玉米虚拟模型GREENLABMAIZE的构建与评估

《玉米虚拟模型GREENLABMAIZE的构建与评估》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、中国农业大学博士学位论文玉米虚拟模型GREENLAB-Maize的构建与评估姓名：马韫韬申请学位级别：博士专业：土壤学指导教师：李保国;郭焱20061101AbstraetFunctional-structuralplantgrowthmodelsCallconsmlcttheplantarchitecturethroughsimulatingtheorganogenesis,biomassproductionandbiomasspartitioningtodifferentorgans．andthusaI℃importantforin-dep

2、thstudyoftheplantgrowthrulesandeultivaroptimization．A邓嘲施cmodelofGREENLAB-Maizew私dcwlopedbased011GREENLABtheory．Wesystemlatieallyevaluatedthestabilityofmodelparametervaluesandvalidityofthesimulationresultsusingpluri-annualandpluri-locationfieldtrialstakingbothfieldheterogenei

3、tyandenvironmentalconditionsvariability．AlibraryofsymbolfilesofmaizeorganswaseonsmaetedanddynamicchangesofmaizeplantlcaTeswel'e柚alyzedinordertoaccuratelysimulatingindividualmaizeplantandstandgrowth．Themajorcontentsmill5follows：(1)Multi-fittingtechniqueusingseveraltargetfiles

4、describingplantsarchitectureandbiomasspartitioningatdifferentgrowthstagesandsinglefittingtechniqueusingonlyo∞targetfiledescribingplantarchitectureandorganweightatfinalstageWCl'ecompared．Theresultsshowedthatmulti-fittingtechniquegavebetterparameteraccuracythansinglefitting，an

5、dpermittedextractinggeneric01．aganexpansionk／neliesfromthestaticobservations．Thus，usingmulti—ratherthansingle矗劬呜w鹳recommendedtoaee,ratelycha“K自妇hiddenparameters．(2)Adeepenedanalysisofoptimizedparametervariabilityamongyears。plantingdensitytrea缸rlents。repetitionsandgrowthstage

6、swascarriedoutinthecaseofonemaizegenotypeeroppcdinVaZiOUSenvironmental，nonlimitingconditions．Astatisticalanalysisshowedthat,despiteof13．hi曲variabilityinplantbiomassanditspartitioningamongorgansbd嘀，黜experiments(withlowvariabilityamongrepetitions)，parametervaluesoptimizedsepar

7、atelyforeachexperiment．repetitionandgrowthstageswerenotsi舯Jfieantlydifferent．Butvaluesofparameterrl，，2andPcvariedbetwⅫdifferentplantingdensitytreatments．whileothersⅥ，e”stable．(3)Multi-fittingwithsparsedataset(inel,ding3growthstagesofwhichthefirsttwo锄sparsedataandthelastOneis

8、completedata)wasusedtoevaluatetheeffectofsuchsimplificationsonparameteropti