Abstract:
Drought indicesareusedtomonitordroughtconditionsofaregion.Variousdroughtindices(DIs)have
been proposedinpastfewdecades,butsomeofthoseareregionspecific andhavelimitationsofap-
plicability inotherclimaticconditions.Also,multipletimestepsofDIsmakeithardtodecidethatwhich
time stepisthebesttoshowthedroughtcondition.
Present studyaimstocompareStandardizedPrecipitationIndex(SPI),EffectiveDroughtIndex(EDI),
statistical Z-Score,China Z-Index(CZI),RainfallDeparture(RD),RainfallDecilebasedDroughtIndex
(RDDI) fortheirsuitabilityindroughtpronedistrictsoftheKenRiverBasin,locatedincentralIndia,
where therainfallisconcentratedinthemonsoonseason(June–September)andfrequentoccurrenceof
severedroughteventsarecommon.AllselectedDIswithmultipletimestepsareappliedtocomputethe
severity for five timestepsof1,3,6,9,and12-month,andcomparedwitheachotherandEDI.
The studyrevealsthat(1)1-monthtimestepinallDIsmayproduceerroneousestimatesofdrought
duration. (2)Thedroughtindicescomputedfor9-monthtimesteparebestcorrelatedwitheachother.
However,thedroughtdurationandthedroughtfrequenciesestimatedusingRDandRDDIareindis-
agreement withotherDIs,therefore,thesearenotsuitableforthisareawherethesummerconcentration
of precipitationisveryhigh.(3)TheDIsarehighlycorrelatedatsametimestepsandcanalternativelybe
used. However,theyarepoorlycorrelatedatdissimilartimesteps,whichmakesithardtoassesswhether
the droughtoccurredornot.(4)Becausetherearenoobjectiverulestoselecttheappropriatetimestep,
and theidentified droughtdurationvariestoomuchwithdifferenttimesteps,itisveryhardagainto
assess whenthedroughtoccurred.
However,EDI,thathasself-defined timestepinitself,andfreefromtimestepproblem,(1)iscor-
relatedbetterwithotherDIsforalltimestepsandeffectiveonlongandshortdroughttogether,with
highest correlationat9-monthtimesteps,(2)identifies thedroughtconditionearlierthananyother
indices, therefore,(3)isfoundtobemoresuitabledroughtindexforthestudybasin.Thisisinagreement
with theresultofEDIapplicationinKorea,Japan,Turkey,AustraliaandIranthoughthemethodsofits
testing aredifferent.