Page 63 - 2022年第53卷第12期
P. 63
156 - 165.
[37] GUPTAH,SOROOSHIAN S,YAPO P.Statusofautomaticcalibrationforhydrologicmodels:Comparisonwith
multilevelexpertcalibration [J].JournalofHydrologicEngineering,1999,4(2):135 - 143.
[38] ZHANGY,GAOX,SUNB,etal.Trackingthermalstructureevolution:Anobjectivepracticeinastratifiedres
ervoirbasedonhigh - frequencymeasurements [J].JournalofHydrology:RegionalStudies,2022,39:100989.
[39] 刘畅,刘晓波,周怀东,等.大流量调度过程对水库缺氧区抑制阈值条件研究[J].水利学报,2021,52
(10):1217 - 1228.
[40] 崔凯鹏,吴吉春.观测数据时空密度 对 集 合 卡 尔 曼 滤 波 计 算 精 度 的 影 响 [J].水 利 学 报,2013,44(8):
915 - 923.
[41] HOUTEKAMERLP,MITCHELLLH.A sequentialEnsembleKalmanFilterforatmosphericdataassimilation
[J].MonthlyWeatherReview,2001,129(1):123 - 137.
[42] EVENSENG.TheEnsembleKalmanFilter:theoreticalformulationandpracticalimplementation[J].OceanDy
namics ,2003,53(4):343 - 367.
[43] 师春香,谢正辉,钱辉,等.基于卫星遥感资料的中国区域土壤湿度 EnKF数据同化[J].中国科学(地球
科学),2011,41(3):375 - 385.
[44] GHAHRAMANIZ.Probabilisticmachinelearningandartificialintelligence[J].Nature,2015,521(7553):
452 - 459.
[45] READJS,JIAX,WILLARDJ,etal.Process - guideddeeplearningpredictionsoflakewatertemperature[J].
WaterResourcesResearch ,2019,55(11):9173 - 9190.
Medium andshortterm predictionofwatertemperatureindeep
waterlake - reservoirbasedondataassimilation
3
1
2
3
1
SUNBowen,YANGXiyu,BAOZhu,LIUXiaobo,LIUChang,GAOXueping 1
(1.StateKeyLaboratoryofHydraulicEngineeringSimulationandSafety,TianjinUniversity,Tianjin 300350,China;
2.LuanheRiverWaterQualityMonitoringCenterofHaiheRiverBasin,LuanheRiverDiversionProjectManagementBureauof
HaiheRiverWaterConservancyCommission ,Tangshan 064309,China;
3.InstituteofWaterEcologyandEnvironment,ChinaInstituteofWaterResourcesandHydropowerResearch,Beijing 100038,China)
Abstract:Thedistributionandevolutionofwatertemperatureindeep - waterlakesandreservoirsaffectwater
movement , biochemicalreactions, and the metabolic processes of aquatic organisms.Forecasting water
temperaturechangesisnecessaryforlakeandreservoirwaterqualitymanagementandecologicalenvironmentsafety.
Inthisstudyalakeandreservoirwatertemperaturedataassimilationsystem isconstructedonthebasisoftheen
sembleKalmanfilteralgorithmandtheCE - QUAL - W2model ,whichisabletocomprehensivelyconsidermodel
parameters ,boundaryconditions,andtheuncertaintyofobservationdata.ThesystemisappliedtoDaheitingRes
ervoirfor1 - 10daysofmediumandshorttermwatertemperatureforecastingusingreservoiroperationdataandme
teorologicaldataasforecastconditions.Theresultsshowthatwhenthenumberofsetsis100 ,thesimulationerror
andtheobservationerrorare10% and1%,theassimilationsystemcanconsiderhighercalculationefficiencyand
simulationaccuracy.Simultaneouslycorrectingthemodelparametersandstatevariablescanincreasethewatertem
peraturesimulationaccuracyofthedataassimilationsystematdifferentwaterdepthsby41.2%- 68.8% comparedto
theresultswithoutdataassimilation.Astheforecastperiodwasextendedfrom 1to10days ,theforecasterrorof
eachwaterdepthincreasedfrom0.22 - 0.35℃ to0.77 - 1.09℃.Regardlessofwhetherthereservoirisinthestrati
fiedormixedperiod,thedataassimilationsystemcanmaintainhighaccuracyundertheinternalandexternalfac
torssuchasmeteorologicalconditionsandreservoirschedulingduringtheforecastperiod.Thehigh - precisionmedi
umandshorttermwatertemperatureforecastmethodcanprovidetheoreticalandtechnicalsupportforlakeandres
ervoirwatersupplyandecologicalsecurity.
Keywords:lake - reservoirwatertemperature;mediumandshorttermforecasting;dataassimilation;CE - QUAL - W2;
ensembleKalmanfilter
(责任编辑:韩 昆)
4
— 1 5 5 —