Comparison and Assessment of Monthly Drought Determined by Two Drought Indices: CZI, SPI

Nature and Science 2015;13(8)

Comparison and assessment of monthly drought determined by two drought indices: CZI, SPI

SeyedVahid Shahoei1, Jahangir Porhemmat2

1. PhD candidate in Water Resources Engineering, Islamic Azad University, Science and Research Branch, Department of Water Resources Engineering, Tehran, Iran

2. Associate Professor of Soil Conservation and Watershed Management Research Center, Tehran, Iran

Abstract: Droughtindexby using rainfalldata, discharge, snowand etc give an understandablepictureofthephenomenonofthedrought.The droughtindexexpressesin anumber, which this numberisthe basis ofthe decision. In this studyassessesand comparesdrought indicesinthreestationsTehranand Qazvinand Khoramabad.To this endfromtwo drought indicesSPI andCZIis used.The data used is monthlyrainfalldata of selectedsynopticstationsduring theperiod of51years (1960-2010). Determiningthreshold of start, the continuation andthe end of thedrought specify byusing the Run theory. To evaluatedroughtindicesin order to monitoringthe droughtfromregression analysisis used. Following the numberandmonthlydroughtperiods have been determined byeach index. Finally,withdeterminingdistribution ofdrought(according tothe number andPeriod of timedrought) this subjectwasstudied that whetheraccording toprevious studiescarried outexistence of ahighcorrelationbetween thetwo droughtindices canbeindicates this subject that the results ofthetwo indicesin calculatingdroughtare the sameor not.

[SeyedVahidShahoei, Jahangir Porhemmat.Comparison and assessment of monthly drought determined by two drought indices: CZI, SPI.Nat Sci2015;13(8):1-7]. (ISSN: 1545-0740).

Keywords: drought; drought indices; Run theory; regression analysis

Nature and Science 2015;13(8)

1. Introduction

In terms ofRichardandHim(2002) Droughtis localphenomena,fromclimatologyphenomenathat everysome years leadsa lot of damages. Rainfallisthe most importantvariablethat its changes reflectdirectlyin thehumidity, surface flows andchanges of undergroundwater reservoirs. Sorainfallisfirstfactorthat canbe consideredin the review ofdrought(Lashnizand, 2003). Indicesthatin own calculationsuseonlyrainfall factorin comparison with thecomplex hydrological indicesgivemore favorableresults(Oladipoo, 1985). Fromdroughtindicesthathave been developedbased onrainfalldatacan bepointedtotwo indexesSPI andCZI.McKeeet al.in 1993 developed Standard Precipitation Index(SPI)for the purpose ofdefiningand monitoringdroughtand determininglackof precipitation intime scalesof 3, 6, 12, 24 and 48 months. Hayeset alin 1999came to the conclusionthattheusers of theindex SPIin the worldis increasing. In Iran alsothe results of studiesYazdaniandcolleaguesshow thatthe standardrainfallindex in seasonalscalehasbetter performance than theotherindexes (Yazdaniet al.,2011).lashniZandandcolleaguesin 2003and moghaddasiin 2005and Akhtarietalin 2006reachedtoa commonconclusion that the performance of SPIdroughtindexin comparison with otherdroughtindexismore appropriate.Wuand colleagues in2001 evaluated threedrought indexes,standardized precipitation(SPI), ZChinese(CZI)andZSIinfour regions inChinawhere had conditions drytowetwithusing48years ofprecipitation Statistics 1998 to1951 andperiod1, 3, 6, 9 and 12 months. In this researchSPI indexwasbasedand otherindices and its reactioninwet and dryyearsthanthisindexwere evaluated.For this study,fromlinear correlationbetween amounts of these indiceswere usedbySPI. Thiscorrelationbetween amounts ofSPI andCZIsuggest that indices SPI andCZIgenerallyshow goodrelationsindifferenttime scales, except intime scale 3monthandin conditionsvery dryof own. In total the results ofthe studyshowedthatthese indiceshave good abilityformonitoring droughtindifferenttime scales. Superiority ofZSIandCZIthantheSPI is in this respectthat thetwo indices unlike SPIaccept statisticaldeficienciesbetweendataseries and have the simpler calculation. On the other hand,CZIthanlack ofrainfall anddrought conditionsthan twootherindicesshowfurtherreactionandprovidesnegativegreateramountsthan the rest. While ZSIdid not showsa lot ofcapabilityforseveredrought. In this studyto evaluating andcomparing the results oftwodroughtsindexCZIand SPIin area studying.

Materials and methods

Study area

Irandue tobeing in geographicaldry ring anddesertstripis locatedat25to 40 degreesnorthern latitude, fromclimatic conditionsenjoysthat is considered partofthe areas with lowrain inworld.Due to thegeographicalpositionof the countryin thedry belt, rainfallin Iran is only the equivalent ofa third of theworld average.In this studythreesynopticstations have beenstudiedin Tehranand Qazvin andKhorramabad. In thisselective range,has beenselectedmonthlystatistics of rainfall3synoptic stationsintheperiod of51years (1960-2010) that characteristicsof selective synopticstationsis shownin Table 1.

Nature and Science 2015;13(8)

Table 1-stations studyingin researchand their properties

Station Name / Stationheight(m) / Latitude(m) / Longitude(m) / Number
Tehran / 1191 / ´41 °35 / ´19 °51 / 1
Khorramabad / 1148 / ´26 °33 / ´17 °48 / 2
Qazvin / 1279 / ´15 °36 / ´03 °50 / 3

Nature and Science 2015;13(8)

The selection of these3stations has been made with consideringproximity of characteristics of heightandgeographicallength and width, all these casecanaffectrainfall.

Droughtindices

As mentioned, in this study is used2drought indexin order toevaluating and comparing theirresultsin the analysis ofthe drought inmonthlytime scale.Each of theseindices with relying onrainfalldataindifferenttime scales(in this studymonthlytime scale) provides quantitative numeral whichisusedin the evaluation ofdrought. The generalcharacteristics2selectivedrought indexbased onrainfalldatacan besummarizedinTable 2.

Nature and Science 2015;13(8)

Table 2.Characteristicsof droughtindicesused in thestudy

Provider Name / Year of providing / Index Title / Symbol
Mckee / 1993 / Standard precipitation Index / SPI
Ju et al / 1997 / Chinese Z Score Index / CZI

Nature and Science 2015;13(8)

Standard Precipitation Index

Standard Precipitation Indexwas introducedin 1993 byMcKeeet al. This indexobtains based ondifferences in level of rainfallfrom the average amountfor aspecifiedtime scaleand thendividing itonstandard deviation. The only factorinvolvedin the calculation ofthe indexisRainfallelement. Thisindexcan be calculated intime scalesof 1, 3, 6, 9, 12, 18, 24 and 48 months anda one-year. Based on thisindex can bedetermine droughtthresholdforany time period. Standard Precipitation Index needs anexpectablefixed amountandmeaningfulvariancein order to comparingindex mentioned in differentstations(McKay et al.,1993). Relation of this indexisin relation 1.

(1)

In which:

: The amount ofprecipitation inthe slightly period : long-termaverage of rainfallfortheslightly periodand is the standard deviation of the amount rainfall.

Chinese ZIndex

This indexwascreatedin 1997by Joeandcolleagues. Relations that usedto calculateCZIindex is theasrelations2 to4:

(2)

(3)

(4)

In which:

- j: slightly month;

-: Coefficient ofskewness;

-: Standardvariable;

-: Precipitation of monthj;

-: Average of rainfalldata;

-: The total number ofmonthsduring thestatistical period;

-: Standard deviationof rainfalldata

Runtheory

FromCommon methods of droughtanalysis, isRUN theory. (Yujnovich, 1967) with usingRUN theory,three important featuresof the droughtnamely continuity,enlargementand intensity ofdrought defined and explained andconcluded that theRUNtheoryhaseffectiveapplicationsin the analysisthe process ofdroughts. Thistheorydeterminesprocess ofstart andends of thedrought,according tohigh limitandbottom limit of acritical threshold. Inall studies of droughtthathave been done, using of RUNtheory have appliedas the firststep monitoringdroughtthatsome of thesestudiescan be refertostudies of (Chung & Salas, 2000), (Morid, 2006,Shiau, 2006),(Lee et al, 2013).For start of drought,everytwo selective indices ofthresholdhave the same start.

Nature and Science 2015;13(8)

Figure 1-Testhomogeneity of Monthlyrainfalldata in stationsstudying

Nature and Science 2015;13(8)

Start of drought in SPI indexandCZIis whenamount of the indexissmallerthan -1. In RUN theorythatisusedto definedrought, droughtbegins whenthatnumerical amount of drought indices are underlevel of threshold of this theory.According to thechart ofthistheorycan be seen thatin some casesdespite neveramounts of drought indexes CZIand SPI do not get to less than -1, butcumulativeamountof thisIndexand therefore theits impactismore thanotherdroughtindexthat amount of index have got toless than -1.On theother side in some casesthatdrought thresholdis -1 and has defined according to thiscertain limit,hasperiod of time veryless than conditionsthatamounts of Index neverarenot got to -1but continuous and prolonged is below thethreshold limit.For solvingthese problems anddefining aspecificthreshold limitfortwoindexof study, the zeroamounts of these indicesareconsideredas the boundary of beginning the drought and with regard tothis border of drought,start ofdrought is when numerical amount of the droughtindexes studying is being lessthan zero.

Initialscrutiny of statistical data

The first step indesigns andhydrologystudies is initialanalyzingofstatistical datausing. Whenthe correctnessandaccuracy of thedatais notreliable, the results obtained can not be reliable. Review andanalysis ofhydrologicaldata is including completingand correctingdatacutoff and incomplete. Checking homogeneity, stability and randomness of data is also including actions thatshouldbe examinedbeforeanalysis andstudy. In thisstudy in order totestingthehomogeneity of themonthly data of rainfallin stations studying from doublemass method, in order to studying therandomness of dataRun test andto investigatestability of used datawasusedSpearmanrankcorrelationtest (Moghaddamet al.,2012). The results ofeach these testsare shownbelow.

Regression analysis

If thetwo sets ofdata are being available, the Regression method determinesdependence ofthese twoseriesrelative to each other and specifies correlationbetween the two setsdata in quantitative. Also this method specifies equation ofthe bestlinearthat can bedrawnbetween the twoseries. A theorythatregressionusesto obtainthe correlation is theory ofthe leastsquares, in fact linear regressionis a linethat the sum of thesquare of the distance of pointsfrom thelinethaneachother line have the minimum amount ofits own. In thisstudy in order to evaluating and comparing theresults oftwodroughtindicesSPI andCZIfromlinear regression was used.

Results and discussion

Homogeneity ofdata

As mentioned before,in eachhydrologicalstudy and researchshould besure from randomness, homogeneity and stabilityof used datain thestudyuntilresults of thestudybeingreliable. In this study with usingmentioned methodcheckedrandomness, homogeneity andstability of themonthlyrainfalldataofselectedstationsover a statistical period of51yearsthatits results have camebelow.

Chartsrelating todoublemass testto evaluate thehomogeneity ofthe monthlyrainfalldata3stationsexaminingin this studyduring the Statistical period51years have came inFigure 1.

Asyou can see, all stationsexamining according to doublemasstestarehomogeneous.

Stability andrandomness of data

Results of Spearmanrankcorrelationtest and runtest to examine stability andrandomness of studiedrainfall data in stationsatthe annualtime scale, issummarized in Table 3.

Nature and Science 2015;13(8)

Table3- resultsofthetest the stability and randomness ofannual rainfalldataat the stationsstudying

data stability test / data Randomnesstest / Station Name
Permitted range / computational amount / Permitted range / computational amount
± 2.01 / - 0.02 / ±1.96 / 1.75 / Tehran
±2.01 / 1.22 / ±1.96 / 0.23 / Khorramabad
±2.01 / -1.85 / ±1.96 / -0.23 / Qazvin

Nature and Science 2015;13(8)

Asis clearfrom the results, allthe rainfalldatainthe studyarerandomandstatic.

Determiningmonthlydroughthappened in theperiod of51years at thestationsexamining

By using runtheory and by using theamounts of critical levelfor eachtwodrought indexconsidered, droughtsoccurredwill be specified.Asbeforementioneddroughthappens whenamounts of indexCZI andSPI are below ofcritical level(zero level). Changes of amountsdroughtindex at each stationisshown in Figure 2.

Nature and Science 2015;13(8)

Station / SPIdroughtindexchanges chart / CZI drought index changes chart
Tehran / /
Khoramabad / /
Qazvin / /

Figure 2-Chart of changesdroughtindicesexamining over the51-year statistical period in theStations studying

Nature and Science 2015;13(8)

Regressionanalysis in order tocompare thedroughtindex the studied

In this study,after identifying thedroughtby2indexesSPI andCZI,by usingregression analysishas beencomparedtwo by two results ofeachindexinall theelected stations..By usingtestthatwasmentioned before have specifiedmeaningfulness or non-meaningfulness ofcorrelation betweenresults ofindicesthat the resultsareshown in Figure3.

Nature and Science 2015;13(8)

Fig. 3. Linearregressionbetweendroughtindicesstudied inselectedstations

Figure 4-Chartfrequency ofmonthlydroughtsoccurredbydrought indexesexamining in stations studying.

Nature and Science 2015;13(8)

Astheresults of theregression analysisshowtwo indicesCZIand SPIin the3stationsexamining havevery highcorrelationsowe should expectthatmonitoringdroughtbythese two indicesbesame.Inthe rest of thisarticlewe willexamine whetherexistence of a highlinear correlationbetween the twoindexesindicatesequality of determination ofdroughts in stationsexaminingor not?

Distribution of drought dispersion

In this section,were examined thedroughtscalculated byeach drought indices andconsidering period ofdrought. In fact,the number ofdroughtsoccurreddue to period ofdrought,by anyindexin allstations studying have been specified thatitschartsisin Figure 4.

Asis clearfrom theresults, in thethreestationsexamining frequency andperiod ofdroughtsdiffers from each otherand in none of thestations,the numberand duration ofmonthlydroughtsdetermined bytwoindexesSPI andCZI is notidentical.

Conclusion

In this studyoverallpresentedassessment and comparison ofthe results of thedroughtindexCZIand SPIandthe relationshipthese indicestogetherin the analysis ofdrought. According to theresults obtainedit was found thatinall thestationsexamining,alwaystwo indexesCZIand SPIhavehighcorrelationand the amount ofthiscorrelationin allstationsareup 0.90. According tothese resultsit can be expectedthattwo droughtindexesSPI andCZI droughtsin monthly scale monitor alike, andconcluded thatinstudiesand research projectscan be sure from equality ofthe results of thetwo indexes. But whenmonthlydroughts determinedweredetermined bytwoindexes,was found those slightly drought indexes monitors number and periods of monthlydroughtcompletelyseparately.So thatnone of thestationsexaminingthe numberand duration ofdrought, which are determinedbytheindexSPI andCZI, are not the same.Thus, fromthisresearchcan beconcluded suchthathighlinear correlationbetweendroughtindices(in this study,SPI andCZI)can not be enough witnessfor confidence of theequality of resultsdetermining droughtbytheseindicesand in estimating droughtarein monthly scale bythese indicesshouldbecareful thatresultsof thesetwo indices, despite the highlinear correlationcan be quitedifferent.

References:

Akhtari, R.; Mahdian, M. and Morid, S. (2006). Analysis ofdroughtindicesSPI andEDIplacein Tehran. Water Resources Research1(4): 27-38.
Yazdani, V.; ZareAbyaneh, H.; and Shademani, M. (2011). "Analysis ofthe frequencyofdroughtsandzoningusingStandardized PrecipitationIndex(SPI).Journal ofWater Resources Engineering, Vol. IV,pp.31-42.
Samiei, M. and Talouri, A. (2005). "Theseverity and duration ofhydrologicaldroughtin thecatchmentareasof Tehran." Researchanddevelopmentofnaturalresources, No. 79, pp. 22-27.
LashniZand, M. (2003). "Theintensity,duration andfrequency ofclimaticdrought." ThirdRegional Conferenceandthe NationalConference onClimate Change, Isfahan.
Moghadasi, M. and Moridi, S. (2005). "Theintensity,duration andfrequency ofclimaticdrought." ThirdRegional Conferenceandthe NationalConference onClimate Change, Isfahan.
Wheredroughtyears1999-2000to2000-2001in TehranmonitoringusingindicatorsDI, SPIand EDIandGIS.QuarterlyJournal of, Vol. 9, No. 1, pp.197-220.
Moghadam, S.; Khani, Z. and Montaseri, M. (2003). "Effects of climate change onthe severity, duration and courseofSPI,droughtandclimatemodels(case studyUrmia Lakebasin)." Fifth Conference ofIran Water Resources Management.
Chung, C.H. and salas, J.D. (2000). Drought occurrence probabilities and risks of dependent hydrologic processes. Journal of Hydrologic Engineering ASCE, 5(3), 259 – 268.
Hayes, M. J., Svoboda, M. D., Wilhite, D. A., and Vanyarkho, O. V. (1999). Monitoring the 1996 Drought Using the Standardized Precipitation Index. Bulletin of the American Meteorological Society, Vol. 80, NO. 3, 427 – 440.
Joe, H. (1997). Multivariate Models and Dependence Concepts. Chapman and Hall, Boca Raton, Fla.
Lee, T. Modarres, R., and Ouarda, T.B.M. J. (2013). Data – based analysis of bivariate copula tail dependence for drought duration and severit, Hydrol. Process. 27: 1454 – 1463.
McKee, T.B., Doesken, N.Y. and Kleist, Y. (1993). The relationship of drought frequency and duration to time scales. Preprints, 8th conference on Applied Climatology, 17 – 22 January, Anaheim, C.A., 179 – 185.
Morid, s., Smakhtin, V., Moghaddasi, M. (2006). Comparison of seven Meteorological Indices for Drought Monitoring in Iran. Int.J.Climatol.26: 971- 985.
Oladipo, E.O. (1985). A comparative performance analysis of three meteorological drought indices. Journal of Climatology, Vol6, 495 – 514.
Richard, R.Heim. (2002). A review of twentieth century drought indices used in the United States. American Meteorogical society, 1148 – 1165.
Shiau, J.T. (2006). Fitting Drought Duration and Severity with Two – Dimensional Copulas. Water Resources Management (2006) 20: 795 – 815.
Wu, H., hayes, MJ. Welss, A., Hu, Q. (2001). An evaluation the standardized precipitation index, the china – z index and statistical Z – score. International Journal of Climatology 21: 743 – 759.
Yevjevich, V. (1967). An objective approach to definitions and investigation of continental hydrological droughts, Hydrology paper 23, Colorado State University, Fort Collins, Co, 18.

Nature and Science 2015;13(8)

6/29/2015