Regiona lLo wFlo wFrequenc yAnalysi sof UpperAwas hRiverBasin ,Ethiopia

Abstract

Estimatesoflowstreamflowstatisticsarerequiredforavarietyofwaterresource applications.Atgaugedriversites,th eestimationoflowstreamflowstatistics requiresestimationofannualn-dayminimumstreamflows ,lowflowestimates, asdeterminedbyprobabilisticmodelingofobserveddatasequences,are commeniyusedtodescribecertainstreamfiowcharacteristics.Unfortunately, however,reliablelow-flo westimatescanbedifficulttocomeby,particularlyfor gaugingsiteswithshortrecordlengths .Theshortnessofrecordslead sto uncertaintiesnotonlyInth eselectionofadistributionformodelingpurposesbut aisointheestimationofth eparametersofachosenmodel.Infloodfrequenc y analysis,thecommonapproachtomitigatesomeoftheseproblemsisthroug h theregionaiizationoffrequenc ybehavior.Thesamegeneralapproachisappiied heretoth ecaseofiow-fio westimation,withthegeneralintentofnotoniy ImprovingIow-fiowestimatesbutalsoillustratin gtii egainsthatmightbeattained indoingso. i nthisstudyio wfio wregionaiizatio nhasbeenperformedusingthre e approaches;fiowdurationcurves,basefiowinde xandiowflowfrequenc y analysis.Lowfio wfrequenc yanalysisofupperawashriverbasinhavebeen carriedoutusing7-dayannualminimumfiowseriesextractedfromtheaverage dailystreamflow sobservedat11streamfiowgaugingsitesacrossupperAwash Riverbasin.L-Momentratiodiagramswereusedforidentifyin gandgroupingof station;intohydrologicalhomogeneousregionsandhencetheunderlying statistinaidistribution.Accordingly,thebasinwasdelineatedintotvz o homocaneousregions .Thehomogeneityofthedelineatedregionswasteste d usingtheCv-basedanddiscordancetests .Thegoodnessoffittest(Z-statistics) andtheminimumstandarderrorofestimateshavebeenusedforassessingthe suitabilityoftheselecteddistribution.Consequently,thelognurmaidistributionis foun dtobeasuitablecandidateforregiononeandthegeneralizedPareto distributionforregiontwo .Themethodofprobabilityweigntedmomentswas consideredasth ebestparameterestimationprocedureascomparedwiththe methofiofmomentsandthemaximumlikelihoo dmethod. V I Flowdurationcurvesof1,7,10,30dayshavebeenderivedfor11slations.Tiic basinsfiowdurationcurvesarecharacterizedintotwohomogeneousclasses basedonthecharacteristicsshownontherighttailendofthebasinsfiow durationcurves.Basefio wseparationhasbeenmadebythestandardmethod calledtheinstitut eofhydrologymethod.Accordinglyth ebaseflowindexis calculatedandthebasinisgroupedintothreeclassesbasedontheirrangeof BFi.Regressionanalyseswereconductedonth edelineatedregionstopredict theinde xiowflow s(mea nannual7-dayminimumfiow )forungaugedcatchments usingcatchmentcharacteristics.ThemultiplecoefficientsofdeterminationsR2 wasusedasameasureofth eabilityoftheregressio nmodeltodescribe variationsinthedependentvariables.ThevalueofR2wasfoun dtobe0.96for regiononeand0.97forregiontwo.inadditionthestandardizediowfiow frequenc ycurveswhichwiiibeusedforestimatingiowfiowquantilesat ungaugedcatchmentshavebeenestablishedforth etw odelineatedregions usingthestandardizedflo wdata.