Page 27 - 2024年第55卷第6期
P. 27
reservoirindicesasexternalcovariates[J].HydrologyandEarthSystem Sciences,2013,17(8):3189 - 3203.
[15] 齐民友.概率论与数理统计[M].2版.北京:高等教育出版社,2011.
[16] 水利部长江水利委员会.长江三峡水利枢纽初步设计报告(枢纽工程)[R].1992.
[17] XIONGB,XIONGL,GUO S, etal.Nonstationaryfrequencyanalysisofcensoreddata: acasestudyofthe
floodsintheYangtzeRiverfrom1470to2017[J].WaterResourcesResearch,2020,56(8):e2020WR027112.
[18] JIANGC,XIONGL,XUCY,etal.Arivernetwork - basedhierarchicalmodelforderivingfloodfrequencydistribu
tionsanditsapplicationtotheUpperYangtzebasin [J].WaterResourcesResearch,2021,57(8):e2020WR029374.
[19] XIONGB,XIONG L, XIA J, etal.Assessingtheimpactsofreservoirsondownstream floodfrequencyby
couplingtheeffectofscheduling - relatedmultivariaterainfallwithanindicatorofreservoireffects [J].Hydrology
andEarthSystem Sciences ,2019,23(11):4453 - 4470.
[20] BATALLARJ,G?MEZCM,KONDOLFGM.Reservoir - inducedhydrologicalchangesintheEbroRiverbasin
(NESpain)[J].JournalofHydrology,2004,290(1):117 - 136.
[21] 李国芳,郝振纯.变化环境下流域水文气象极端事件演变规律及超标准洪水致灾机理[M].武汉:长江出
版社,2022.
[22] STANFORDJL,VARDEMANSB.StatisticalMethodsforPhysicalScience[M].Cambridge:AcademicPress,1994.
[23] 熊丰,郭生练,陈柯兵,等.金沙江下游梯级水库运行期设计洪水及汛控水 位 [J].水 科 学 进 展,2019,
30(3):401 - 410.
[24] AKAIKEH.Anewlookatthestatisticalmodelidentification[J].IEEETransactionsonAutomaticControl,1974,
19(6):716 - 723.
[25] 三峡工程泥沙专家组.三峡工程运行后泥沙冲淤与调控[M].北京:中国水利水电出版社,2020.
[26] 徐长江,徐高洪,戴明龙,等.三峡 水 库 蓄 水 期 洞 庭 湖 区 水 文 情 势 变 化 研 究 [J].人 民 长 江,2019,50
(2):6 - 12.
Non - stationarydesignfloodestimationofThreeGorgesReservoirin
operationperiodconsideringhistoricalinformation
XIEYuzuo,GUOShenglian,XIONGLihua,WANGJun,
ZHONGSirui,YANGYuanting
(StateKeyLaboratoryofWaterResourcesEngineeringandManagement,WuhanUniversity,Wuhan 430072,China)
Abstract:At - sitenaturalannualmaximumfloodseriesandthePearsonType Ⅲ(P - Ⅲ)distributioncurvefitting
methodareusedtoestimatefloodquantilesanddeterminethereservoircharacteristicwaterlevelsintheplanning
anddesignstageofwaterconservancyandhydropowerprojects.However ,thestationaryassumptionofflooddata
serieshasbeendestroyedbecauseoftheconstructionandoperationofupstreamreservoirs.Inthispaper,atime -
varyingmomentmodelfornon - stationaryfloodfrequencyanalysiswasbuilt ,andtheReservoirIndex(RI)ofkey
large - scalereservoirsintheupperreachesoftheYangtzeRiverwascalculated.Basedonthemonotonicallyin
creasingpropertyofthedistributionfunctionofthecontinuousfloodvariable,thetime - varyingP - Ⅲ curvefitting
methodwasproposedtofittheQ - Qplots.Thenon - stationarydesignfloodsduringoperationperiodoftheThree
GorgesReservoir (TGR)wereinferredandcomparedwiththoseofthemaximum likelihood(ML)method.The
resultsshowthat:(1)TheRIeffectivelyreflectsthenon - stationarityoffloodseries,servingascovariatesinthe
time - varyingP - Ⅲ model;(2)Thetime - varyingP - Ⅲ curvefittingmethodyieldsasmallerAICvaluethanthe
MLmethod,indicatingabetterfitnessforhistoricalfloodseries;(3)Comparedwiththeoriginalresults,Thenon -
stationarydesignfloodpeak ,alongwiththe3 - dayand7 - dayfloodvolumeofTGRaredecreasedbyabout18%,
whilethe15 - dayand30 - daydesignfloodvolumesaredecreasedbyabout14%.Theseresultsdemonstratethatthe
regulationofthecascadereservoirsintheupperYangtzeRiverdirectlyaffectsthefloodcontrolcapacityandcharac
teristicwaterlevelofTGRduringoperationperiod.
Keywords:designflood;non - stationary;non - continuousdataseries;time - varyingmoment;reservoirindex;
P - Ⅲ curvefittingmethod;maximumlikelihoodmethod;ThreeGorgesReservoir
(责任编辑:韩 昆)
— 6 5 3 —