Improvement of the drought indicators system in the Júcar River Basin, Spain

2018·3.8 MB·English

by  Tatiana Ortega-Gómez

ScienceoftheTotalEnvironment610–611(2018)276–290
ContentslistsavailableatScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
Improvement of the drought indicators system in the Júcar River
Basin, Spain
TatianaOrtega-Gómeza,⁎,MiguelA.Pérez-Martínb,TeodoroEstrelaa,b
aConfederaciónHidrográficadelJúcar(CHJ)JúcarRiverBasinAuthority,Avd.BlascoIbáñezno48,46010,Valencia,Spain
bResearchInstituteofWaterandEnvironmentalEngineering(IIAMA),UniversitatPolitècnicadeValencia,CaminodeVeras/n,46022,Valencia,Spain
H I G H L I G H T S G R A P H I C A L A B S T R A C T
• SPI-12orSPI-24canbeusedtodefine
the “prolonged drought” required by
EU.
• There is a high correlation between
operationaldroughtindicesandlong-
termSPI.
• Moisture content index is correlated
withtheshort-termprecipitationindex.
• Variousindicesarenecessarytodetect
differenttypesofdrought.
• A unique aggregated indicator could
hideasignificantdroughtinaspecific
area.
a r t i c l e i n f o a b s t r a c t
Articlehistory: Droughtsareoneofthegravestnaturalthreatscurrentlyexistingintheworldandtheiroccurrenceandintensity
Received15June2017 mightbeexacerbatedinthecomingyearsduetoclimatechange.Thesevereimpactsthatdroughtscausetoin-
Receivedinrevisedform27July2017 landwaterresourcesandtotheassociatedsocio-economicactivitiesjustifythecontinuousmonitoringofthe
Accepted28July2017
drought.Thecasestudypresentedshowsapracticalapplicationofadistributeddroughtmonitoringsystemim-
Availableonline11August2017
plementedinarealriverbasindistrict,theJúcarRiverBasinDistrict(43,000km2),wheredroughtperiodsof
markedintensityhaveoccurredhistoricallyandtheclimaterangesfromhumidinthenorthtosemiaridinthe
Editor:D.Barcelo
south.Fivedroughtindiceshavebeenapplied:StandardisedPrecipitationIndex(SPI)formeteorological
Keywords: drought;PalmerDroughtSeverityIndex(PDSI)andanewsoilmoistureindex(HI),foredaphicdrought;Normal-
distributeddroughtmonitoring isedDifferenceVegetationIndex(NDVI)forthevegetationactivity;andSpanishStatusIndex(SI),fortheoper-
droughtindices ationaldrought.Allindicesarestandardisedtocomparethem.
remotesensing TherelationshipbetweenthestandardisedoperationaldroughtindexSIandthelong-termmeteorologicalindi-
prolongeddrought ces,SPI-12orSPI-24,showthatinamediumsizebasintheconceptof“prolongeddrought”requiredbytheEu-
ropeanCommission undertheWaterFrameworkDirectivecouldbedefinedbytheuseof accumulated
precipitationindices.Thenumberofmonthstobeaccumulateddependsonthesizeofthebasinandthewater
managementsystemproperties.Inlargebasins,suchastheJúcarriverbasin(22,000km2),therearesignificant
deviationsduetothespatialdistributionofthedrought.Theuseofauniqueaggregatedindicatorcouldhidea
significantdroughtinaspecificarea,orontheotherhandshowanon-realdrought.Evolutionofdroughtindices
foreachwatermanagementsystemmustbeaccompaniedbyspatiallydistributeddroughtmapstobetterunder-
standthedroughtstatusanditsevolution.
©2017ElsevierB.V.Allrightsreserved.
⁎ Correspondingauthor.
E-mailaddress:[email protected](T.Ortega-Gómez).
http://dx.doi.org/10.1016/j.scitotenv.2017.07.250
0048-9697/©2017ElsevierB.V.Allrightsreserved.
T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290 277
1.Introduction circumstances are exceptional or could have not reasonably been
foreseen.
Droughtsareanaturalhazardfortheenvironment,theeconomy Thismanuscriptincludestheapplicationofasetofspatiallydistribut-
andthesocialdevelopment.Althoughbyitselfisnotadisaster,whether eddroughtindicatorsinarealcase,theJúcarRiverBasinDistrict(RBD),
itbecomesonedependsonitsimpactsonsocietyandenvironment whichisalargedistrict(43,000km2)formedbytheaggregationofseveral
(WilhiteandBuchanan-Smith,2005).Droughtsarepresentallover riverbasinsandislocatedintheSpanishMediterraneanarea.TheJúcar
theworldandaffectthearidandsemi-aridregionsandalsohumid RBD includes nine water resources systems with different climates,
andsub-humidregions(MishraandSingh,2010).Thespatialextension, fromhumidtosemiarid,sothedroughtindiceswouldbetestedindiffer-
fromlocaltoregionalscale,andthedurationintime,fromweeksto entclimateconditions.Climatechangecouldbeasignificantimpactinthe
years,alsocanbeveryvariable(Andreuetal.,2015).Droughtsarea naturalresourcesofthisriverbasin(Estrelaetal.,2012)anddrought
slow-onsetnatural hazard(Sivakumar et al., 2011),whichappears couldbeincreasedinnumberandintensityinthefuture(Pérez-Martín
whenprecipitationisunderthenormalvaluesduringaperiodoftime. etal.,2015).Thestandardisedformoffivetypesofindicesareapplied:
Theunusuallowvaluesofprecipitationcanbetransferredinafast formeteorologicaldrought,theStandardisedPrecipitationIndex(SPI;
way(weeksorafewmonths)toothercomponentsofthehydrological McKeeetal.,1993);foredaphicdrought(agricultural)amodifiedPalmer
cycle(soilmoisture,riverflows),oronthecontrary,inamuchslower DroughtSeverityIndex(PDSI,Palmer,1965)andanexpresslycreated
one(manymonthsorevenyears)toothercomponentsofhydrological HumidityIndexinsoil(iHI);forthevegetationresponse,thestandardised
cyclesuchaspiezometriclevelsorgroundwaterdischarges.Dueto NormalisedDifferenceVegetationIndex(iNDVI;Jordan,1969),andfor
thesecharacteristics,theDroughtManagementPlanshavedemonstrat- theoperationaldrought,thestandardisedStatusIndex(iSI).
edtobevaluablemanagementtoolstofightthistypeofanomalies TheiSIandiNDVindicesaredirectlyderivedfromobserveddataand
(EstrelaandVargas,2012). reflecttherealsituationobservedintheriverbasin.Thefirstonemainly
TheDroughtIndicatorSystemhasasignificantroleinthedeclara- indicatestheamountofavailablewatertosupplywaterdemands.The
tionofadroughtandintheapplicationofthemeasuresdefinedinthe secondoneindicatesthevegetationactivityinamonthiNDVI,inasea-
DroughtManagementPlans(EstrelaandSancho,2016).TheDrought son(threemonths)iNDVI-3orduringsixconsecutivemonthsiNDVI-6,
IndicatorSystemisapowerfultooltosupportthewatermanagers derivedfromtheEOS-AquasatelliteequippedwiththeMODISsensor.
andfinaluserstounderstandhowandwherethedroughthasbeenoc- Themeteorological(SPI)andedaphic(PDSIandiHI)indicesarederived
curring.Thereisawiderangeofdroughtindicatorstodetermineeach fromprecipitationandtemperaturedata.Differentmonthlyaccumula-
typeofdrought:meteorological,definedasalackofprecipitationover tionsforSPIareconsideredfortheshort-term(SPI,SPI-3,SPI-6)and
aregionforaperiodoftime;hydrological,relatedtoaperiodwithinad- thelong-term(SPI-12andSPI-24).ThemodifiedPDSIandtheiHIare
equatesurfaceandsubsurfacewaterresourcesforestablishedwater obtained by the water balance in soil calculated with the Patrical
usesofagivenwaterresourcesmanagementsystem;agricultural,refers model(Pérez-Martínetal.,2014).
toaperiodwithdecliningsoilmoistureandconsequentcropfailure TheremaindermanuscriptincludesinsectiontwotheDataSetand
withoutanyreferencetosurfacewaterresources;andsocio-economic theindicesused.Thethirdsectiondescribesthestudycaseresults.Dis-
drought,associatedwithfailureofwaterresourcessystemstomeet cussionisincludedinthefourthsectionandfinally,thefifthsectioncon-
waterdemands(MishraandSingh,2010).Anextensiverevisionof tainstheconclusions.
droughtindicesisincludedinMishraandSingh(2010)andPedro-
Monzonísetal.(2015).Besides,remotesensingprovidesasynoptic 2.Datasetandindicesused
viewoftheEarth,andisanadvantageoussourceofinformationineval-
uatingthedroughtimpacts(QuiringandGanesh,2010). TheJúcarRiverBasinDistrict(JRBD)islocatedintheeasternpartof
ThecurrentDroughtIndicatorSysteminSpainisformedbasicallyby theIberianPeninsulainSpainandisformedbytheaggregationofriver
anoperationaldroughtindex,theStatusIndex(SI).Thisindexreflects basinsthatflowintotheMediterraneanSea.Thewholeterritoryin-
theamountofavailablewaterfortheendusersineachmonth,inrela- cludes nine water resources systems (WRS or system), 1) Cenia-
tionwiththeamountofavailablewaterforthatmonthhistorically.This Maestrazgo,2)Mijares-PlanadeCastellón,3)PalanciaylosValles,4)
indexisusefultocharacterisethesocio-economicdrought.Theindex Turia, 5) Júcar, 6) Serpis, 7) Marina Alta, 8) Marina Baja and 9)
consistsofselectedcontrolpointsdistributedthroughouttheSpanish Vinalopó-Alacantí,anditstotalareaisaround42.735km2(Fig.1).The
RiverBasinDistrictswhichincludethefollowinginformation:volume JúcarRBDpresentsaMediterraneanclimate.Thetotalannualprecipita-
storedinsurfacereservoirs;groundwaterinaquifers;riverflowdis- tionisaround500mm,oscillatingbetweenmaximumannualvaluesof
charges; reservoir inflows and precipitations in those areas where 780mmforthewetyearsandjustover300mmforthedryyears.Pre-
theyaresignificantinrelationtothewaterresourcesavailabilitytosup- cipitationintheautumnisalmosthalfoftheannualprecipitationinthe
plythemainwaterdemands(EstrelaandVargas,2012).Thisindexisan coastalarea.Thesecondhighestvalueoccursduringthespringand,
aggregatedindexforsocio-economicdroughtineachwaterresources duringthesummer,rainisalmostnon-existent.Thesamevariability
system,whichshouldbecomplementedwithasetofotherindicators canbeobservedinregardstospatialdistribution.Thereareareassuch
thatshowtherestofdroughttypesandtheirspatialvariability. asMarinaAltawithaverageannualvaluesaround730mm,andmaxi-
Recently,in2016,theRoyalDecreebywhichSpanishriverbasin mumvaluesof1.325mm,whereasotherareaslikeVinalopó-Alacantí
managementplanswereapprovedestablishedthatindicatorsystems receivemuchlessrainfall,withaverageannualvaluesof345mmand
ofRiverBasinAuthoritiesshouldbeabletoseparatelydiagnosesitua- minimumvaluesof190mm.Approximately80%ofthetotalwaterre-
tionsofdroughtandwaterscarcity.TheEuropeanWaterFrameworkDi- ceived in the form of precipitation is returned to the atmosphere
rectivedeterminesthattemporarydeteriorationinthestatusofwater throughevaporation,whiletheremaining20%isgroundwaterandsur-
bodiesshallnotbeinbreachoftherequirementsofthisDirectiveif facerunoff(Pérez-Martínetal.,2014).Theannualvalueofpotential
thisistheresultofcircumstancesofnaturalcauseorforcemajeure evapotranspirationfortheperiod1940/41–2010/11is890mm/year,
whichareexceptionalorcouldnotreasonablyhavebeenforeseen,in actual evapotranspiration is around 409 mm/year, infiltration is
particularextremefloodsandprolongeddroughts,ortheresultofcir- 64mm/year,surfacerunoffis31mm/yearandtotalrunoffisabout
cumstancesduetoaccidentswhichcouldnotreasonablyhavebeen 95mm/year(Pérez-Martínetal.,2014).
foreseen.TheEuropeanCommission(EC,2007)indicatesthattheiden- TheJúcarRBDhasapopulationofaround5millionpeople(11%of
tificationofsituationsofprolongeddroughtshouldbeperformedusing the Spanish population) and has an irrigated surface of about
naturalindicatorsbasedonprecipitationasthemainunderlyingparam- 389.000hamainlyconcentratedinthelowerMijares,Palancia,Turia
etertoindicatethatitisa‘naturalcauseorforcemajeure’,andthatthe andJúcarbasins,theregionofManchaOriental,andtheirrigatedvalleys
278 T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290
Fig.1.LocationoftheJucarRiverBasinDistrictandtheninewaterresourcessystems:1)Cenia-Maestrazgo,2)Mijares-PlanadeCastellón,3)PalanciaylosValles,4)Turia,5)Júcar,6)
Serpis,7)MarinaAlta,8)MarinaBajaand9)Vinalopó-Alacantí.
oftheVinalopóandMonegrerivers.Thisterritoryhashistoricallybeen VegetationIndex(NDVI).Themeteorologicaldatainrealtimecome
markedbyperiodsofintensivedroughtsthathavecompromisedthe fromtheJúcarAutomaticHydrologicalInformationSystem(SAIHin
management of water resources in the Júcar RBD. The Júcar River Spanish;Estrela,etal.,2007).TheSAIHincludes181raingaugesand
BasinAuthorityguaranteesthesupplyoflargepopulationslikethe 24thermometers(Fig.2)with5-minutesdatafrom1991to2015and
cityofValencia,AlbaceteandTeruel,andisresponsibleforthewater othertemporalaggregations,suchasdailyormonthlytimestep.The
supplyofirrigatedareas,andotherrecreational,industrialandenergy sparsedensityofavailablethermometersatthestartoftheseriesjusti-
wateruses,aswellasforthecompliancewiththeenvironmentalre- fiedtheuseofothersourcesofdailytemperature,sotheSAIHtemper-
quirements.Inordertosatisfythesedemands,theDistrictcountson ature series has been completed with data from the Spanish
27majorreservoirsandatotalwaterstoragecapacityof3,300hm3. MeteorologicalStateAgency(AemetinSpanish).
Alarcón, Contreras and Tous are the biggest reservoirs in the river MonthlymeteorologicaldatacomingfromSAIH,precipitationand
JúcarandBenagéberintheTuria(Fig.1).Afurtherdescriptionisinclud- temperature,werepreviouslyvalidated.Thevalidationprocessincludes
edinFerreretal.(2012). twoprocesses:thefirstprocessistheautomaticidentificationofpossi-
Thebasicgeographicalunitusedinthisstudyisthewaterresources bleoutliersbasedontheapplicationoftwooutliers'detectionmethods:
system(WRSorsystem).Thisallowsdiscretisingthespatialareaof aseasonal-trenddecompositionprocedurebasedonLOESS(Cleveland
droughtinsmallerterritorialunitswithacharacteristichydro-meteoro- etal.,1990)andaLOFmethod(Breuningetal.,2000).Thesecondpro-
logicalbehaviourwithintheglobalscopeoftheDistrict.Inaccordance cessisadetailedanalysisoftimeseries,dailyand5-minutestimestep,
withtheSpanishHydrologicalPlanningRegulation,awaterresources ofthepossiblemonthlyoutliersdetected.
systemisformedbysurfacewaterbodies,groundwaterbodies,hydrau- Itwasfoundthatsomeoftheobservationsshowapatternofunusual
licinfrastructureworksandwateroperationrulesthatallowestablish- behaviour.Inthoseraingaugeslocationswhereextremevalueswere
ingthewatersupplybasedontheavailableresourcesandincompliance recorded,intensitycurveswerereviewedandthereturnperiodfor
withenvironmentalobjectives. eachrecordwascalculated(CEDEX,1999)and,whereappropriate,
someanomalieswerecorrectedincollaborationwithtechniciansfrom
2.1.Dataset theSAIHoftheJúcarRBA.
TheNormalisedDifferenceVegetationIndex,NDVI,comesfromthe
Indicesarecalculatedfrommonthlymeteorologicaldata,precipita- product MYD13Q1 from the EOS-Aqua satellite which is equipped
tionandtemperature,andsatellitedata,15daysNormalisedDifference withtheMODIS(ModerateResolutionImagingSprectroradiometer)
T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290 279
Fig.2.PrecipitationandtemperatureSAIHnetwork.
sensor.Thesatellitedataseriescoverstheperiod2002–2015.TheNDVI 2.2.Indicesused
valueswereextractedwhenthesatellitesweepsthesurfaceoftheJúcar
RBDaround13h30UTC,whenvegetationismorestressed.Thesensor Fivedroughtindicesareanalysedinthemanuscript(Table1):the
hasaviewingswathwidthof2330kmthatcoverstheentiresurface StandardisedPrecipitationIndex(SPI;McKeeetal.,1993),amodified
oftheEartheveryday.Theimagesaresuppliedin.hdfformat,designed PalmerDroughtSeverityIndex(PDSI,Palmer,1965),thestandardised
tostorelargeamountsofdigitaldataandaregeographicallyreferenced NormalisedDifferenceVegetation Index (iNDVI; Jordan,1969), the
toasphericalcoordinatessystem.Thetemporalresolutionistherefore standardisedStatusIndex(iSI)derivedfromtheSpanishStatusIndex
16daysandthespatialis250m2. (SI)andtheexpresslycreatedHumidityIndexinsoil.Adetailedreview
ThedownloadingoftheimagescoveringtheterritoryoftheJúcar of the indices is included in Mishra and Singh (2010) and Pedro-
RBDwasautomated.SinceJuly2002ithasprovidedoneortwoimages Monzonísetal.(2015).
permonth.Eachimageisformedbytheunionoffourscenes:h17v04 ThefirsttwoindicesareintegratedintheGIS-basedmonthlywater
(Scene1),h17v05(Scene2),h18v04(Scene3)andh18v05(Scene4). balancemodelPatrical,whichisahydrologicalandwaterqualitydis-
Next,thetwomonthlyimages(eightscenes)coveringthescopeof tributedsimulationmodel(Pérez-Martínetal.,2014;Pérez-Martínet
theJúcarRBDwereunifiedinasinglerasterimageaveragingNDVI al.,2016).Themodelisusedforthewaterresourcesassessmentofthe
valuesineachpixeloverlap. riverbasin,includinglong-termresources,spatialvariability,inter-
TheNDVIsatelliteimagesprovidedirectvalueofthisvariableorigi- andintra-annualvariability,climatechangeimpactandfortheanalysis
nallyformulatedinaccordancewiththefollowingrelationshipbetween ofthetemporalevolutionofprocesseswithhigherinertia,suchasni-
thereflectancedetectedbythesensorinthespectralbandsofredand tratepollutioningroundwaterbodies.Themodelhasaresolutionof
near-infrared(Eq.(1)): 1000×1000mandhasbeenappliedthroughoutSpain;itsresultsare
usedbydifferentSpanishRiverBasinAuthorities.ThePatricalmodel
ρNIR−ρred hasdevelopedanextensionforthecreationoffullydistributedmaps
NDVI¼ ð1Þ
ρNIRþρred (1km×1km)ofSPI,PDSIandiHI,whichcanalsobeprocessedwith
theArc-GISprogrammeuntilthedesireddisplayisobtained.
where:ρNIRisthereflectancedetectedinthespectralbandofnear-in- MonthlymapsofStandardisedPrecipitationIndexfordifferentaccu-
fraredandρredisthereflectancedetectedinthespectralbandofred. mulationperiods(SPI-1,SPI-3,SPI-6,SPI-12ySPI-24)arecalculated
Table1
Droughtindicesanddataemployed.
StandardisedIndex Typeofdrought Datasource Process
monitoring
SPI(McKeeetal.,1993)Shorttime:SPI,SPI-3,SPI-6Long Meteorological SAIHPrecipitationstations Areal
time:SPI-12,SPI-24 precipitation
PDSI(Palmer,1965) Edaphic SAIHPrecipitationandTemperaturestationsfromandNational Waterbalance
MeteorologicalAgency(Aemet) insoil
iHI(Ortega,2015)iHI,IHI-3 Edaphic SoilmoistureobtainedbyPatricalWaterBalanceModel. Waterbalance
insoil
iNDVI(derivedfromNDVI,Jordan,1969)iNDVI,iNDVI-3, Edaphic Greennessofvegetation(NDVI)recordedbysatelliteimages.MODIS Standardised
iNDVI6 sensor index
iSI(Ortega,2015)(derivedfromdeSI) Operational ReservoirStoragesfromSAIHPiezometriclevels.Groundwater Standardised
monitoringnetwork index
3-monthStream-flowsfromSAIH
12-monthPrecipitationfromSAIH
280 T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290
basedonnormalised(log-normaldistributionfunction)accumulated explained above is applied to obtain the spatial distributed iHI-3,
precipitation(1,3,6,12and24months).Shortaccumulationperiods iNDVI-3andiNDVI-6indices.
determinewhenthemeteorologicaldroughtstartsanditsdistribution TheStandardisedStatusIndex(iSI)wascalculatedbythemonthly
overtheterritory.Long-accumulationperiods,SPI-12orSPI-24,reflects standardisation(Eq.(2))oftheSpanishStatusIndex(SI).TheStatus
thepatternofbehaviourofthelong-termrainfall,andcouldalsoserve Index(SI)isappliedinalltheSpanishRiverBasinsandismonthlyup-
inthedefinitionoftheconceptof“prolongeddrought”(EC,2007)in dated(Fig.3)intheNationalDroughtIndicatorSystem(MMA,2007).
theJúcarRBD.Also,theSpanishMeteorologicalStateAgencycarries Thissystemallowsmakingtheformaldeclarationofadroughtinthe
outthemonthlyfollow-upofmeteorologicaldroughtandpublishes SpanishRiverbasinorganisations.Thisstatementimpliestheentry
theSPIindexmapsforthenationalscope(www.aemet.es). into force of the drought management plans called “Special action
MonthlymapsofthemodifiedPalmerDroughtSeverityIndex,PDSI, plansinsituationsofalertandtemporarydrought”,wherethewaterre-
areobtainedfromthewaterbalanceinsoiloriginatedbythePatrical sourcessystems'exploitationrulesandmeasurestobeappliedare
model,whichisthebestcalibratedandvalidatedmodelfortheJúcar establishedinregardtotheuseofwaterduringthedifferentphasesof
RBD (Pérez-Martín et al., 2014). The water balance proposed by adrought.
Palmer(1965)thatisdevelopedpreviouslyinordertocalculatethe Thisindicatorssystemisformedbyadistributednetworkofcontrol
PDSI is replaced by the more accurate water balance for this river pointsandcollectsinformationaboutreservoirstorages,groundwater
basincalculatedbythePatricalmodel.PDSIinputdataarethemonthly piezometric levels, stream flows, and precipitation (MMA, 2007;
precipitationandtemperaturedata. EstrelaandSancho,2016).EachSpanishriverbasinorganisationhas
Droughtindicesthatincludeprecipitationandtemperaturearecur- chosenfortheSIdefinition,theindicators(variables)thatbestdescribe
rentlyrequiredduetothechangesobservedandpredictedbyclimate thewaterresourcessystems'behaviourineachterritory(Estrelaand
changeintemperature.ThePDSIhasbeenusedintheJúcarRBDtode- Vargas,2012),ensuringageneralhomogeneityatthenationalscale.
terminetheeffectofclimatechangeinthedurationandintensityofthe TheStatusIndexiscalculatedbythefollowingexpressions(Eq.(3)):
currentdroughts(Pérez-Martínetal.,2015)andalsotheStandardised (cid:1) (cid:3)
PrecipitationEvaporationIndex(SPEI;Vicente-Serranoetal.,2010), 1 Vi;t−Vmed;t
Vi;t≥Vmed;t→SIi;t¼ 1þ
whiAchnieswamdrooduigfihctatiinodneoxfbthaeseSdPIotnhatthienvclaurdiaebslteemsopilemratouisretu.reispro- Vi;tbVmed;t→SIi;t¼ 2 Vi;t−VVmmaxi;nt;−tVmed;t ð3Þ
posed,theStandardisedHumidityIndex(iHI).TheiHIhasasinput 2ðVmed;t−V min;tÞ
data monthly precipitation and temperature. The index, for one
monthiHIandforthree-averagemonthsiHI-3,iscalculatedfromthe where:V istheStatusindexfortheyeariandmontht;V istheindi-
i,t i,t
soilmoistureobtainedbythePatricalmodel.Theaimofthisindexis catorvaluefortheyeariandmontht,V V andV arethe
med,t, max,t min,t
toknow,inaneasytounderstandanddirectway,therelativeamount monthly(t=1to12)statistics(mean,maximumandminimum)of
ofwaterinthesoil,whichcouldbeusedbythevegetationinamonth theindicatorvaluesinthereferenceperiod.
orin aseason(threemonths) andexploreitrelationship withthe TheintegratedvaluesofSIforthedifferentRiverBasinsorfortheen-
otherindices.TheiHI-3reflectstheseasonalbehaviourofsoilmoisture, tireRiverBasinDistrictareobtainedastheweightedsumoftheindivid-
whichisassociatedwithmedium-termprecipitation,andthereforeit ualindexesrepresentingtheavailablewaterresources.Theweightof
shouldbelinkedwiththebehaviourofPDSIandiNDVI. eachindividualindexisbasedmainlyontheamountofwaterdemand
Themonthlystandardisation(Salasetal.,1980)isdevelopedfor thatwaterresourcesrepresentedbytheindexissupplyingandthe
monthlysoilmoistureorthree-averagesoilmoisture(Eq.(2)). levelofguaranteerequiredbyeachwaterdemandunit.Theintegrated
indexallowsclassifyingtheriverbasinsinfourtypesofhydrologicsta-
Xv;t‐μt tuses(normality,pre-alert,alertandemergency)thatassesstherestric-
Yv;t¼ ð2Þ
σt tionrisksinthevariouswaterdemandunitsofthesystems.
The SI has therefore an advantage over the meteorological and
where:Y istheresultingstandardisedvariablefortheyearvand edaphicindicesinthedroughtmanagementofawaterresourcessys-
v,t
month t;X is theoriginal variable; μ andσ are respectively the tem,becauseitislinkedtothedemandstobemetandthewaterre-
v,t t t
monthlyt(1to12)meanandstandarddeviationoftheoriginalvariable sources that are used to meet them. In addition, the weighting
forthereferenceperiod. coefficientsandthethresholdsdefiningthetypesofhydrologicalstatus
Maps of Standardised Normalised Difference Vegetation Index, havebeenestimatedtakingintoaccountactualwaterresourcessys-
iNDVI,areobtained by applyingthestandardisation procedurede- tems'deficits.
scribedabove(Eq.(2))totheNDVIvariable(greennessofvegetation) IntheJúcarRBD,theSIsummarisesdroughtsituationsbymeansof
directlyextractedfromthesatelliteimages.TheJúcarRBDirrigated 34indicatorslinkedtowaterresourcesavailabilitytosatisfywaterde-
areasmightaffecttheinterpretationintheevolutionoftheNDVIvari- mands(Fig.4a).Arealprecipitationgaugingstations,aquiferwater
able.Theseareasinvolveanalterationofthenaturalhydrologicalcycle levelmonitoringnetwork,stream-flowsandreservoirinflowsrecorded
regimeduetotheartificialsupplyofwaterforirrigation,whichaffects bygaugingstationsandreservoirstoragesobtainedfromSAIH.Those
thevegetatedlandarea.Theproceduredevelopedforcalculatingthe indicatorsareallweightedatthewaterresourcessystemscale,depend-
zonal(waterresourcessystem)monthlyNDVIvaluesincludetwover- ingonthedemandstheymeet,uptoalimitlowerthan10hm3/year.
sions,withandwithouttheirrigationlandexistingintheJúcarRBD. TheresultsoftheirmonthlyapplicationintheJúcarRBDarepublished
Thus,thezonaliNDVIhasasecondversionwithouttheirrigationland inthedroughtindicatorsfollow-upreportoftheJúcarRiverBasinAu-
(iNDVI ).MonthlystatisticvaluesoftheNDVIvariable(range,average, thority(Fig.4b),whichisaccessiblethroughthewebpageofJúcar
w
minimum,maximumandstandarddeviation)werethencalculatedper RiverBasinAuthority(www.chj.es).
waterresourcessystem.
TheiHI-3,iNDVI-3andiNDVI-6indicesrequirethreemonthsaccu- 3.Results
mulatedvalues,soitisnecessarytouseaPythonscriptcreatedfor
this purpose to automate the task in ArcGIS. Basically, the script ThestudycaseshowsapracticalapplicationofaDroughtMonitoring
consistedofamovingwindowof3itemswhichrunstheentireseries Systemimplementedinarealriverbasindistrict,theJúcarRiverBasin
oftheinputrastermaps(soilmoistureandNDVI),creatinganewout- District.DroughtindicesdefinedinTable1areappliedtotheentire
putrasterthatincludesthethree-monthaccumulatedoriginalvariable. riverbasin.Droughtshavethree-dimensionalcomponents,spatialdis-
Oncetheaccumulatedvariableiscreated,thestandardisationprocess tribution(twodimensional)andtemporalevolution(thirddimension).
T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290 281
Fig.3.NationaldroughtindicatorsystemAprilandMay2017(Source:www.mapama.es)andStatusindexdefinitionandriskclassification(MMA,2007).
Twoanalysesaredeveloped:MappingDroughtIndicesalongtimeand meteorologicalconditionsoftheJúcarRiverBasinDistrict(Table2).In
aggregated TimeSeries DroughtIndices for adetermined area,the thecaseofSPI,iHIandiHI-3indicestheoriginalvaluerangewasmod-
waterresourcessystem. ifiedfromseventoninelevelsandthenormaltypewasdividedinto
Mappingdroughtindicesallowsknowingthespatialdistribution threecategories.InthecaseofPDSIandiNDVI-3indicesaclassification
andthespatialdroughtevolution,andcanhelptobetterunderstand isdividedintoelevenlevels,whichseparatesthenormaltypeintofive
thedroughtevolutionanditsconsequencesalongtime. categories.
TimeSeriesofDroughtIndicesforaspecificarea,suchasthewater Recently,ameteorologicalandedaphicdroughtoccurredinthemid-
resourcessystem,areusefultoexplorethedependencyandrelation- dleandlowerbasinsoftheJúcarRBDduringtheyear2014(theanom-
shipsbetweendifferentindices.Fromtheseresultstherelationships alyendedinDecemberof2014).Themainsystemsaffectedfromnorth
thatmaybeconsideredstatisticallysignificant,havebeenestablished. tosouthwere:middleandlowerMijares,TuriaandJúcarandtheentire
Vinalopó system. Howevertheupperriverbasinsof thesesystems
3.1.MappingGIS-basedindices (Mijares,TuriaandJúcar)wereunderwetconditions,sothewatergen-
erationintheseareaswasnormalorabovenormalinthatyear.
Differentcolourclassificationsareadoptedtobettercomparethe Thespatialdistributionofthedroughthasdeterminantimplications
drought indicators. A more detailed classification to the usually onwatermanagement.Duringthatyear,2014,thewaterresourcesgen-
employedbytheUniversityofNebraska-Lincolnisadoptedtorepresent erationintheupperriverbasinswasnotaffected.Theseareasprovide
themapsinordertoadaptthemappingresultstotheregionalhydro- thenaturalinflowstothemajorreservoirsoftheRiverBasinDistrict.
Fig.4.Locationofthe34indicatorsselectedintheJúcarRiverBasinDistrict.IndividualSIMayof2017.
282 T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290
Table2
Droughtintensityclassification,valuesandcolourrange,forSPIandiHI(leftside)andPDSIandiNDVI(rightside).
SPI and iHI Classification Colour PDSI and iNDVI Classification Colour
−4.00 and under Extremely dry
−2.00 and under Extremely dry −3.99 a −3.00 Very dry
−1.99 a −1.50 Very dry −2,99 a −2.00 Moderately dry
−1.49 a −1.00 Moderately dry −1,99 a −1.00 Slightly dry
−0.99 a −0.50 Slightly dry −0.99 a −0.50 Incipient dry
−0.49 a +0.49 Near normal −0.49 a +0.49 Near normal
+0.50 a +0.99 Slightly wet +0.50 a +0.99 Incipient wet
+1.00 a +1.49 Unusually wet +1.00 a +1.99 Slightly wet
+1.50 a +1.99 Very wet +2.00 a +2.99 Unusually wet
+2.00 and above Extremely wet +3.00 a +3.99 Very wet
+4.00 and above Extremely wet
Fig.5.Mapsofthedroughtindicesinthedryseasonof2014.
T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290 283
Waterdemandsforcropswereincreasedthatyear,duetotheminor TheoperationaldroughtindexiSIandthelong-termmeteorological
precipitationinthecoastandthehighertemperature,andwaterderiva- drought,SPI-24,haveasimilarperformanceduringthisperiod(Figs.6b
tionsforirrigationareincreasedaround10%inrespectofpreviousyears and7b)inbothsystems,TuriaandJúcar.Themainreasonisthatapre-
(CHJ,2016).Themajorwaterresourcessystems(Mijares,Turia,Júcar) cipitationdeficitprolongedintimeistransmittedintothehydrological
werenotsignificantlyaffectedbythismeteorological-edaphicdrought. cycleandfinallyimpliesareductionofwaterreservesandtheavailable
However,rain-fedirrigation systems(Magroriver,MiddleJúcaror resourcesforthefinalusers.However,therearesomedeviationsinthe
Vinalopósystem)wereclearlyimpactedbythedrought. Júcarsystem,thelargestsystemoftheJúcarRBD.SPI-24couldbeagood
That year saw the paradox that aggregated drought indicators indextodefinetheprolongeddroughtbecauseitonlytakesintoaccount
showedthattheyearwasnormalorbetterthannormal,whileinlarge precipitationdataasrequiredbytheEuropeanCommission(EC,2007)
areasoftheRiverBasinDistrictsignificantimpacts duetodrought andadditionallyitreflectsthebehaviourofthewaterresourcesystems.
wereoccurring(Fig.5). Inthelastnormal-drycycle,fromthebeginningof2012untilthe
endof2014,manyindices-inbothsystems-showaninitialdrycondi-
tion(iHIandSPI;SPI-24andiSI;andPDSI,iHI-3)thatfinallycomestobe
3.2.Timeseriesofdroughtindices inaverydrycondition(SPI-3attheendof2014andthePDSIduring
2014).WhiletheoperationaldroughtindexiSIoftheTuriasystem
TemporalevolutionofthedroughtindicesforTuriaandJucarsys- alsoreflectsthissituation,theiSIoftheJúcarsystemdoesnotreflect
tems(thelargestsystemsoftheJúcarRBD),includingtheclassification it.TounderstandthesedeviationsintheJúcarsystemitisnecessaryto
of“verydry”and“extremelydry”conditionsaccordingtoTable2,have usethedistributedmapsofthedroughtindicators(Fig.5).Inthecase
beenobtainedfromOctober1991untilDecember2014(Figs.6and7). oftheTuriasystem,themaps(SPI-24,SPI-12andPDSI)showthatthe
InthecaseofiNDVI,dataseriesbeganin2002. droughtextensioncoversthemediumandlowerbasinandalsoapart
Short-termmeteorologicaldrought,SPI-3,isverywellcorrelated oftheupperbasin,whichisalsoobservedintheiNDVI-3,sointhissys-
withthesoilmoisture,iHI.Bothindiceshaveaverysimilarbehaviour temthereisnotasignificantdeviationbetweenindices.Howeverinthe
alongthe23yearsstudiedfortheTuria(Fig.6b)andJucar(Fig.7a)sys- caseoftheJúcarsystem,themaps(Fig.5)showthatalthoughdrought
tems.Sothe3-monthaccumulatedprecipitationisasimpleandavery conditionscoveralargeextensiontheydonotcoverthewholeofthe
goodindicatorofthesoilmoistureconditions. riverbasin,especiallytheupperbasin.Ifitisalsotakenintoaccount
Fig.6.TimeseriesofdroughtindicesfortheTuriasystem.a)RelationshipbetweensoilmoistureindexiHIandshort-termmeteorologicalindexSPI-3.b)Relationshipbetweenoperational
indexiSIandlong-termmeteorologicalindexSPI-24.c)RelationshipbetweenedaphicindicesPDSIandiHI-3andvegetationindexiNDVI-6.
284 T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290
Fig.7.TimeseriesofdroughtindicesfortheJúcarsystem.a)RelationshipbetweensoilmoistureindexiHIandshort-termmeteorologicalindexSPI-3.b)Relationshipbetweenoperational
indexiSIandlong-termmeteorologicalindexSPI-24.c)RelationshipbetweenedaphicindicesPDSIandiHI-3andvegetationindexiNDVI-6.
thattheupperbasinistheareawhereasignificantpartofthewaterre- Forthesecases,theindicesSPI-3,iHIandiNDVIarethefirsttodetect
sourcesoftheriverbasinaregenerated,thenitispossiblethatanaggre- theonsetofadryperiod(October2004,June2009,August2011),with
gatedindexfortheentireriverbasinreflectsagoodsituationforthe anestimatedanticipationof5to8monthscomparedtotheSPI-12and
totalresources,whileitispossiblethatthereisanintensivedrought PDSI,(October2004,August2011),andof10to18monthswhencom-
inspecificandsignificantareasinsidethebasin.Thereversecanalso paredtotheSPI-24(October2004,August2011).Thesesoilorshort-
happen. termmeteorologicalindiceshavedemonstratedtheiranticipationin
The vegetation index iNDVI-6 has a similar performance to the droughtdetection,respectivelytootherindicesstudiedforlong-term
edaphicdroughtindicatorPDSI,inbothsystemsanalysed,Júcar(Fig. meteorological(SPI-12,SPI-24),edaphic(PDSI)oroperational(iSI),so
6c)andTuria(Fig.7c).ThisbehaviourisslightlyworsewiththeiHI-3 theyimprovethedroughtriskdetection.
index.
MonthlyevolutionofthedifferentdroughtindicesfortheJúcarsys- 4.Discussion
temispresentedthroughamatrix(Table3).Thematrixincludesfor
each index, iSI, SPI-24 SPI-12 SPI-6, SPI-3, PDSI, iHI, iHI-3, iNDVI, Correlationsbetweenindices,iSI,SPI-24SPI-12SPI-6,SPI-3,PDSI,
iNDVI-3andiNDVI-6,thenumericalvalueandtheclassinwhichitisin- iHI,iHI-3,iNDVI,iNDVI-3andiNDVI-6,areanalysedforthetwolarger
cludedbasedonTable2.Thistableisaneasyunderstandingtoolto waterresourcessystems,TuriaandJúcar,oftheJúcarRBD,(Table4).
knowthebeginningandtheendingofthedroughts,theirintensity Inbothsystems,thesamestructureaboutwhatarethebettercorrela-
andscope:meteorological,edaphicoroperational. tions is produced (see bold numbers in table 4). The operational
IntheJúcarsystemthereisanextremedryperiodduringthewater droughtindexiSIisbettercorrelatedwithSPI-24andthePDSIwith
years1993/94–1994/95andtwoverydryperiods,oneoftwoyears SPI-12.Inthecaseofmoistureindices,iHIitismoresimilartothe
1998/99–1999/2000andanotherofthreeyears2004/05–2006/07.The short-termmeteorologicalindexSPI-3andtheniHI-6isbettercorrelat-
greater manifested drought occurred in the early nineties (1994– edwithSPI-6.Forthevegetationindices,iNDVI,iNDVI-3iNDVI-6are
1995)whenallstandardisedindicesfordifferenttimeoffsetsreveal bettercorrelatedwithSPI-12andPDSI,inbothcaseswithsimilarvalues.
theexistenceof“extremelydry”meteorological,edaphicandoperation- Inbothsystemstheoperationaldroughtindex,iSI,isbettercorrelat-
alconditionspersistentoveraround24months.Othershorterdrype- edwiththelong-termmeteorologicalindexSPI-24,0.81fortheTuria
riodsaredetectedintheyears2011/12and2013/14. systemand0.62fortheJúcarsystem.ThecorrelationintheJúcarsystem
T.Ortega-Gómezetal./ScienceoftheTotalEnvironment610–611(2018)276–290 285
Table3
EvMolounttihonofiSIdroughSPI-24tindicaSPI-12torsinSPI-6theJúcSPI-3arsystePDSIm. iIH iIH-3 iNDVI iNDVI-3 iNDVI-6 AMuogn-9th7 0iSI.81 1SPI-24.00 1SPI-12.53 1SPI-6.27 1SPI-3.64 0PDSI.12 2iIH.11 2iIH-3.18 iNDVI iNDVI-3 iNDVI-6
Sep-97 0.59 1.43 1.68 2.30 1.86 1.12 2.35 2.99
Oct-93 −1.54 −1.23 −0.40 0.11 0.13 −1.29 0.66 0.35
Oct-97 0.71 1.47 1.75 1.85 1.12 0.06 −0.54 1.28
Nov-93 −1.28 −1.10 0.02 0.50 0.37 −0.83 0.88 0.77
Nov-97 1.06 1.67 1.78 1.77 0.81 0.70 0.32 0.49
Dec-93 −1.61 −1.31 −0.40 −0.22 0.00 −1.61 −0.31 0.50
Dec-97 1.38 1.66 1.90 1.89 0.78 1.63 1.18 0.48
Jan-94 −0.93 −1.35 −0.29 −0.56 −0.65 −1.71 −0.41 0.01
Jan-98 1.34 1.65 1.46 1.46 1.17 1.59 0.91 0.94
Feb-94 −1.24 −1.46 −0.91 −0.69 −1.37 −1.89 −0.76 −0.54
Feb-98 1.22 1.53 1.65 1.19 0.85 1.37 0.60 0.92
Mar-94 −0.48 −1.50 −1.27 −1.09 −1.94 −3.11 −1.18 −0.81
Mar-98 1.11 1.40 1.63 0.28 −0.22 0.44 −0.21 0.50
Apr-94 −0.77 −1.21 −0.88 −1.10 −1.07 −2.66 −0.76 −0.94
Apr-98 0.98 1.43 1.30 0.50 −0.65 0.33 0.05 0.18
May-94 −0.87 −1.37 −0.82 −1.60 −1.12 −3.32 −1.07 −1.08
May-98 0.99 1.51 1.34 0.69 0.44 1.88 1.41 0.27
Jun-94 −0.87 −1.72 −1.25 −1.90 −1.10 −3.34 −1.67 −1.19
Jun-98 1.33 1.48 1.07 0.23 0.61 1.56 0.19 0.60
Jul-94 −0.83 −1.69 −1.47 −2.07 −2.09 −2.61 −1.61 −1.55
Jul-98 0.94 1.40 0.89 0.01 0.73 1.62 −0.60 0.81
Aug-94 −0.52 −1.77 −1.46 −2.08 −2.12 −2.19 −1.06 −1.67
Aug-98 0.81 1.41 0.63 −0.06 −0.89 1.53 −0.35 −0.20
Sep-94 −0.55 −1.35 −1.23 −1.14 −0.15 −1.73 0.05 −0.99
Sep-98 0.63 1.20 0.21 0.11 −0.61 1.15 −0.49 −0.63
Oct-94 −0.83 −1.19 −1.20 −0.71 0.56 −0.91 0.58 0.25
Oct-98 1.41 1.24 0.16 −0.43 −1.54 0.08 −1.16 −1.30
Nov-94 −0.50 −1.05 −1.54 −0.32 0.47 −1.12 −0.27 0.16
Nov-98 0.89 0.97 −0.43 −2.56 −1.75 −0.59 −1.56 −1.55
Dec-94 −0.77 −1.44 −1.75 −0.65 −0.55 −2.05 −1.58 −0.63
Dec-98 0.45 0.87 −0.78 −1.92 −1.88 −0.37 −1.07 −1.60
Jan-95 −0.47 −1.54 −2.03 −0.70 −1.56 −2.39 −1.36 −1.27
Jan-99 −1.59 0.30 −1.18 −1.59 −0.60 −0.70 −1.00 −1.39
Feb-95 −0.60 −1.93 −2.01 −0.81 −2.00 −3.05 −1.59 −1.59
Feb-99 −0.31 0.42 −1.29 −1.72 −0.10 −0.76 −1.11 −1.13
Mar-95 −1.46 −2.04 −1.69 −1.48 −1.87 −3.72 −1.32 −1.50
Mar-99 0.01 0.74 −0.53 −0.88 0.22 0.08 −0.40 −0.88
Apr-95 −1.64 −2.12 −2.28 −2.52 −2.59 −4.40 −1.52 −1.53
Apr-99 −0.41 0.50 −0.60 −0.28 0.23 −0.50 −0.68 −0.76
May-95 −1.70 −2.10 −1.89 −2.58 −2.07 −4.97 −1.38 −1.49
May-99 −0.38 0.19 −1.33 −0.21 −0.07 −1.54 −1.03 −0.67
Jun-95 −1.67 −1.92 −1.37 −1.63 −0.81 −4.29 0.05 −1.35
Jun-99 −0.74 −0.04 −1.27 −0.51 −1.03 −1.50 −0.79 −0.94
Jul-95 −1.56 −1.99 −1.30 −1.48 0.10 −3.81 −0.13 −0.85
Jul-99 −0.61 −0.05 −0.99 −0.10 −0.27 −1.45 0.50 −0.81
Aug-95 0.41 −1.68 −0.69 −0.33 1.31 −3.85 1.83 0.69
Aug-99 −0.57 −0.31 −1.00 −0.11 0.22 −1.18 −0.45 −0.43
Sep-95 0.49 −1.76 −1.21 −0.67 0.18 −3.95 −0.90 0.31
Sep-99 −0.82 −0.44 −0.70 −0.35 0.71 −0.79 0.72 0.35
Oct-95 0.40 −2.23 −1.93 −0.58 −0.65 −4.62 −1.08 −0.65
Oct-99 0.18 −0.13 −0.13 0.24 0.69 −0.35 0.61 0.77
Nov-95 −1.71 −2.53 −2.06 −0.26 −1.64 −4.61 −1.17 −1.36
Nov-99 0.07 −0.47 −0.09 0.34 0.41 −0.54 0.01 0.49
Dec-95 0.31 −1.84 −0.96 0.27 0.39 −3.02 0.26 −0.80
Dec-99 −0.35 −0.86 −0.38 0.11 −0.42 −0.83 −0.42 0.03
Jan-96 1.26 −1.62 −0.39 0.71 1.29 −2.04 0.59 −0.04
Jan-00 −0.56 −1.16 −0.43 −0.31 −1.00 −0.96 −0.31 −0.28
Feb-96 1.18 −1.31 −0.08 0.34 1.44 −1.56 0.95 0.67
Feb-00 −0.35 −1.33 −0.66 −0.66 −1.42 −1.93 −1.33 −0.74
Mar-96 1.03 −1.08 0.06 0.63 0.68 −1.68 0.47 0.74
Mar-00 −0.59 −1.17 −1.07 −1.28 −1.64 −2.44 −1.06 −0.93
Apr-96 0.59 −1.24 0.21 0.69 −0.40 −2.03 −0.06 0.55
Apr-00 −0.37 −0.96 −0.69 −0.95 −0.45 −1.75 −0.34 −1.01
May-96 0.53 −0.93 0.47 0.81 −0.21 −1.61 0.17 0.26
May-00 −0.09 −1.29 −0.27 −0.49 0.31 −1.74 −0.14 −0.66
Jun-96 0.54 −0.81 0.22 0.23 −0.27 −1.59 −0.38 −0.06
Jun-00 0.10 −1.26 −0.43 −0.69 0.21 −1.83 −0.82 −0.43
Jul-96 0.68 −0.74 0.24 −0.27 0.18 −1.41 −0.19 −0.07
Jul-00 −0.31 −1.23 −0.65 −0.70 −0.41 −1.34 −1.19 −0.60
Aug-96 0.54 −0.70 −0.07 −0.42 −0.14 −1.19 −0.16 −0.30
Aug-00 −0.70 −1.33 −0.63 −0.36 −1.62 −1.04 −1.00 −1.11
Sep-96 0.72 −0.40 0.61 0.51 1.12 −0.04 1.70 0.80
Sep-00 −1.05 −1.43 −1.26 −0.96 −2.58 −1.58 −1.70 −1.79
Oct-96 0.62 −0.80 0.56 0.14 0.21 −1.05 −0.81 −0.03
Oct-00 −1.05 −0.81 −0.89 −0.33 0.15 −0.28 0.68 −0.36
Nov-96 0.59 −0.59 0.85 0.43 0.66 −0.29 0.35 0.25
Nov-00 −1.06 −0.71 −0.82 −0.51 0.26 −0.23 0.36 0.23
Dec-96 1.24 −0.13 0.77 1.17 0.70 0.96 1.63 0.60
Dec-00 −0.63 −0.70 −0.53 0.07 1.29 0.24 0.61 0.67
Jan-97 1.46 0.53 1.24 1.78 1.96 3.03 2.12 1.69
Jan-01 −0.19 −0.56 −0.24 0.43 0.62 0.65 0.55 0.59
Feb-97 1.50 0.45 0.90 1.57 1.39 1.89 0.71 1.58
Feb-01 0.17 −0.44 0.11 0.59 0.71 0.51 0.53 0.59
Mar-97 1.33 0.37 0.72 0.73 0.57 0.59 −0.58 0.86
Mar-01 0.43 −0.60 0.26 1.04 0.50 0.04 0.17 0.47
Apr-97 0.94 0.73 1.00 1.21 −0.72 0.66 −0.17 0.02
Apr-01 0.63 −0.58 −0.06 0.29 −0.10 −0.55 −0.25 0.22
May-97 0.78 0.88 0.95 1.03 0.36 1.03 0.27 −0.27
May-01 0.53 −0.38 −0.02 0.38 0.07 −0.35 −0.05 0.00
Jun-97 0.80 0.95 1.32 1.23 1.42 1.41 1.57 0.35
Jun-01 0.10 −0.48 −0.07 −0.07 −0.58 −0.65 −0.99 −0.38
Jul-97 0.99 1.02 1.39 0.61 1.41 0.59 2.19 1.14
Jul-01 −0.18 −0.60 −0.06 −0.50 −0.51 −0.32 −1.19 −0.61
Aug-01 −0.21 −0.58 0.03 −0.55 −1.24 −0.21 −0.53 −1.02