Recent changes in climate, hydrology and sediment load in the Wadi Abd, Algeria

18 Pages·2016·3.08 MB·English

Hydrol.EarthSyst.Sci.,20,1355–1372,2016
www.hydrol-earth-syst-sci.net/20/1355/2016/
doi:10.5194/hess-20-1355-2016
©Author(s)2016.CCAttribution3.0License.
Recent changes in climate, hydrology and sediment load
in the Wadi Abd, Algeria (1970–2010)
MohammedAchite1andSylvainOuillon2,3
1LaboratoireEau-Environnement,UniversitéHassibaBenBouali,BP151,HayEs-Salem,Chlef,Algeria
2LEGOS,UniversitédeToulouse,IRD,CNRS,CNES,UPS,14avenueE.Belin,31400Toulouse,France
3UniversityofScienceandTechnologyofHanoi,Dept.Water-Environment-Oceanography,18HoangQuocViet,
CauGiay,Hanoi,Vietnam
Correspondenceto:SylvainOuillon([email protected])
Received:20September2015–PublishedinHydrol.EarthSyst.Sci.Discuss.:14October2015
Revised:7March2016–Accepted:19March2016–Published:6April2016
Abstract. Here we investigate the changes of temperature, centchangesshowedatrendtowardsincreasinglanderosion
precipitation,riverrunoffandsedimenttransportintheWadi and decreasing fluxes to coastal waters (Walling and Fang,
Abd in northwest Algeria over a time series of 40 hydro- 2003;Vörösmartyetal.,2003;Wangetal.,2006).Thesedi-
logical years (1970–2010). Temperature increased and pre- mentfluxtrappedinregulatedbasinswithreservoirsishigher
cipitationdecreasedwiththereductioninrainfallbeingrela- than 50% (Vörösmarty et al., 2003). Locally, it can reach
tivelyhigherduringtherainyseason.Ashifttowardsanear- more than 60% after the impoundment of one single dam,
lier onset of first rains during summer was also found with likeontheRedRiver(Vinhetal.,2014),andmorethan80%
cascadingeffectsonhydrology(hydrologicalregimes,vege- onriverswithmanydams(86%ontheYellowRiver,Wang
tation, etc.)and thus onerosion and sedimentyield. During etal.,2007; >95%ontheEbroRiver,Durandetal.,2002).
the 1980s, the flow regime shifted from perennial to inter- Other engineering activities (meander cutoffs, river-training
mittent with an amplification of the variations of discharge structures, bank revetments, soil erosion controls) also sig-
and a modification of the sediment regime with higher and nificantly affect sediment fluxes and can participate in the
moreirregularsuspendedparticulateflux.Sedimentfluxwas shiftfromatransport-limitedsystemtoasupply-limitedsys-
showntoalmostdoubleeverydecadefromthe1970stothe tem, like on the Missouri–Mississipi River system (Meade
2000s.Thesedimentregimeshiftedfromtwoequivalentsea- andMoody,2010).
sons of sediment yield (spring and fall) to a single major Climate change, through increasing temperatures and
seasonregime.Inthe2000s,autumnproducedover4times evaporation, tends to accelerate the water cycle and mod-
more sediment than spring. The enhanced scatter of the C– ifyhydrologicregimes(Batesetal.,2008).Precipitationin-
Q pairs denotes an increase of hysteresis phenomena in the tensities and the frequency of extreme events are projected
WadiAbdthatisprobablyrelatedtothechangeinthehydro- to increase under climate change, leading to more frequent
logicregime.Attheendoftheperiod,duetoirregularityof floodeventsofhighermagnitudethatwill,inturn,affectpat-
thedischarge,theabilityofaratingcurvetoderivesuspended terns of erosion and deposition within river basins (Tucker
sedimentconcentrationfromriverdischargewaspoor. and Slingerland, 1997; Pruski and Nearing, 2002; Tockner
and Stanford, 2002; Coulthard et al., 2012). Recent stud-
ies focused on the impact of climate change on sediment
transport(e.g.Gomezetal.,2009;Hancock,2009;Walling,
1 Introduction 2009; Hancock and Coulthard, 2011; Knight and Harrison,
2013;Luetal.,2013).Syvitski(2003)showedonanexam-
Fluvial and estuarine suspended sediment fluxes are chang- plethatsedimenttransportmayincreaseduetotheincreas-
ing dramatically under the combined effects of anthro- ingdischargeordecreasebecauseoftheenhancedtempera-
pogenicactivitiesandclimatechange.Onaglobalscale,re-
PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion.
1356 M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd
ture.Studieshavecomparedtrendsinhydrologicalandsed- shipbetweensedimentloadandrunoffoverthelast40years;
iment time series to land use changes (Wang et al., 2007; (4)detectwhenashiftoccurredintherunoff–sedimentload
Memariam et al., 2012; Gao et al., 2012). Climate projec- relationship; (5) analyse the possible causes of the change
tions are consistent regarding warming and the acceleration inflowregimeanditsconsequencesonsuspendedsediment
ofthewatercycle(IPCC,2013),however,theyremaintobe discharge; and (6) assess the use of rating curves and the
definedonsedimenttransportwhereprojectionsshowahigh physicalsignificationoftheirparameterswhenariverisex-
uncertainty (Shrestha et al., 2013; Lu et al., 2013). This is periencingatransitionandshiftsfromaperennialregimeto
inpartduetothefactthatclimateaffectsmanyfactorscon- anintermittentregime.
trolling sediment yield, such as surface moisture availabil-
ity,weatheringprocessesandrates,andthenatureofriparian
2 Studyarea:theWadiAbd
vegetation(Nansonetal.,2002).
While sediment transport is well documented in peren-
2.1 Generalinformation
nialriversinhumidortemperateclimates,itsstudyinsemi-
arid areas is still fragmentary due to the difficulty of sam-
TheWadiAbd,locatedinnorthwestAlgeria,isatributaryof
plingduringflashfloods.Amongstthefactorsfavouringero-
theWadiCheliff,theprincipalriverofAlgeria(Fig.1).The
sion (slope, nature of rocks, relief, climate, human activi-
length of the Wadi Abd’s main stream is 118km, the basin
ties),climateisrecognizedtobethemainfactorinsemi-arid area is 2480km2 and the drainage density is 3.70kmkm−2
Mediterranean areas of Algeria which experience short and
(Fig.2a).TheWadiAbdsuppliesthedownstreamSidiMo-
intense rain episodes, high evaporating power of wind, pro-
hamedBenaouda(SMB)reservoirwhichhasabasinareaof
longeddroughtsandfreezingandthawingcycles(Touaibia, 4900km2.TheWadiAbdcatchmentareaisformedoferodi-
2010;Houyouetal.,2014).Erosionisextremelyactiveand
ble sedimentary rocks from the Upper Jurassic (45.9% of
the average concentration is at least 1 order of magnitude
its surface), Middle Jurassic (20.2%) and Pliocene (7.4%)
higher than the global average (Achite and Ouillon, 2007).
(Fig.2b).SoftbottomsedimentarydepositsfromtheQuater-
One of the main impacts of this high erosion is the rapid
narycover13%ofthebasinalongthewadi(TecsultInterna-
silting up of reservoirs (up to 2–5% per year, Kassoul et
tional,2004).
al., 1997; Remini et al., 2009; Touaibia, 2010) with impor-
TheclimateisMediterraneanandischaracterizedbyadry
tantconsequencesonwaterresourcemanagementinaregion
season from April to August/September, and a wet season
where 85% of rain evaporates (Benhamiche et al., 2014).
fromSeptembertoMarch.Thehydraulicdeficitisveryhigh.
Thehightemporalvariabilityandrecentchangesinforcings
Annualprecipitationis264mm,onaverage,whilethemean
meanthatitisnecessarytostudysedimentdynamicsinsuch
potentialevapotranspirationovertheSMBbasinis1525mm
environments over time periods of several decades in order
(TecsultInternational,2004).
todocumentandunderstandthechangesinsedimentregime.
The watershed mainly consists of steep slopes (Fig. 2c)
Achite and Ouillon (2007 hereafter referred as AO2007)
withverysparsevegetationorbaresoil(Fig.2d).Themain
analysedsedimenttransportchangesintheWadiAbd,anAl- landuseisnaturalenvironment(73%;17%offorests+56%
gerianwadiovera22-yearperiod(1973–1995).Hereweex-
ofscrubandbaresteppesoils),cultivatedlandscoverabout
tend this analysis of sediment transport changes to cover a
26%andcities0.4%.Sevenhillreservoirswerebuiltinthe
40-yearperiod(1970–2010).Thehydrologicgaugingstation
WadiAbdbasinfrom1986to2004foragriculture(irrigation
islocatedupstreamfromadamandisnotaffectedbyanyma-
and livestock watering) or for fire-fighting measures. Their
jormanagement.Thisriversub-basinisalsoparticularlysuit- total cumulated capacity is 0.88hm3, representing 2.3% of
ableforsuchstudybecauseitshydrologicregimewasshown
theyearlyaverageddischargeatAinHamarastation.These
tohavedrasticallychangedbetweenthe1970sandthe1980s.
smallreservoirsarenowsiltedupto70%oftheirvolume.
Precipitation decreased and became more irregular and the
In2008,123000inhabitantswerelivingintheWadiAbd
flowregimeshiftedfromperennialtointermittentwith26% basin (average density: 49inhabitantskm−2), 44% of them
drydays,onaverage,in1990–1995.Variationsofdischarge
living in the city of Takhmaret. The Wadi Abd is thus little
were amplified, and a modified sediment regime occurred
influenced by human activities, in view of its extensive sur-
withahigherandmoreirregularsuspendedparticulateflux,
facethatissubjecttoseverenaturalerosion.
that was 4.7 times higher over the period 1985–1995 than
In the plain, sheet (interrill) and rill erosion dominates
over 1973–1985. AO2007, showing the advantage of work-
(Fig. 3b and f). Gully erosion is mainly restricted to the
ingwithover22yearsofmeasurement,however,stressedthe
mountainous regions of Frenda and Tiaret in the north
difficultyofdefiningareferenceperiod,andtheneedtoex-
(Figs.3c,dand2c),whilesomemid-slopeareasaregullying
tend the study over a longer period of time. The objectives
(Fig.3aande).
ofthisadditionalstudyareto(1)describeprecipitation,dis-
charge and sediment flux variability of the Wadi Abd basin
overa40-yearperiod;(2)detecttheshift,ifany,intemper-
ature,runoffandsedimentyield;(3)determinetherelation-
Hydrol.EarthSyst.Sci.,20,1355–1372,2016 www.hydrol-earth-syst-sci.net/20/1355/2016/
M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd 1357
Figure1.LocationoftheWadiAbdsub-basinwithintheMinaandCheliffbasins,andtheothermainbasinsofAlgeria.
Figure2.TheWadiAbdcatchmentarea.(a)displaysrainandhydrometricstationsincludingHS1atTakhmaretandHS2atAinHamara,
(b)geology,(c)slopesfromthedigitalelevationmodelofnorthAlgeria,(d)vegetationcoverfromLandsatETM+dataof2009.
2.2 Data 2012;OsbornandJones,2014).Thesestationsarelocatedat
Chlef (36.20◦N, 1.30◦E – 1951–2011), Miliana (36.30◦N,
2.20◦E – 1922–2011) and Dar El Beida (36.70◦N, 3.30◦E
Long-term series of temperature measured at three stations
–1856–2011).Annualaveragetemperatureswerecalculated
in Algeria were extracted from CRUTEM4 (Jones et al.,
www.hydrol-earth-syst-sci.net/20/1355/2016/ Hydrol.EarthSyst.Sci.,20,1355–1372,2016
1358 M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd
Figure3.LinearerosionformsintheWadiAbdbasin.(a)and(e)displaygullying(depth:30–50cm,width>1m),(c)and(d)gullyerosion
(depth:50–200cm),(b)and(f)interillandrillerosion.
Rainfall and hydrometric records were provided by the
22
National Agency of Hydraulic Resources (ANRH). Time
series of rainfall data are available at six stations within

T Chlef = 0.699 T Miliana + 7.335
the basin (see Fig. 2a): S1 Ain Kermes (altitude: 1162m),
20
r ² = 0.724 S2 Rosfa (960m), S3 Sidi Youcef (1100m), S4 Tiricine
(1070m), S5 Takhmaret (655m) and S6 Ain Hamara
2 n T Chlef = 0.944 T Dar El Beida + 2.545
oitats ta )C°( T 18 r ² = 0.731 p(m2ee8nn8dtemsd)o.fsTewdhieamrteeerwntdeircseochn9ac0re7gn6etrca(ontiianomcnisedle(ynCtQ,ini,nsitnagnLmta−n31es)o−ur1es)cmaonreddaesdsuuresa--t
the Ain Hamara gauging station between September 1970
16 andAugust2010.Waterdepthsweremeasuredcontinuously
T Miliana = 0.911 T Dar El Beida + 0.866 andacalibrationbetweenwaterlevelanddischargewasreg-
r ² = 0.684
ularly performed from velocity profiles. During flow mea-
14 surements,waterwasmanuallysampledonceortwiceusing
14 16 18 20 22 a1Ldipattheedgeofthewadi.Thenumberofsampleswas
T (°C) at station 1 adapted to the flow regime. During baseflow samples were
collected every other day, whereas during floods samples
Figure 4. Relationships between mean annual temperatures at
the three stations of Dar El Beida, Miliana and Chlef (from werecollectedathigherrates(uptooneevery30min).Water
CRUTEM4). sampleswerefilteredonpreweighedWhatmanglassfiberfil-
ters(GFF),ovendriedat105◦Cfor24h,andweighedagain
todeterminetheconcentration.Thismethod,usedbyANRH
foreachstationfromthe12monthlyaverages.The20miss- atallhydrologicstationsinAlgeria,underestimatesthesus-
ing monthly data (out of 480) at Chlef, the nearest station pendedloadascomparedtoitsvalueaveragedoverthecross
fromtheWadiAbd,wereextrapolatedfromthemonthlytem- section under low turbulence (i.e. at low flow) since water
peratures measured at Miliana and Dar El Beida using the is sampling near the surface (Touat, 1989). During floods,
relationships between the monthly average temperatures at whichtransportmostofthesedimentload,turbulenceishigh
ChlefandMiliana,andChlefandDarElBeida(Fig.4).The enoughtohomogenizesuspensionload.Whilethisunderes-
resulting estimates of temperature at Chlef on seasonal and timation may slightly affect the budget, it does not severely
yearlyscalesallowedustoestimatechangesbydecadeover affectthetimevariabilityofsuspendedmatterwhichisanal-
theperiod1970–2010. ysedinthispaper.From9076coincidentinstantaneousdata
Hydrol.EarthSyst.Sci.,20,1355–1372,2016 www.hydrol-earth-syst-sci.net/20/1355/2016/
M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd 1359
measuredduring1213days,averagearithmeticvalueswere
105
calculatedperdaysoastoobtain1213pairsof“meandaily”
(C,Q)values.Theresulting“meandailyQ”differsfromthe 104
(ctoruneti)nduaoiulysidnisstcahnatrQge.obtainedfromtheaveragingof24hof 1-s g) 103
The Atlantic Multidecadal Oscillation (AMO) index is k( sQ 102
an index of North Atlantic temperatures. The monthly un- ylia
d
d 101
smoothed values used in this study were calculated by e
tam
NOAA, Earth System Research Laboratory, Physical Sci- itsE 100
ences Division/ESRL/PSD1 (http://www.esrl.noaa.gov/psd/
data/timeseries/AMO/). 10-1
10-2
10-2 10-1 100 101 102 103 104 105
3 Modelsandmethods Measured daily Qs (kg s-1)
3.1 Trends Figure 5. Comparison between estimates of Qs obtained from Q
andtheglobalratingcurve,andmeasuredQs.
The analysis of trends was conducted following a method
fullydescribedbyStahletal.(2010)andDéryetal.(2005)
for river runoff. The Kendall–Theil robust line furnishes a by the global relationship established from 40 years of data
linearequationfromatimeseriesofnmeasurementssuchas (methodA).TheaverageerrorfordailyQs valueswas51%
using method A and 42.1% using method B. However, the
y=mt+b (1) cumulativefluxofsuspendedmatteroverthe1213daysfor
which daily data are available was overestimated by 3.1%
wheret istime(year),y denotesthehydrologicalparameter
using method A while it was underestimated by 5% using
(precipitation, river discharge and sediment discharge) and
methodB.Acomparisonoftheestimatesbythesetwometh-
misthemagnitudeofthetrendoverthisperiod.miscalcu-
ods showed that method B is not reliable at high discharge
lated as the median of all slopes m of consecutive pairs of
k
duringthelastdecadebecauseofanincreaseinscatteringof
values:
theC,Qpairs.Therelationshipobtainedoverthelastdecade
y −y
mk = tjj−tii, (2) (th2e00301–420d1a0y)slfeodrtwohainchudnadielryesCtimanadtioQn oafreQksnoowf 2n3. I%n coovne-r
where k=[1, n(n−1)/2], i=[1, n−1], j=[2, n]. This trast,theglobalalgorithmfrommethodAledtoanunderes-
timation of the same cumulated Q by only 3.5% over the
slope is often referred to as the Sen slope (Sen, 1968). The s
sameperiod.Therelationshipestablishedover40yearswas
significanceofthistrendatalevelp wascalculatedfollow-
thereforeusedforthisstudy.
ingZiegleretal.(2003).
ItshouldbenotedthatalthoughmethodAprovidessome
3.2 Ratingcurves daily solid discharges from the 1213 daily Q values with
a high error (the average error being 51%), it enabled the
C and Q measurements were used to define rating curves reconstruction of good trends of Q values over more than
s
that estimate C from measured values of Q, according to a 7 orders of magnitude (Fig. 5). The temporal variability of
commonapproach(e.g.Walling,1977a;Asselman,2000;El the coefficients a and b of the rating curves calculated over
Mahi et al., 2012; Tebbi et al., 2012; Louamri et al., 2013). yearsordecadeswillbediscussedinlightofthevariability
ThemostsuitablemodelisapowerlawofthetypeC=aQb oftheforcingsandtheirconsequencesonsedimenttransport
for which the coefficients (a, b) determined empirically ac- inordertobetterunderstandtheirphysicalmeaning.
countfortheeffectivenessoferosionandtransport.Therat-
ingcurveestablishedfromthe1213dailyaveragesofC and 3.3 Averageloads
Q data enabled the estimation of C then Q (Q =C×Q)
s s
Inordertoanalysethetemporalvariabilityofsuspendedsed-
for the whole period 1970–2010 from the measured daily
iment flux, we use the average concentration resulting from
Qvalues.
theratiobetweenthesolidandtheliquidflowrate,denoted
Considering the change in hydrologic regime during the
as SPM∗, which can be defined for any integration period
study period, we wondered if the estimate of C and Q per
s
(day,month,season,year).
subperiods such as decades would be more appropriate. We
therefore applied the four rating curves established for the
4 decades to the time series of daily Q to obtain daily C
and then daily Q . This method (B) enabled us to com-
s
pare the estimated solid discharge with the value provided
www.hydrol-earth-syst-sci.net/20/1355/2016/ Hydrol.EarthSyst.Sci.,20,1355–1372,2016
1360 M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd
22 Theaverageprecipitationoverthesixrainfallgaugingsta-
)C°( T egareva ylraeY 112860 TTT MCDhialilrae nEfal Beida asatittnoaidontinso1swn9as8ist5shhc–ioo2nmw0t1hpa0ea,drbeteahdcseritenoaavwtsehearesainAg2ep7inr3deHemcfiiampcmiittyaabrtr−iaeoi1snn,tgawbteeiiottqwhnu.ecaFeolnnitvos1ei39sot.7e7u0nt%–to1v.f9as8riix5-

14 4.3 Riverdischargeandflowregime
1850 1870 1890 1910 1930 1950 1970 1990 2010
Figure 6. Interannual variations of mean yearly temperature (cal- The annual discharge was 1.18m3s−1, on average, and ex-
culatedfromSeptembertoAugustmonthlytemperatures)atthree hibited a higher interannual variability (CV=44.4%) than
stations in northern Algeria: Dar El Beida, Miliana and Chlef
annualprecipitation(Table1).Yearlyvaluesshowedatrend
(from measurements of CRUTEM4 only, extrapolated values are
towardsanincreaseofriverflow(+11.3Ls−1yr−1,onaver-
notshown).
age,over40years,p<0.01;Fig.7),withdecreasingdecadal
valuesbetweenthe1970sandthe1980s,thenincreasingval-
3.4 Studyofbreaks:double-masscurve uesafterwards,similartoP (Table2).
Detailed analysis of daily river discharge shows that the
Double-masscurveswereusedtodeterminelong-termtrends riverwasperennialinthe1970sandthenbecameintermittent
and changes in the hydrosedimentary regime (Searcy and duringthe1980s(Fig.8).Thedriestyearoccurredin1993–
Hardison,1960;Walling,1977b,2006). 1994with117daysofthefullydryriver.InFig.8,verylow
riverdischarges(around0.01m3s−1)werenotconsideredas
daysofdryriver.
4 Interannualvariationsoftemperature,precipitation,
Fromthe1970sto2000s,whenQaveragedover10years
riverdischargeandflowregime
increasedby25%,thewetdischargeQ (i.e.theyearlyaver-
w
agedischargeofthedaysofrunningriver)increasedbymore
The statistics of hydrological parameters at Ain Hamara
than35%(Table2).Twoindicatorsofintra-annualdischarge
gauging station over the period 1970–2010 are reported in
variabilityareshowninFig.7:Q ,the98thpercentileofan-
Table1. 98
nual flows calculated from daily discharge and the standard
deviation of daily discharge within each year (σ ). Q in-
4.1 Temperature Q 98
creasedbyafactor3.2between1970–1980and2000–2010
Temperature in northern Algeria at the three stations of (Table 2). Q98 is a good indicator of changes in sediment
Chlef, Miliana and Dar El Beida increased from the 1970s transportasitoccursduringthehighestfloodeventsthatoc-
onwards (Fig. 6), with higher values at Chlef than at Dar cureachyear.
El Beida and Miliana. On average, temperature at Chlef in-
creased by 1.17◦C from the 1970s to the 2000s (Table 2).
The increase was 0.87◦C between the 1970s and the 1980s 5 Interannualvariationofsedimentload
whichismorethan4timesthedifferencebetweenthe1980s
5.1 Ratingcurve
and the 1990s (+0.18◦C) and the 1990s and the 2000s
(+0.12◦C).Ashasbeenshownonaglobalscale,thedecade
Theratingcurveobtainedfrom1213pairsofdailyaverages
ofthe2000swasthewarmest(IPCC,2013).
gavethefollowing:
4.2 Precipitation (cid:16) (cid:17)
C=2.270Q0.647 r2=0.431 . (3)
AnnualprecipitationatAinHamarastationwashighlyirreg-
ular,varyingbetween165and506mmyr−1(Table1,Fig.7). Of the variations of C, 43% are explained by those of Q.
Mean annual precipitation (P) was 264mm, with a coef- TheratingcurveobtainedbetweenQandQs showsamuch
ficient of variation (CV) of 27% between 1970–1971 and highercoefficientofdetermination(r2=0.831)butisbiased
2009–2010.TheinterannualvariationsofP (Fig.7)showed sinceQs=C×Q.Nevertheless,bothrelationshipsgivees-
trendstowardsadecreaseofrainfall(−1.86mmyr−1,onav- timates of Qs values from Q with less than 1% difference
erage,over40years,p<0.05).P decreasedby15%(from whichislessthantheuncertaintyofQs.
310 to 264mm) between the 1970s and 2000s (Table 2). A
5.2 Yearlysedimentfluxesandconcentrations
more precise analysis shows that rainfall greatly decreased
from the 1970s to the next decade (from 310 to 231mm,
5.2.1 DecadalvariabilityofQ
−25%), then slightly increased in the 2 following decades s
(seeTable2).
Q increasedfrom180to1130×103tperyearbetweenthe
s
1970sandthe2000s(Table2).Theincreasefromonedecade
Hydrol.EarthSyst.Sci.,20,1355–1372,2016 www.hydrol-earth-syst-sci.net/20/1355/2016/
M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd 1361
Table1.GeneralstatisticsontheyearlyaveragesofhydrologicparametersfromtheWadiAbdatAinHamaragaugingstationovertheperiod
1970–2010(note:T atChlefwasestimatedfrommeasurementsatDarElBeidaandMilianafor20monthsover480months).
Statistic T (Chlef) P Q Qw M SPM∗
◦C mmyr−1 m3s−1 m3s−1 103tyr−1 gL−1
Mean 19.09 264 1.18 1.29 564 12.3
Min 17.52 165 0.37 0.46 33.1 2.56
(Year) 1971–1972 1999–2000 1992–1993 1983–1984 1992–1993 1975–1976
Max 20.32 506 2.19 2.98 3266 50.25
(Year) 1989–1990 1995–1996 1994–1995 1994–1995 2007–2008 2007–2008
SD 0.69 71.2 0.52 0.59 696 10.6
CV(%) 27.0 44.4 45.6 123.3 86.0
600 200 3,5
500 Precipitation 3,0
ry mm( noitatipice)1- 234000000 AAls 110500mrHyA() 13-l 122,,,505 1-s6ry sn0o1At()
rP 50 1,0
100 0,5
0 0 0,0
1970 – 71 1980 – 81 1990 – 91 2000 – 01 2010 – 11
Figure7.Interannualvariationsofannualprecipitation,waterdischargeandsedimentyieldatAinHamarastation.
100 35 5.2.3 Analysisofbreakpoints
)%
( retaw ni emit launnA 468000 %A n Qtnimju ael iQn 9w8 ater 1122305050 m( egrahcsisD) 13- Tsoinechdce1ium9dr9roee1nud–tb1lir9ene-9sm2p1,ao9sbn8sus5etp–lt1ooh9fte8et6enh.neatbAirsleetrdsepeaeucmrsoiontod(dFai1irdgy9e.8n5b9ti–r)fe.1ya9Akc8h6wma/2naa0gsjo0esr9n–oibn2tri0ect1haekd0e
20 may be considered as a single period (with the relationship
5
“cumQ ”=0.021דcumQ”−9.417,r2=0.989).Thepe-
0 0 s
1970/71 1975/76 1980/81 1985/86 1990/91 1995/96 2000/01 2005/06 2009/10 rtrioandsi1e9n8t5e–v1en9t8t6o/w19a9rd1s–1a9n9e2wmreagyimthe.us be considered as a
Figure8.Variationofhydrologicalregimewithannualpercentage The response of sediment flow to various constraints dif-
oftimeofflowingwater,Q98(amongstdailydischarges,peryear) fersclearlyfromthatofdischargefromtheyear1985–1986
andannualstandarddeviationofdailyriverdischarge. onwards.Thisbreakcorrespondstothefirstyearofdryriver
overalongperiodinsummer(49days).Thisinitiatesaphase
ofintermittentflowregime.Theaveragedparametersforthe
tothenextisremarkablyregular:+85%betweenthe1970s
twoperiods1970–1985and1985–2010wereaddedtotheta-
and the 1980s, +84% between the 1980s and the 1990s,
bles, in addition to average values throughout the full study
+84% between the 1990s and the 2000s and is statistically
period and values for decades to illustrate the dynamics of
significant (+19.7×103tyr−1, on average, p<0.05). Spe-
thehydrologicalandhydrosedimentarychange.
cificsedimentyieldfollowsthesametrend(Table2).
5.2.2 VariabilityofmeanannualloadSPM∗ 5.3 Highdependencyofthesoliddischargeon
Qvariability
The average value of SPM∗ over the period 1970–2010 is
12.3gL−1, with annual values comprised between 2.5 and ThevariabilityofQandQ orSPM∗ atdifferenttimescales
s
50.2gL−1 (Tables1and2).Theirinterannualvariationwas werecompared.AO2007showedthat,over22years,71%of
smallerthanthatofsoliddischargebecauseannualSPM∗ is thevarianceoftheannualSPM∗valueswasaccountedforby
theratiooftheannualQs totheannualQ(whichincreased annualdischargeand73%bythe95thpercentileofdailydis-
lessthanQs). chargewithinthegivenyearQ95.ThismeansthatSPM∗was
www.hydrol-earth-syst-sci.net/20/1355/2016/ Hydrol.EarthSyst.Sci.,20,1355–1372,2016
1362 M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd
1121111 P toatTa 25
985–201019.47251.9629.044.81.2845.11.4543.7970–198518.51284.3423.10.81.0237.81.0238.2000–201019.49264.2219.728.11.4543.31.5742.2990–200019.37250.4240.559.91.1355.11.3455.2980–199019.19231.2316.824.10.9836.81.0741.5970–198018.32310.5319.41.21.1632.91.1632.9970–201019.09264.1027.028.31.1844.41.2945.7 (0)Q=daysofdry(C)(mm)(%)number(ms)(%)(ms)(%)3131◦−−AverageAverageCVyearlyAverageCVAverageCV daysChlefprecipitationaveragedischargedischargeofwetweriodat,yearlyNDD,,yearly,yearlyTPQQ thetwosubperiodsofperennialregime(1970–1985)orintermittentregime(19ChlefwasestimatedfrommeasurementsatDarElBeidaand/orMilianafor20ble2.Generalstatisticsonthehydrologicparameters(averages)fromtheWad dedFmsuaiiissagrgYcciiugtnhheryelaaae6yQ)s dneotta cluirmt0eu1mC( srbrrst9lggdiey.neertDwgi112Svpv05050oePeea0auenrrMQ b in ssy al b =t e∗ be rSy 0- roa .i2mP0a lt=nr0ihn 111MQ 4a0 tg999 (9s.ed ys 9897 σ= Qs8520 .∗ 1c00 ––– Qy4r- p.F 2121o0000a e=909l1.)0i0rn –918o4a0.n4r2054.dt10r9 eaQ8lo5ll9 ytC-afl h4hyut.sσe4mi,ie3ogQu7tdsnhhltaiQ 8a mete s0se b nys=t0 de ie td0 Qnlr .st2dl0a tw =h2(ar sH2y0ode4i.hmi9 lwQee9oyd2 ln-3we d)e1d0d1vyvei.29sied00eaac0rathsrairueoalsysmrngtwraQoeoarsfntkse1itganr6hab0flelna0lioe dndyweaelt.iaiahnlrrye--,
85miA ityover40-yearperiod(r2=0.956,Fig.10a).Ahighercor-
80897.015978.9113090.361498.333491.718078.8564123.3 (10tyr)(%)31−AverageCV sedimentloads,yearlyQ –2010),whilenononthswithmissingbdatAinHamara rcccceiolhfiolnaascrctegilsolueyensdvidioawmelnupa,eseenfsnoto.drybeittnehaltiidnso,ernadivntebdhree,ytwetdhaieerselcyynheσayarQreglaey(rrlvsy2oa=lrQiida0sb.d9ioil9sirtc1ySh,SaFthrYigag,en.t1ihos0enbms)d.opiIresne--
12.2161.1325.64.1358.864.213.9444.5455.611.0388.5247.57.3968.0134.54.3766.972.79.1878.6227.6 (tkmyr)21−−(ms)(%)sed.yield31−AverageCVspecific yearlyvaluesaverage98,averageofSSY,Q italicnumbersrefertotheaveravaluesover480months).Italicgaugingstationperdecadeand 6Tbpm1ue9hurt7eamV0httyhusyaeoerdraafeirrmnailonfdytorlcioonvo2rgmaetn0hila0culosMya0eeflssatai)pnohv,racfeeiJhrtrtuseaanremgtmeoaeempses(Noe+tasrneioahn3ravtose.ule3iwdrtm0eyeq◦bdaouCetfirhC,tceioihglnniclhmceoarfdnveagisaetsetisrcacnaernegdeatrepnaw,iadlnblniytecDhtiiweneascce.reemeTmnaeasbtmxheeide--r
16.6567.42.9740.5760.4158675.1658.91.2130.8180.51934620.5568.74.4400.4120.38432414.3669.22.7530.6590.4183439.9357.02.0490.6490.4493164.5447.91.0210.8900.57324012.386.02.2700.6470.4311213 (gL)(%)1−AverageCVabRN2 SPMRatingcurveparameters∗ gevaluesperdecade.numbersrefereithertothefull40-yearperiodofortheentireperiodfrom1970to2010(note:T tM2ImaAaastauwenatnnr.euur2eimaddseAnsrg2yevgeiueviuFQ)mn◦anismesvpeCilrturtnsueebmT.haimcnersragaceirusainobeeneir1atnndddlaeyg(re9Dts.,myasr7fseCe3baeaeh0pdadnal.scoasenlosbnenrTrdaoavlpmyuasuhnnnalitad1tybdueddlaleeu..eerl6cttseeaTrrvvhe6e–rissvahamneeaFe◦ldlea(gsuurC2epSosoaeibte0eew,ennhgssrr0dspcaeeuanau0otrldtabeanemsfusrvrmy,adlroyPeimesewgv)lb0irea,nheeaehd.tdQ9wrrtgriiiplev–l8fveyae9efceNaa◦esvrrasnlcCrecoacunmpedeaovloedrnunsioneinQsncenetmmsst(edsgrteosJJsiabcpmfubafvxbeueoinuntaiaynrthrrumtel)irciuves–akeo0rbueeerAaeeny.ayyms3cdaeyush3s(oaoo1egiMtefnrev◦nd.audl2mCery,ePasJ9srwctwurpv),ca◦lbtaieiabhyhdCQnterrly–h–eee----.
r
Hydrol.EarthSyst.Sci.,20,1355–1372,2016 www.hydrol-earth-syst-sci.net/20/1355/2016/
M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd 1363
Table3.Variationofprecipitation,waterdischargeandsedimentyieldaveragedperseasonovereachdecade.
Precipitation(mm) Waterdischarge(m3s−1) Sedimentyield(103t)
Autumn Winter Spring Summer Autumn Winter Spring Summer Autumn Winter Spring Summer
1970–1980 68.5 102.6 128.2 11.2 1.26 1.38 1.41 0.58 62.2 43.7 66.1 8.4
1980–1990 56.0 94.4 70.7 10.1 1.15 1.29 1.08 0.40 128.8 61.0 97.2 46.8
1990–2000 67.0 81.1 86.9 15.5 1.86 0.99 1.11 0.54 279.1 57.8 130.9 146.0
2000–2010 78.6 98.4 77.7 9.5 3.04 1.35 1.06 0.35 804.9 94.4 195.3 35.4
Precipitation(%) Waterdischarge(%) Sedimentyield(%)
Autumn Winter Spring Summer Autumn Winter Spring Summer Autumn Winter Spring Summer
1970–1980 22.1 33.0 41.3 3.6 27.3 29.8 30.3 12.6 34.5 24.2 36.6 4.7
1980–1990 24.2 40.8 30.6 4.4 29.4 32.8 27.7 10.1 38.6 18.3 29.1 14.0
1990–2000 26.7 32.4 34.7 6.2 41.3 22.1 24.6 12.0 45.5 9.4 21.3 23.8
2000–2010 29.7 37.3 29.4 3.6 52.5 23.2 18.2 6.1 71.2 8.4 17.3 3.1
60 1600
SPM* = 2.749 Q + 0.642 SSY = 24.989 Q1.401
50
r 2 = 0.956 1-ry) 1200 r 2 = 0.991
1-L) 40 2-m
g( *MPS ylraeY 2300 k t( YSS ylraeY 800
400
10 (a) (b)
0 0
0 4 8 12 16 0 4 8 12 16
Yearly Q (m3 s-1) Yearly Q (m3 s-1)
Figure10.Yearlyaverageofrelatedsedimentloadparametersvs.intra-annualvariabilityofdailyriverdischarge,characterizedbytheir
∗
annualstandarddeviation.(a)displaysSPM ,(b)specificsedimentyield.
presented in Fig. 11. The monthly values of P, Q and Q season(November–December),aswellasthestrength-
s
per decade over 40 years also clearly illustrate the absolute ening of rain in the warm season (July–October) and
changes in intensity and in seasonality of the river regime in winter (January–February)); (2) advancement of the
(Fig.12).ThemainconclusionsoftheanalysisofP,Qand rainy season as evidenced by precipitation in Septem-
Q seasonalvariationsarethefollowing: berandOctober;(3)spreadingoftherainyseasonover
s
9months(September–May)for1985–2010fromprevi-
– Rainfall decreased in spring and increased in autumn.
ously 7 or 8 months (from October or November on-
Precipitationinautumnincreasedfrom22to30%atthe
wards);(4)increasedregularityofrainyseasonprecipi-
expense of spring rains (decreasing from 41 to 29%).
tation.
For the decade 2000–2010 precipitation was the same
in autumn and in spring (78mm) while for the decade
– Proportionally, flow decreased from winter to summer
1970–1980springrainfallwas87%higherthaninfall
and increased dramatically in autumn from just over
(seeTable3andFig.11a).
a quarter (27.3%) of the flow delivered during 1970–
– Average monthly rainfall from six weather stations in 1980tomorethanonehalf(52.5%)during2000–2010
theriverbasinfor1970–1985and1985–2010(Fig.13) (Table3andFig.11b).Flowdecreasedinsummerand
illustrates the changes. Two marked seasons typical of theriverbecamedryformuchofthesummer.Overthe
aMediterraneanclimatearepresent(adryseasonanda last decade, it is striking to see the difference between
rainyseason)butthefollowingchangesareobservable: the average flow rates in fall and spring: the fall rate
(1) differences between seasons decrease, as indicated is almost 3 times that of spring with almost the same
bytheCVofmonthlyrainfallfrom57.3 in1970–1985 rainfall. This trend is evident over the 40-year period
to 45.9% in 1985–2000 (there is a decrease of spring (Fig.11b).
rains (March–May) and at the beginning of the cold
www.hydrol-earth-syst-sci.net/20/1355/2016/ Hydrol.EarthSyst.Sci.,20,1355–1372,2016
1364 M.AchiteandS.Ouillon:Recentchangesinclimate,hydrologyandsedimentloadintheWadiAbd
% ni noitubirtnoc lanosaes noitatip)egareva raey 9( 43321124505055005 (PWSSAapuui)nrmtiutnemmgrenr edaced rep degareva noitatipicerp ylhtnoM1-htnom mm() 6453120000000 (a) P 1211909980790000–––– 1212909091800000
ice S O N D J F M A M J J A
rP 0
% ni noitubirtnoc lanosaes egrah)egareva raey 9( 4531260000001970 1980 1990 2000 2010 (QWSAbpui)nrtiutnemgrn edaced rep degareva egrahcsid ylhtnoM31-ms() 0123456S O (bN) DQ J F M A M J J A
launna naem ot noitubirtnoc lanosaeScsiD)egareva raey 9( )%( daol tnemides 86420000001970 1980 1990 2000 2010 (QSWSSAcupuui)msnrmtiutnmemmgreenrr aFvigeruargeedaced rep degareva xulf tnemides ylhtnoMed113-htnom sn0o1t() 2o.v110M0e100.10r100odSnetchalOdyevsaNilnu(etchs)DeoQWfsJpardeicAFipbidtaMtbiaosniAn(.a)M,QJ(b)JandAQ s (c)
1970 1980 1990 2000 2010
Figure11.Trendsoftheseasonalindexesofprecipitation(a),dis-
– In absolute values, solid discharge has been increasing
charge(b)andsedimentdischarge(c)intheWadiAbdbasin.
in all seasons over 4 decades, but more so in the fall
than in the other seasons (Table 3 and Fig. 12c). Dur-
– These results point towards a change in runoff as ing autumn, it more than doubled from one decade to
defined by the ratio Q/P. Considering the whole another.Duringtheotherseasons,itdoubledortripled
basin area, the river discharge at the Ain Hamara sta- betweenthe1970sand2000s(seeTable2).Whiledur-
tion averaged over 40 years corresponds to a water ing the 1970s the Wadi Abd had two major periods of
depth of 15mmyr−1, while the average precipitation roughly equivalent sediment discharge in the fall and
is 264mmyr−1. For comparison, on average, 85% of spring,suspendedsedimentloadsweregreaterintheau-
rain in this region evaporates and the remaining 15% tumnduringthe2000s(>70%oftheyearlydischarge).
runsintosurfacewatersorinfiltratesundergroundstor- The Wadi shifted from a regime with two equivalent
age(Sari,2009,quotedbyBenhamicheetal.,2014).On seasons of sediment production to a regime with one
theWadiAbd,Q/P averages5.7%.Wecalculatedthe dominant season in the 2000s. Autumn produced over
value of Q/P averaged over 3 consecutive years and 4timesmoresedimentthanspringinthe2000s(Table3,
over 3 consecutive months (centred) and then took the Fig.11c).Thisphenomenondoesnotseemtobedueto
averageperdecade(Fig.14).ItappearsthattheQ/P ra- someexceptionalfloodsbecausethetrendisobservable
tio remains constant during the months from Decem- over4consecutivedecades(Fig.11c).
ber to April (around 4.4%, on average), it increased
slightlyinNovemberandMayduringthedecade2000–
2010 and it increased significantly from September
to November. In other words, runoff increased, rain
decreased slightly and the temperature (and therefore
ETP)increased.Asaconsequence,infiltrationwillde-
crease and the water level in the aquifers will be low-
ered.
Hydrol.EarthSyst.Sci.,20,1355–1372,2016 www.hydrol-earth-syst-sci.net/20/1355/2016/