資源描述:
《基于統(tǒng)計模型的汛期降水預(yù)測分析——以義烏市為例》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、萬方數(shù)據(jù)Basedonthefloodseasonprecipitationofyiwuastheresearchobject,thispaperattemptstouseneuralnetworkpredictionmethodandthresholdautoregressivemodelrespectivelytostudythefloodseasonprecipitationofyiwu.Thispaperchosetheelninoeventsandthesubtropicalhighridgelinepositionasantecedentinfluencedfac
2、torswhichhavegoodresponseoffloodseasonprecipitationofyiwu,49yearsof1959-2007floodseasonprecipitationasthetrainingsample,and4yearsoffloodseasonprecipitationof2008—201astestsamples.Thetwomodelswereanalyzedbythreeaspects:thehistoricalsamplefittingprecision,theforecastresultsofindependentsample
3、andactualforecastability.Thefinalresultisusedtoverifythepossibilitytousetheneuralnetworkmodelandthethresholdautoregressivemodelforfloodseasonprecipitationforecastinyiwu,andalsoprovidesthebasisofthetheoryandpracticeoffloodseasonprecipitationforecastmodelinthecomingyears.Themainconclusionsoft
4、hisstudyaresummarizedasfollows:(1)Thispaperstudiesthecharacteristicsoffloodseasonprecipitationofyiwu,andalsothemainfactorsinfluencingprecipitation,provedthefloodseasonprecipitationofyiwuhasagoodresponsetosolaractivity,elninoevents,andpositionofsubtropicalhighridgeline.Onthebasisofmanyresear
5、cherstostudyconclusion,thispaperchoseelninoeventsandthesubtropicalhighridgelinepositionasimpactfactorstoestablishmodels.Thefinaldataresultsconfirmedthatitcanignificantlyreducetheprecipitationforecasterrorwhileselectingtheelninoeventsandpositionofsubtropicalhighridgelineastheearlystageofthei
6、mpactfactor.(2)Neuralnetworkforecastmodelandthethresholdautoregressivemodelareeffectiveprocessingnonlinearhydrologicsystemmodeloffloodseasonprecipitationforecastofyiwucity.BPneuralnetworkforecastingmodelisbetterwhenitcomestothehistoricalsamplefittingprecision,theforecastresultsofindependent
7、sampleandactualpredictionability.However,thedataofneuralnetworkforecastmodelandthresholdpredictionmodelareinsufficient,SOtheprecisionoftheirresultsneedslmprovmg.Keywords:NeuralNetworkForecastModel;ThresholdAutoregressiveModel;FloodSeasonPrecipitation;Imp