Bayesian多層模型評(píng)價(jià)杭州西湖富營(yíng)養(yǎng)化風(fēng)險(xiǎn).pdf

1、浙江大學(xué)碩士學(xué)位論文Bayesian多層模型評(píng)價(jià)杭州西湖富營(yíng)養(yǎng)化風(fēng)險(xiǎn)姓名：洪嵐申請(qǐng)學(xué)位級(jí)別：碩士專(zhuān)業(yè)：環(huán)境科學(xué)指導(dǎo)教師：裴洪平200605011V浙江夫掌磺士攀位論震糕篇鼉暑詈皇鼉篁煳舞魯烹!曼皇鼉?nèi)既?!!!暑曼黑姻堂詈毫皇皇曼s搠鼎!皇曼鼉寰燃_攔自!喜!!!皇!!!!!!!徽AbstractModelsareoftenf|ttodatasetscomposedofobservationsfrommultiple“samplingu

2、nits”，orcalled“subsystems”comparedtothewholesystemThesesubsystemsmaysharesome‘commonalitybuteachofthemisuniqueinbehaviourmoreorless，SOthatassumptionofthesamesetofmodelparametervaluesacrossallsubsystemsmaynotalwaysbevalid

3、ThisproblemcallbeovercomebyadoptingahierarchicalapproachUnderthehierarchicalstructure，eachsubsystemhasitsownsetofparametervalues，butsomecommonalireinvaluesiSassumedacrossthewholesystem，whichcanbestructuredbyaIlunderlying

4、populationdistributionWeappliedthishierarchicalmethodtoassessingeutrophicationriskinWestLake，Hangzhou，Chinaatypicalshalloweutrophicationlake，artificiallydividedinto5subsystemsTheprocedurefollowedwasdevelopingahierarchica

5、lmodelrelatingeutrophicationresponse—thelevelofChlorophylla(Chla)一tomultipleroutinelymonitoringvariables，andthenintroducinganewvariable，named“theprobabilityofstandardviolation”，indicatingtheexpectedfrequencyofChlastandar

6、dexceedanceForcomparison，bothglobalandsubsystemspecificparameterswereestimatedusingBayesTheoremResultsshowedthatthehierarchicalmodelwasmorerealisticthantheglobalmodelFurthermore，inBayesianperspective，predictionsexpressed