简介：Nonlinearmixedeffectsmodel(NLMEM)isbuiltontherelationshipofthefixedandrandomeffectsintheregressionfunction.TheNLMEMhasanobviouscomparativeadvantageinanalyzingthelongitudinaldata,repeatedmeasuresdataandmultileveldata.Two-levelNLMEMisusedtoanalyzethedominantheightforChinesefir(Cunninghamialanceolata).Theauthorsoutlinethetwo-levelNLMEMandintroducetheparametersestimationmethodofthemodel.BasedonfivecommonRichardandLogisticmodels,themixedmodelisbuilt.Themodelingdataareusedtocalculateandcomparewith19modelsderivedfromeachbasedmodel,and5optimalmixedmodelsarebuilt.Comparedthe5optimalmixedmodelswithtraditionalregressionmodels,itisshowedthatthetwo-levelNLMEMhasabetterfittingeffectthantheregressionmodel.

简介：Nonlinearmixedeffectsmodel（NLMEM）isbasedontherelationshipbetweenthefixedandrandomeffectsintheregressionfunction.TheNLMEMhasacompetitiveadvantageinanalyzingrepeatedmeasuresdata,thelongitudinaldataandmultileveldata.Thispaperchosetwokindsoftwo-levelnonlinearmixedmodeltoanalyzebasalareagrowthforChineseFir（Cunninghamialanceolata）.Model1isageneraltwo-levelNLMEMandModel2isbasedonModel1tofurtherconsiderthefixedeffectsparameterschangeswithaspecificfactor.Firstly,throughtheanalysisofthesetwomodels,thispaperdefinedthebasicmodeltobuildthetwo-levelNLMEM.Secondly,665kindsofmodelsderivedfromModel1and2703kindsofmodelsderivedfromModel2werecalculatedandcompared.Theresultsshowedthat:forModel1,therewere57kindsofmodelsconverging,andwhentheformalparameterb0consideredtheblockeffectsandploteffects,b1andb4onlyconsideredtheblockeffects,themodelfittedthebest;andforModel2,therewere24kindsofmodelconverging,andwhentheformalparameterbsconsideredtheblockeffectsandploteffects,b1onlyconsideredblockeffectsandthefixedeffectsb0changedwithanylevelofblocklevel,Model2fittedthebest.Finally,bycomparingthetraditionalnonlinearregressionmodel,Model1andModel2,theresultsshowedthatModel1andModel2fittedbetterthanthetraditionalnonlinearregression,andModel2wasbestfittingmodel.