Page 118 - 2025年第56卷第7期
P. 118
[32] 祁诗阳,沈宸,刘小标,等.基于小波变换的郑州降雨量与太阳黑子活动关系分析[J].现代地质,2023,
37(1):184 - 196.
[33] TORRENCEC,COMPOGP.Apracticalguidetowaveletanalysis[J].BulletinoftheAmericanMeteorological
Society,1998,79(1):61 - 78.
[34] 王秀杰,封桂敏,耿庆柱.小波分析组合模型在日径流预测中的应用研究[J].自然资源学报,2014,29
(5):885 - 893.
[35] 李红燕,王志平,羊华,等.基于 Morlet小波分析的弥苴河径流量变化及其影响因素分析[J].人民珠江,
2021,42(3):53 - 60.
[36] 梁中耀,刘永,盛虎,等.滇池水质 时 间 序 列 变 化 趋 势 识 别 及 特 征 分 析 [J].环 境 科 学 学 报,2014,34
(3):754 - 762.
[37] CLEVELANDRB,CLEVELANDW S,TERPENNINGI.STL:aseasonal - trenddecompositionprocedurebased
onloess[J].JournalofOfficialStatistics,1990,6(1):3 - 73.
[38] DRAGOMIRETSKIYK,ZOSSOD.Variationalmodedecomposition[J].IEEETransactionsonSignalProcessing,
2014,62(3):531 - 544.
[39] 江星星,宋秋昱,杜贵 府,等.变 分 模 式 分 解 方 法 研 究 与 应 用 综 述 [J].仪 器 仪 表 学 报,2023,44(1):
55 - 73.
[40] 徐岩,向益锋,马天祥.基于粒子群算法优化参数的 VMD - GRU短期电力负荷预测模型[J].华北电力大
学学报(自然科学版),2023,50(1):38 - 47.
[41] PARKM J,HA R,KIM N W,etal.Assessmentoffutureclimateandvegetationcanopychangeimpactson
hydrologicalbehaviorofChungjudam watershedusingSWAT model [J].KSCE JournalofCivilEngineering,
2014,18(4):1185 - 1196.
[42] YOOC,KIM J,YOONJ.Uncertaintyofarealaveragerainfallanditseffectonrunoffsimulation:Acasestudy
fortheChungju - dam basin ,Korea[J].KSCEJournalofCivilEngineering,2012,16(6):1085 - 1092.
[43] KIM S,TACHIKAWAY,TAKARA K.Recentflooddisastersandprogressofdisastermanagementsystem in
Korea [J].AnnualsofDisasterPreventionResearchInstituteofKyotoUniversity,2007,50(B):15 - 31.
Atwo - stagedynamicdecomposition - prediction - reconstructionmodelfor
medium- longterm runoffforecasting
4
1
1
1,2,3
1
CHENXiaoze,WANGZhongjing ,LIUDan,SHIYujia,KANGBoosik
(1.DepartmentofHydraulicEngineering,TsinghuaUniversity,Beijing 100084,China;
2.StateKeyLaboratoryofHydro - scienceandEngineering,TsinghuaUniversity,Beijing 100084,China;
3.SchoolofCivilandHydraulicEngineering,NingxiaUniversity,Yinchuan 750021,China;
4.DepartmentofCivil&EnvironmentalEngineering,DankookUniversity,Yongin 16890,SouthKorea)
Abstract:Medium - longtermrunoffforecastingisacriticalfoundationforoptimizingwaterresourcemanagement.
Withtheintensificationofclimatechange ,thevariabilityofrunoffhasincreased,makingtheforecastingmore
challenging.Toimprovetheaccuracyofrunoffforecastsandextendtheforecastperiod ,inthispaper,byintegrat
edSeasonalandTrenddecompositionusingLoess (STL),VariationalModeDecomposition(VMD)andthedeep
learningmodelInformer ,atwo - stagedynamicdecomposition - prediction - reconstructionmodel(STL - VMD - In
former)formedium - longtermrunoffforecastwasdevelopment.TheapplicationattheShixialihydrologicalstation
ofYongdingRiverBasinshowsthattheforecastNash - Sutcliffeefficiency(NSE)reaches0.897,0.843,and0.
796forpredictionperiodsof1 ,3,and6months,respectively,indicatingtheproposedmethodwiththepotential
inseparatingthetimeseries ,extendingtheforecastinghorizon,andimprovingforecastaccuracy.Theapproach
willbenefittothemedium - longtermrunoffforecasting.
Keywords:runoffforecast;STL;VMD;Informer;YongdingRiver
(责任编辑:韩 昆)
— 9 4 —
4

