)FuzzyLogicmodel(variousoperators)SpatialDataModelerExtension:Arc-SDMandGeoDASWeightsofEvidenceFuzzyWeightsofEvidenceLogisticRegressionFuzzyLogicNeuralNetwork1234567