Restrepo&Kjerfve

Magdalena river: interannual variability (19751995) and revisedwater discharge and sediment load estimatesJ.D. Restrepoa,b, B. Kjerfvea,c,*aMarine Science Program, Department of Geological Sciences and the Belle W. Baruch Institute for Marine Biology and Coastal Research,University of South Carolina, Columbia, SC 29208, USAbDepartamento de Geolog a, Area de Ciencias del Mar, Universidad EAFIT (Escuela de Administracion, Finanzas y Tecnolog as), A.A. 3300Medell n, ColombiacDepartamento de Geoqu mica, Universidade Federal Fluminense, CEP-24.020-007 Niteroi, BrazilReceived 23 December 1999; revised 22 May 2000; accepted 25 May 2000AbstractMagdalena river discharges 228 km3of water annually, or on the average 7200 m3s1, into the western Caribbean based on21 years of daily data, 19751995. The mean sediment load for the Magdalena is 144 106t yr1, corresponding to a sedimentyield of 560 t km2yr1for the 257,438 km2basin. Magdalena is the largest river discharging directly into the Caribbean Sea,andit hasthehighest sediment yieldofanymedium-sizedorlargeriver alongtheentireeast coast ofSouthAmerica.Magdalena sediment load is well correlated with water discharge (R2 0.76). Regression of water discharge on the SouthernOscillation Index (SOI) shows that 69% of the variability of the Magdalena streamow is explained by the SOI, with highdischarge occurring during La Nina phase and low discharge during El Nino phase. Cross-spectral analysis between Magdalenadischarge and the SOI indicates an average recurrence interval of 3.0 years with a coherence, Y2 0.90, and that the dischargeanomaly is in phase with the SOI anomaly. Analysis of time series of sediment load, 19751995, indicates that la Nina phase ischaracterized by a sediment load as high as 1600 103t day1. Sediment load is also correlated with SOI with a coefcient ofvariation of R2 0.54. Clearly, the Magdalena water discharge and sediment load are strongly coupled to the El NinoLa Ninacycle. 2000 Elsevier Science B.V. All rights reserved.Keywords: Magdalena river; Discharge; Sediment load; Sediment yield; Time series analysis; El Nino1. IntroductionWater discharge and sediment load from theMagdalena River have previously been reported as apart of global water and sediment budgets (Baumgart-ner and Reichel, 1975; Meybeck, 1976; Alekin, 1978;UNESCO, 1978, 1979, 1985, 1992; Milliman andMeade, 1983;Millimanetal., 1995). Mostofthesedischarge and load estimates were based on a 19711972studybyNetherlandsEngineeringConsultants(NEDECO, 1973). However, the validity of theNEDECOdata set is questionable because of theshort record and the possible inclusion of bothbedloadandsuspendedload(MillimanandMeade,1983).The primary focus on rivers draining South Amer-ica has been directed towards three large uvialsystems, Amazon (Meade et al., 1979, 1985; RicheyJournal of Hydrology 235 (2000) 1371490022-1694/00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved.PII: S0022-1694(00)00269-9www.elsevier.com/locate/jhydrol*Corresponding author. Address. Marine Science Program,Department of Geological Sciences and the Belle W. Baruch Insti-tute for Marine Biology and Coastal Research, University of SouthCarolina, Columbia, SC 29208, USA. Fax: 1-803-777-4600.E-mail address: [email protected] (B. Kjerfve).J.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 138Fig. 1. Map of the Magdalena River drainage basin, Colombia, showing the principal tributaries, the headwaters at La Magdalena lagoon(triangle), and the hydrological station (circle) at Calamar, where water discharge and sediment concentration were measured. The differentrainfall regions in Colombia are marked A, B, C, D and E.et al., 1986, 1989), Orinoco (Eisma et al., 1978;Paolini and Ittekkot, 1990; Depetris and Paolini,1991) and Parana (Depetris et al., 1996; DepetrisandGaiero, 1998; Goniadzki, 1999). However, thesediment load for the Magdalena River is of thesame magnitude as these three rivers, which all havemuch larger drainage basins (Bassin, 1976; Millimanand Meade, 1983; Javelaud, 1987). Milliman (1990)calculated that the Amazon, Orinoco and Paranariverscontribute11%of theglobal sediment load,andinourestimateareresponsiblefor24%oftheglobal water discharge tothe oceans. Inaddition,Milliman and Meade (1983) claimed that the Magda-lena River transports more sediment than the Orinocoand Parana rivers, although it has much smaller waterdischarge and drainage area. They estimated anannual sediment load of 220 106t. Whether theMagdalenaRiver indeedhassuchahighsedimentload warrants re-examination.Several studies have shown that the El Nino-South-ern Oscillation (ENSO) is a major forcing mechanismof climatic and hydrological anomalies (Kousky et al.,1984; Ropelewski and Halpert, 1987; Rogers, 1988;Barnett, 1991; Rasmusson, 1991; Mechoso and Perez-Iribarren, 1992). In tropical South America, particu-larly Colombia, interannual hydro-climatology isstrongly inuenced by extreme phases of ENSO(Poveda, 1997). Long-term rainfall analysis conrmsthat El Nino events are associated with negative rain-fall anomalies and La Nina with positive anomalies innorthernSouthAmerica(PovedaandMesa, 1997).Althoughseveral studieshavefocusedontheinter-annualvariability and hydrologic anomalies of rain-fall andriver dischargesinColombia(PovedaandMesa, 1993, 1995, 1996; Poveda, 1994, 1997;Restrepo and Kjerfve, 2000), the impact on theMagdalena River has received scant attention.It is our objective to explore the variability of waterdischarge and sediment load of the Magdalena River(Fig. 1), the principal uvial system of Colombia, andrelate this variability to the ENSO cycle.2. The Magdalena river systemTheAndesof Colombiaconsistsof threenearlyparallel andnorthsouthorientedmountainranges,which merge into a single range near the Ecuadorianborder. Between these ranges lie two river valleys, thehigh and narrow Cauca valley to the west, and the lowand broad Magdalena valley to the east (Fig. 1). TheMagdalena River is 1612 kmlong and drains a257,438 km2basin. It is the largest uvial system inColombia and originates fromheadwaters in theAndeanCordilleraat anelevationof 3300 m. ThemaintributariesaretheCauca(secondlargest riverin Colombia), Sogamoso, San Jorge and Cesar rivers(Fig. 1). AccordingtoPotter(1997), LateMiocenedeposits in the Magdalena Valley between the Easternand Western Cordilleras of Colombia indicate a LateMiocene age for the Magdalena River. The paleo-Magdalena, and its principal tributary, the paleo-Cauca, both developed in tectonic lows when the East-ern and Central Cordilleras were uplifted. Thus tectoniccontrol is evident along the Magdalena watershed.The Magdalena basin occupies a major portion oftheColombianAndes. Theareaischaracterizedbymoderate rainfall, which averages 1500 mm in regionA, 2600 mmin region B, 1200 mmin region C,3100 mmin region Dand 1350 mmin region E(Fig. 1). The mean rainfall for the drainage basin asa whole is 2050 mm yr1. The annual distribution issimilar throughout the watershed. There are two wetandtwodryseasons. DecemberMarchandJuneSeptember are low rainfall periods and MarchMayand OctoberNovember are high rainfall periods. Thetwo wet seasons are comparable in length and inten-sity, except in the upper Magdalena valley, where theearly rst wet season is more prolonged (Snow, 1976;HIMAT; Instituto Colombiano de Hidrolog a, Meter-eolog a y Adecuacion de Tierras, 1989). The oor ofthe Magdalena valley has an annual mean temperatureof 2728C, and a mean elevation of 74 m. The meanrelativehumidity is 7075%. The annualvariabilityof mean monthly temperature measures only ^3C inspite of large variations in rainfall (COLCIENCIAS;Instituto Colombiano para el Desarrollo de la Cienciay la Tecnolog a Francisco Jose de Caldas and FEN;FinancieraEnergeticaNacional, 1990; IGAC; Insti-tuto Geograco Agust n Codazzi, 1995).The MagdalenaRiver discharges into the westernCaribbean and forms a 1690 km2triangular delta(Coleman, 1976). The delta plain consists of alluvialplains, marginal lagoon systems and beach ridges(Vernette, 1985). The receiving basin is characterizedby sedimentation, slumping and compressionalJ.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 139tectonics that causes the presence of mud diapirism inthe delta front (Shepard et al., 1968; Kolla and Bufer,1984; Vernette et al., 1992). The onshore plainsbetweenCartagenaandtheactual Magdalenadeltaare coveredbythickPlioceneQuaternaryterrige-neous deposits which resulted from successive migra-tions of the Magdalena River (Vernette et al., 1992).According to Bordyne (1974), the mouth of the river,situated near Galerazamba during the Pliocene epoch,migrated westwardduringthe Pleistocene towardsCartagena, then migrated to the north in the Recent.The present delta mouth empties intoanoffshorecanyon with a steep slope (40) (Shepard, 1973).The shoreline is extensively reworked by waveprocesses.Averagewavepoweris206 ergs s1m1ofcoastline(ColemanandWright, 1975; Coleman,1976). The delta front experiences a microtidalrangelessthan0.5 m(Kjerfve, 1981; Mart nezandMolina, 1992). Strong littoral currents predominantlyow towards the west and are the result of open oceanswells, generatedbyNEtradewinds(Lorinet al.,1973).J.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 140Fig. 2. (A) Monthly mean and standard deviation of water discharge; and (B) sediment load, in the Magdalena River at Calamar, 19751995.3. Mean water dischargeWeobtaineddailywater discharge, 19751995,from the downstream station at Calamar fromInstituto de Hidrolog a, Metereolog a y EstudiosAmbientales (IDEAM, Colombia). Calamar islocated 112 km upstream from the Caribbean(Fig. 1). Discharges are based on daily waterstage measurements 19751995. Simultaneousmeasurements of water level, river discharge andsedimentconcentrationweredoneon55occasionsbyIDEAMbetween1975and1995duringhigh,intermediate and lowriver discharge conditions.The daily stage readings for the 21 year recordJ.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 141Fig. 3. Based on 55 sets of measurements in the Magdalena River at Calamar 19751995, we plot: (A) measured sediment load (S, t day1) vs.measured water discharge (Q, m3s1); and (B) measured suspended sediment concentration (kg m3) vs. measured water discharge (m3s1),distinguishing between high, low, rising and falling stage and river discharge.were converted to discharges via the establishedratingcurve.Themeanannual water dischargeof MagdalenaRiver at Calamar is7200 m3s1withameanlowdischarge of 4068 m3s1in March and a mean highdischargeof10,287 m3s1inNovember(Fig. 2A).The annual volume of water discharged into theCaribbeanSeais 228 km3. Thisdischarge is similarto earlier estimates of 7500 m3s1(Wright and Cole-man, 1973;Coleman, 1976;Meybeck, 1976, 1979),7474 m3s1(Milliman and Meade, 1983),7600 m3s1(Vernette, 1985) and 7421 m3s1(Mar n, 1992), 7100 m3s1(Alvarado, 1992, 1998).4. Sediment loadSediment load estimates at Calamar were based onthemeasuredsediment concentrations, cross-multi-pliedwithwater discharge. Regressionofsedimentload on water discharge for the 55 measurement occa-sions yieldedarelationshipwhichwasusedtoesti-mate daily sediment loads 19751995 (Fig. 3A). Forthe major Magdalenatributaries, wealsogatheredmonthlydischargeandloaddata, 19751993, fromIDEAM (1995) but are not showing these. Sedimentloadis well explainedbywater discharge for theMagdalena River. Curvilinear regression of instanta-neous daily sediment load on water discharge yieldedacoefcient ofdeterminationof0.76, signicantat95%. ThemeasuredsedimentloadatCalamarindi-cates values as high as 910 103t day1(Fig. 3A).Measured sediment concentrations during high,intermediateandlowstreamowconditions 19751995, indicate that in the Magdalena River, the seaso-nal relation of sediment concentration on waterdischarge formsa clockwiselooporhysterisis(Fig.3B). That is, mean sediment concentrations(0.67 ^0.20 kg m3) duringrisingstage level anddischarge were greater than mean concentrations(0.42 ^0.18 kg m3) duringfallingstagelevel anddischargeat equal discharges. Meanhighandlowstage/discharge means were 0.56 ^0.15 and0.43 ^0.11 kg m3, respectively. According toMeade (1988), the relation between sediment concen-trationandwater dischargeisnot aperfect powerfunction. Therelationbetweensediment concentra-tion and water discharge, when plotted on a loglogscale graph, often forms a clockwise loop. Such clock-wise hysterisis is observed in the Magdalena data andis typical of many large rivers, including the Missis-sippi, Amazon, OrinocoandParana (Meadeet al.,1985; Meade, 1988; Depetris and Kempe, 1991;Depetris and Gaiero, 1998).Dailysediment loaddata, averagedfor 21 years,yieldedanestimate of 143.9 106t yr1, equal to86% of the total sediment load of all Colombian riversdrainingintotheCaribbean(RestrepoandKjerfve,2000). Themaintributary, theCaucaRiver, contri-butes 31%of the total Magdalena sediment load.The seasonal distribution of sediment load at Calamar(Fig. 2B)indicateshighvaluesof690 103t day1and 678 103t day1during November and Decem-ber. Secondary high sediment loads occur duringJuneJuly with loads reaching 443 103t day1.Theannualsedimentloadof143.9 106t yr1ishigher than the 133.06 106t yr1reported by Mar n(1992), butconsiderablylowerthantheestimatebyMilliman and Meade (1983) of 220 106t yr1basedon the NEDECO (1973) study. Alvarado (1992, 1998)estimatedatotal loadof180 106t yr1, including150 106t yr1of suspended sediment load and30 106t yr1of bedload,whichalsois lowerthantheNEDECO(1973) estimate. Ourestimateof thesediment load for the Rio Magdalena implies a sedi-ment yieldof 560 t km2yr1for the257,438 km2upstreambasin. Becauseof the recordlength, ouryieldismorerealisticthanthepreviouslyreportedyields of 1000 t km2yr1(Meybeck, 1976;Meybeck, 1988), 900 t km2yr1(Milliman andMeade, 1983) and 920 t km2yr1(Milliman andSyvitski, 1992).5. Interannual variability 19751995ENSOanomalies have been linked to climaticanomaliesandriverowworldwide(Kouskyetal.,1984; Ropelewski and Halpert, 1987; Probst andTardy, 1987; Probst andTardy, 1989). IntropicalSouth Americathereis a coherent patternof hydro-logical anomaliesduringextremephasesofENSO.This is clear inCostaRica(Waylenet al., 1996),Panama (Estoque et al., 1985), Venezuela (Pulwartyet al., 1992), Ecuador (Gessler, 1995) and the Amazonbasin (Richey et al., 1989; Vorosmarty et al., 1996).J.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 142Northeastern South America, including north equator-ialBrazil, French Guyana,Surinam,Guyana,Vene-zuela and northeastern Colombia, has one of the mostconsistent ENSO-precipitation relationships of anyfound anywhere (Ropelewski and Halpert, 1987). Inthe Magdalena and Cauca basins, El Nino results inrelatively dry periods, and La Nina is associated withexcessive rainfall.Inthe Magdalena River, water discharge variessignicantly interannually. The mean discharge isJ.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 143Fig. 4. Time series plots of mean monthly (thin lines) and low-frequency pass lter with zero-phase (bold lines) (A) discharge for MagdalenaRiver 19751995 at Calamar; (B) the Southern Oscillation Index (SOI) (National Oceanic and Atmospheric Administration-NOAA, 1999,data-base on the Internet at http://ftp.ncep.noaa.gov/pub/cpc/wd52dg/data/indices); and (C) sediment load of the Magdalena River at Calamar,19751995.7200 m3s1, and the seasonal root mean square (rms)variability 2020 m3s1. The Magdalena discharge atthe Calamar station (Fig. 4A), the smoothed monthlyvalues of the Southern OscillationIndex (dened asthe sea level pressure differencebetweenTahitiandDarwin) (Glantz, 1997) (Fig. 4B) and the Magdalenasediment load at Calamar (Fig. 4C) show very goodcoherencefor the21 year period19751995. Peakowsusuallyexceed12,000 m3s1duringLaNinayearsandlowdischargesof 20003000 m3s1areobserved during El Nino years (Fig. 4A). Mean annualdischarges during El Nino and La Nina years are 5512and 8747 m3s1, respectively.A low-pass frequency Butterworth lter wasappliedtoremove highfrequencyoscillations andemphasizethe interannual variability. The lter wasofeighthorderwithahalf-gainfrequencyof0.045cycles per month. The lter was applied by perform-ingzero-phasedigital ltering, processingtheinputdatainboththeforwardandreversedirectionsandyielding zero-phase distortion (Oppenheim and Scha-fer, 1989). The seasonal variability, with highdischarge in NovemberDecember and low dischargein FebruaryMarch is pronounced in the interannualcycle. Regression analysis between the smootheddischarge and the smoothed SOI yielded a coefcientof variation of R2 0.69, signicant at the 95% con-dence level, which indicates that variations in the SOIexplain 69% of the variability in discharge, with highvalues of theSOI correspondingtopeakLaNinaconditions and peak Magdalena discharge.Strong relationships between hydrologicalanomalies andtheSOI indexhavebeenfoundinother rivers in South America. For the Amazon,Sao Francisco, and Paranarivers, regression analy-sisofdischargeontheSOIindexyieldedacorre-lation coefcient of 0.41 (Probst and Tardy, 1989).Kouskyet al. (1984) foundthat the level of theParana River at Rosario (Argentina) was corre-lated with the SOI index with a coefcient ofvariation of R2 0.56. Other good correlationshavebeendocumentedinRioTrompetas (easternAmazonia) (Molion and Moraes, 1987 cited byRicheyet al., 1989), theAmazonRiver at Mana-capuru (Richey et al., 1989) and the Amazonbasin (Poveda and Mesa, 1997). For the SanJuanRiver, Paciccoast of Colombia, regressionanalysis of water discharge on the SOI indexyielded a coefcient of variation of R2 0.64(RestrepoandKjerfve, 2000).Several studies have focusedonthe mechanismresponsible for the hydrological anomalies in northernandsouthernSouthAmerica(Kouskyet al., 1984;Ropelewski and Halpert, 1987; Aceituno, 1989;Mechoso and Perez-Iribarren, 1992). During the posi-tiveLa Nina phase of the Southern Oscillation withTahiti pressure higher than at Darwin, southeast tradewinds are stronger than usual, and the Inter TropicalConvergence Zone (ITCZ) remains north of its typicalpositionintheEasternPacic. Thisresultsindrierthan normal conditions in southeastern South Amer-ica, but induces strong precipitation in the northeast-ern parts of the continent. During the negative El Ninophase of the SO, the southeast trades are weakened,permitting a more southward displacement of theITCZin southeastern SouthAmerica, resulting inabnormallyheavyrainfallthere. Duringthisperiod,ascending atmospheric motions are strengthened,promoting increased precipitation over southernportions of the continent.La Nina conditions were pronounced in 19751976, 19811982, 19881989and19992000, andEl Ninoconditions in19771978, 19911992and19971998 in the Magdalena discharge. Our data donot cover the most recent El Nino and La Nina events.Althoughthe19821983El Ninoevent haddrasticeffectsontherainfalldistributionwithinthetropicsand was generally the strongest on record (Kousky etal.,1984), itdidnotcauseintensedryanomaliesinnorthern Colombia (Poveda and Mesa, 1997). In fact,the El Nino of 19821983 failed to have the expectedresponse on Magdalena discharge (Fig. 4A).However, during the 19911992 El Nino, Colombiaexperiencedoneofthemoredrasticdryseasonsonrecord (Poveda andMesa, 1997). This eventcausedprolonged electricity shortages and produced losses ofabout US $1 billion to the economy of Colombia. LaNina conditions, on the other hand, are usually asso-ciated with high rainfall anomalies and ood condi-tionsontheMagdalenaRiver. ThestrongLaNinaduringNovemberDecember1999causedoodsinthe lower Magdalena valley, which reportedlyproduced the strongest oods during the past 40 years.Fourier analysis (Welch, 1967) of the ltered timeseries of water discharge, 19751995, reveals aspectral peakat aperiodof 36months (Fig. 5B).J.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 144The tendency for oscillations with a 3.0 year period isevidentinthelteredtimeseries(Fig. 5A). Similarperiodicities have been shown for the Amazon(2.4 years; Richey et al., 1989) and Parana rivers(3.3 years; Depetriset al., 1996). Also, PovedaandMesa (1997) identied spectral peaks of rainfallanomalies in Colombia associated with ENSO at 54,43, and 26 months.To calculate the statistical signicance of the rela-tionship between ENSO events and Magdalena Riverdischarge anomalies, we performed cross-spectralanalysis (Panofsky and Brier, 1968) between theltered 19751995 Magdalena discharge and theSOI index. The coherence function (Y2) estimatebetweenthelteredMagdalenaandSOIindexwassignicant at the 95% condence level at a period of3 years with a coherence of Y2 0.90 (Fig. 5C). Also,the cross spectrum magnitude(not shown) reveals apeak at a frequency of 0.028 cycles/month or a period of36 months. The corresponding phase spectrum showsthat discharge anomalies at Calamar (10150 N, 74550W) are almost exactly in phase with SOI anomalies ataperiodof 36months(Fig. 5D), implyingnolagbetween the SOI and the discharge of the river.For South America, the phase lag of the SOIincreaseseastwardanddecreasesnorthward(Probstand Tardy, 1987). Thus our phase lead/lag of 0-month at a period of 36 months appears to be reason-able when it is compared to other phase valuesobserved in other South American rivers, e.g. Suma-pazRiver inColombia, Magdalenabasin, 4000N,74300W(4 months, Poveda and Mesa, 1997),J.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 145Fig. 5. (A) Demeaned water discharge of the Magdalena River at Calamar, 19751995. A ltered discharge was calculated by subtracting the21 year mean discharge for each month (Q) from the respective monthly mean discharge in each year (Q) for the ith month of the jth year toform the deviation from the long-term monthly mean discharge Q0ij QijQi(Richey et al., 1989); (B) fast Fourier transform (periodogram)of the ltered discharge. The period (months) is estimated as the total record length (252 months)/relative frequency; (C) coherence spectrum atthe 95% condence level between the SOI index and the Magdalena discharge based on ltered monthly means 19751995; and (D) phase(months) of the cross-spectrum magnitude between the SOI index and the Magdalena ltered water discharge. We have chosen not to show thevariance spectrum of the SOI and the magnitude of the cross-spectrum.Amazon (6 months, Richey et al., 1989) and Parana (8months, Depetris et al., 1996). Although Calamar andtheSumapazRiver stationsarelocatedat differentlatitudes, but almost at the same longitude, ourphasevalueappearstobecoherent withrespect tothe progressive delay northward of the ENSO.The sediment load at Calamar also varies interan-nually (Fig. 4C). The low-frequency pass ltered sedi-ment load for the 21 year period 19751995, showsaninterannual oscillation well correlated withtheENSO cycle. Regression analysis of the low-frequency sediment load on the smoothed SOI yieldeda coefcient of variation ofR2 0.54, signicantat95%, indicatingthat variations intheSOI explain54%of the variability in sediment load. The LaNina high owin 19881989 caused a markedincreaseinsedimentload, withoneprominentpeakof 1600 103t day1. Other ood events are clear butless pronounced, e.g. in 19751976, 19811982 and1995. Lowsediment loadsoccurredduringElNinoevents in19771978, 19821983and19911992.ThemeandailysedimentloadsduringElNinoandLa Nina years are 256 t day1and 511 t day1,respectively.6. DiscussionOur data indicate that the Magdalena River contri-butesapproximately9%of thetotal sediment loaddischarged fromthe east coast of South America(Table 1). In addition, the Magdalena River appearsto have the highest sediment yield of the large riversalong the Caribbean and Atlantic coasts. It is almostthree times greater thantheyieldof theAmazon,190 t km2yr1, Orinoco, 150 t km2yr1, Negro(Argentina), 140 t km2yr1(Milliman and Syvitski,1992) and much greater than the yield of the Parana,30 t km2yr1(Milliman and Syvitski, 1992;Goniadzki, 1999), Uruguay, 45 t km2yr1andSaoFrancisco, 10 t km2yr1(Milliman and Syvitski,1992) (Table 1).According to Milliman and Syvitski (1992), basinareaandrelief arethemajor controlsonsedimentyield, with climate, geology and land-use beingsecond-order inuences. In the Magdalena basin,slopes steeper than40 lead toexcessive erosion.Besidestectonicactivityandmorphological factors,J.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 146Table1Drainagebasin,waterdischarge,sedimentload,calculatedyieldsandreceivingbasinforselectedriversinSouthAmerica(Normalizedsedimentyieldfortheriverbasinswasestimatedbydividingsedimentload(tyr1)bydrainagebasinareas(km2);Arg

Argentina;Bra

Brazil;Col

Colombia;Uru

Uruguay;Ven

Venezuela.)RiverBasinarea(106km2)Waterdischarge(km3yr1)Sedimentload(106tyr1)Sedimentyield(tkm2yr1)ReceivingbasinR.Amazon(Bra)a,b6.163001200190N.AtlanticR.Orinoco(Ven)a,b0.991100150150N.AtlanticR.Parana (Arg)b,c2.64707930(?)S.AtlanticR.Magdalena(Col)d0.25228144560CaribbeanR.Atrato(Col)d0.0358111315CaribbeanR.Uruguay(Uru)c,d0.2425311(?)45(?)S.AtlanticR.Negro(Arg)b0.103013140S.AtlanticR.S.Francisco(Bra)b0.6497610S.AtlanticR.SanJuan(Col)d0.01482161150N.PacicR.Pat a(Col)d0.0141014972N.PacicR.Chira(Peru)a,b0.025201000S.PacicaDataweregatheredfromMillimanandMeade(1983).bDataweregatheredfromMillimanandSyvitski(1992).cDataweregatheredfromGoniadzki(1999).dDataweregatheredfromRestrepoandKjerfve(2000).forest cover in the Andes section has greatlydecreased due to population expansion. Deforestationhasleadtoseveresoilerosion. Theonlyremainingrainforest areaislocatedinthelower Magdalenavalley, whereasmost ofthelandonthelower andmiddle slopes is undercultivation. In addition, highconcentrations of suspended sediments, often greaterthan 800 mg l1, have resulted from the rapid erosionof the lowlands, partly because of ongoing goldmining in the Cauca basin (IGAC; Instituto Geogra-co Agust n Codazzi, 1995).The time series of Magdalena water dischargeconrms that La Nina events are associated with posi-tive discharge and sediment load anomalies. Theimpactof La Nina is stronger than the impactof ElNino. Phase analysis between discharge and SOIindex indicates that Magdalena discharge is in phasewith the SOI anomalies at a period of 36 months. InColombia, extreme climatic anomalies, includingprolonged drought and excessive rainfall, have adramatic impact on the economy, as well as thelivesof theinhabitants of affectedregions.A betterunderstanding of hydrological anomalies can improvemodels and forecasts that contribute to minimize risksduring ood and drought events. In addition, linkagesbetween the Magdalena discharge and ENSO anoma-lies reinforce the importance of determining thefactors controllingthe hydrology of the basin in thepresence of extensive man-inducedalterations andland-use change.The water and sediment discharges of the Magda-lenaRiver havegreat environmental andeconomicimpacts on the adjacent coastal ecosystems. TheCanaldelDique(Fig.1),a114 kmlongman-madechannel fromthe Magdalena River at Calamar toCartagena Bay, was constructed in 1514 by indigen-ous slaves supervised by Spanish conquistadors. Thecanal has a mean annual water discharge of299 m3s1andsediment loadof 4.76 106t yr1.Discharges as highas of 800 m3s1andsedimentloads as high as 600 103t month1often occurduring November (Restrepo and Kjerfve, 2000).Since 1954, the government of Colombia has dredgedCanal del Dique, whichhasincreasedthesedimentloadintoCartagenaBay. Becauseof theincreasedsedimentation in the bay during the 1970s, new canalswereconstructedfromEl DiquetoBarbacoasBay(Fig. 1), and since then, the suspended sedimentloadintoBarbacoas has reachedandimpactedtheEl Rosario Islands, a coral reef ecosystem southwestof Cartagena (Vernette, 1985), and is probablyresponsible for most of the observed coral reefmortality.AcknowledgementsThis study was done with support from the InstitutoColombiano para el Desarrollo de la Ciencia y Tecno-log aFranciscoJose deCaldas-COLCIENCIAS,andUniversidad EAFIT-Departamentode Geolog a.We thank Dr George Voulgaris and Dr L.R. Gardnerfortheir constructivecommentsonthemanuscript.WewouldliketothankthedirectoroftheMarineScience andTechnology Program-COLCIENCIAS,Leonor Botero, andEduardoZamudioof IDEAM,fortheirassistanceandsupport withtheriverdata.We also thank Dr John D. Millimanfor his sugges-tions on an earlier draft of this article.ReferencesAceituno, P., 1989. On the functioning of the Southern Oscillationin the South American Sector. Part II: upper-air circulation. J.Climate 2, 341355.Alekin,O.A., 1978. Watererosionoflandsurface.In:UNESCO(Ed.), World Water Balance and Water Resources of the Earth.Stud. Repts. Hydrol. vol. 25. UNESCO Press, Paris, pp. 645663.Alvarado, M., 1998. Cambios antropicos en la desembocadura delR o Magdalena y su efecto en la zona. In: IOCARIBE-EAFIT(Ed.), Memorias del Primer Seminario Nacional sobre ManejoIntegrado de Zonas Costeras. EAFIT Press, Medell n, pp. 36Alvarado, M., 1992. RioMagdalena (sector Calamar-Bocas deCeniza): caracterizacionhidrosedimentologica ycuna salinabasada en mediciones realizadas durante 20 campanas. Techni-cal Report, Ministerio de Obras Publicas y Transporte, Barran-quilla, Colombia.Barnett, T.P., 1991. The interaction of multiple time scales in thetropical climate system. J. Climate 4, 269283.Bassin, N.J., 1976. Sources and transport of suspended particulatesin the Caribbean Sea. Mar. Geol. 21, 289310.Baumgartner, A., Reichel, E., 1975. TheWorldWater Balance,Elsevier, Amsterdam.Bordyne, B., 1974. Neogene biostratigraphy and paleoenvironment,lower Magdalena basin. PhD thesis, Louisiana State University,Baton Rouge.COLCIENCIAS, FEN, 1990. Perl Ambiental Colombia. EscalaEditores, Bogota.Coleman, J.M., 1976. Deltas: processes of deposition and modelsJ.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 147for exploration, Burgess PublishingCompany, Minneapolis,MN.Coleman, J.M., Wright, L.D., 1975. Modern river deltas: variabilityof processes and sand bodies. In: Broussard, M.L. (Ed.). Deltas:Models for exploration, Houston Geological Society, Houston,TX, pp. 99149.Depetris, P.J., Paolini, J.E., 1991. Biogeochemical aspects of SouthAmericanRivers:TheParana andOrinoco.In:Degens,T.E.,Kempe, S., Richey, J.E. (Eds.). Biogeochemistry of major worldrivers, Wiley, Chichester, UK, pp. 105125.Depetris, P.J., Gaiero, D.M., 1998. Water-surface slope, totalsuspended sediment and particulateorganic carbon variabilityin the Parana River during extreme ooding. Naturwissenschaf-ten 85, 2628.Depetris, P.J., Kempe, S., Latif, M., Mook, W.G., 1996. ENSO-Controlled ooding in the Parana river (19041991). Naturwis-senschaften 83, 127129.Eisma, D., van der Gaast, S.J., Martin, J.M., Thomas, A.J., 1978.Suspendedmatter andbottomdepositsoftheOrinocodelta:turbidity, mineralogy, and elementary composition. NetherlandsJ. Sea Res. 12, 224251.Estoque, M.A., Luque, J., Chandeck-Monteza, M., Garc a, J., 1985.Effects of ElNino on Panama rainfalls.Geos. Int. 24, 355381.Gessler, R.D., 1995. PrecipitationvariabilityoverEcuadorasso-ciatedwithEl Nino/SouthernOscillation. MSthesis, Depart-ment of Geography, University of Florida.Glantz, M.H., 1997. Currents of Change, El Ninos Impact onClimate and Society, Cambridge University Press, Cambridge.Goniadzki, D., 1999. Hydrological warning and information systemfor the Plata basin. Publicacion del Instituto Nacional del Aguay del Ambiente, Buenos Aires.HIMAT, 1989. Regimen de la precipitacion en Colombia. Publica-cion del HIMAT, Bogota.IDEAM, 1995. Estad sticas Hidrologicas de Colombia (19901993), Diego Samper Ediciones, Bogota (Tomos III).IGAC, (1995). Suelos de Colombia. Canal Ram rez Antana Ltda,Bogota.Javelaud, O., 1987. La sedimentation du plateau continental de laColombie Caraibe, au cours du quaternarie terminal. PhDthesis,Universite Bordeaux I, Bordeaux, France.Kjerfve, B., 1981. Tides of the Caribbean Sea. J. Geophys. Res. 86,42434247.Kolla, V., Bufer, T., 1984. Seismic stratigraphy and sedimentationofMagdalenaFan, SouthernColombiabasin, CaribbeanSea.AAPG Bull. 68, 316332.Kousky, V.E., Kagano, M.T., Cavalcanti, I.F., 1984. A review of theSouthern Oscillation: oceanicatmospheric circulation changesand related rainfall anomalies. Tellus 36A, 490504.Lorin, J., Hernandez, C., Rouault, A., Bottagisio, J., 1973. Estudiosedimentologicodelaplataformacontinental entreBocasdeCeniza y Santa Marta. Technical Report, Puertos de Colombia,Barranquilla, Colombia.Mar n, R., 1992. Estad sticas sobre el recurso agua en Colombia.Publicacion del Ministerio de Agricultura, Instituto Colombianode Hidrolog a, Metereolog a y Adecuacion de Tierras, Bogota.Mart nez, J.O., Molina, A., 1992. Geomorfolog a y aspectoserosivos del litoral Caribe colombiano, sector Bocas deCeniza-ParqueTayrona. Technical Report, CIOH, Cartagena,Colombia.Meade, R.H., 1988. Movement andstorageofsediment inriversystems. In: Lerman, A., Meybeck, M. (Eds.), Physical andchemical weatheringingeochemical cycles, NorthAmericanTreaty Organization. Advanced Scientic Series C Math.Phys. Sci. Kluwer Academic Publishers, Dordrecht, The Nether-lands. vol. 51. pp. 165179.Meade, R.H., Nordin, C.F., Curtis, W.F., Costa-Rodr guez, F.M., doVale, C.M., Edmond, J.M., 1979. Sediment loads in the AmazonRiver. Nature 278, 161163.Meade, R.H., Dunne, T., Richey, J.E., Santos, U.M., Salati, E.,1985. Storage andremobilizationof sediment inthe lowerAmazon River of Brazil. Science 228, 488490.Mechoso, C.R., Perez-Iribarren, G., 1992. Streamow in the South-eastern South America and the southern oscillation. J. Climate5, 15351539.Meybeck, M., 1976. Total mineral transport by world rivers.Hydrol. Sci. Bull. 21, 265284.Meybeck, M., 1979. Concentration des eaux uviales en elementsmajeurset apportsensolutionauxoceans. Rev. Geol. Dyn.Geogr. Phys. 21, 215246.Meybeck, M., 1988. Howtoestablishanduseworldbudgetsofriverine materials. In: Lerman, A., Meybeck, M. (Eds.), Physicaland chemical weathering in geochemical cycles, North Amer-ican Treaty Organization. Advanced Scientic Series C Math.Phys. Sci. vol. 51. Kluwer Academic Publishers, Dordrecht, TheNetherlands, pp. 247272.Milliman, J.D., 1990. Fluvial sediment in coastal seas: ux and fate.Nature Resour. (UNESCO) 26, 1222.Milliman,J.D., Meade, R.H., 1983. World-wide deliveryof riversediment to the oceans. J. Geol. 91, 121.Milliman, J.D., Syvitski, J.P.M., 1992. Geomorphic/tectonic controlof sediment transport totheocean: theimportanceof smallmountainous rivers. J. Geol. 100, 525544.Milliman, J.D., Rutkowski, C., Meybeck, M., 1995. River dischargeto the sea: a global river index (GLORI). LOICZ Reports andStudies, Netherlands Institute for Sea Research, The Hague, TheNetherlands.NEDECO, 1973. Rio Magdalena and Canal del Dique SurveyProject. Technical Report, Netherlands EngineeringConsul-tants, The Hague, The Netherlands.NOAA, 1999. TahitiDarwin Sea Level Pressure Anomaly. http://ftp.ncep.noaa.gov/pub/cpc/wd52dg/data/indices.Oppenheim, A.V., Schafer, R.W., 1989. Discrete-Time SignalProcessing, Prentice-Hall, New York.Panofsky, H.A., Brier, G.W., 1968. Some applications of statisticsto meteorology, Pennsylvania State University Press, Pennsyl-vania, PA.Paolini, J., Ittekkot, V., 1990. Particulateorganicmatter intheOrinoco River. A pristine example. Naturwissenschaffen 77,8081.Potter,P.E.,1997.TheMesozoicandCenozoicpaleodrainageofSouth America: a natural history. J. S. Am. Earth Sci. 10, 331344.Poveda, G., 1994. Empirical orthogonal functions in the analysis ofJ.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 148the relationship between mean discharges in Colombia and thePacic and Atlantic Oceans surface temperatures (in Spanish).FinalProc. XVI Latin-AmericanCongress onHydraulicsandHydrology, Santiago, Chile, pp. 131144.Poveda, G., 1997. Atractores extranos (CAOS) en la hidro-climato-log a de Colombia? Rev. Acad. Colomb. Cienc. 21, 431444.Poveda, G., Mesa, O.J., 1993. Methodologies to predict the Colom-bian hydrology considering the ENSO event. Rev. Atmos. 16,2639.Poveda, G., Mesa, O.J., 1995. The relationship between ENSO andthe hydrology of tropical South America. The case of Colombia.Final Proc. Fifteenth Annual American Geophysical UnionHydrology Days, pp. 227236.Poveda, G., Mesa, O.J., 1996. The North Atlantic Oscillation and itsinuence on the hydro-climatology of Colombia (in Spanish).Final Proc. XVII Latin-American Congress on Hydraulics andHydrology, Santiago, Chile, pp. 343354.Poveda, G., Mesa, O.J., 1997. Feedbacks betweenhydrologicalprocesses intropical SouthAmerica andlarge-scale ocean-atmospheric phenomena. J. Climate 10, 26902702.Probst, J.L., Tardy, Y., 1987. Longrangestreamowandworldcontinental runoff uctuations since the beginning of thiscentury. J. Hydrol. 94, 289311.Probst, J.L., Tardy, Y., 1989. Global runoff uctuations during thelast 80 years in relation to world temperature change. Am. J. Sci.289, 267285.Pulwarty, R.S., Barry, R.G., Riehl, H., 1992. Annual and seasonalpatternsof rainfall variabilityover Venezuela. Erdkunke46,273289.Rasmusson, E.M., 1991. (Observational aspects of ENSOcycleteleconnections). In: Glantz, M., Katz, R.W., Nicholls, N.(Eds.). TeleconnectionsLinkingWorldwideClimateAnoma-lies, Scientic Basis and Societal Impacts, Cambridge Univer-sity Press, Cambridge, pp. 309343.Restrepo, J.D., Kjerfve, B., 2000. Waterdischargeandsedimentloadfromthewesternslopesof theColombianAndeswithfocus on Rio San Juan. J. Geol. 108, 1733.Richey, J.E., Meade, R.H., Salati, E., Devol, A.H., Nordin, C.F., dosSantos, U., 1986. Water discharge andsuspendedsedimentconcentrations in the Amazon River: 19821984. WaterResour. Res. 22, 756764.Richey, J.E., Nobre, C., Deser, C., 1989. Amazon river dischargeand climate variability: 1903 to 1985. Science 246, 101103.Rogers, J.C., 1988. Precipitation variability over the Caribbean andtropical Americasassociatedwiththesouthernoscillation. J.Climate 1, 172182.Ropelewski, C.F., Halpert, M.S., 1987. Global and regional scalesprecipitation associated with El Nino-southern oscillation. Mon.Weather Rev. 115, 16061626.Shepard, F.P., 1973. Sea oor off the Magdalena Delta and SantaMarta Area, Colombia. Geol. Soc. Am. Bull. 84, 19551972.Shepard, F.P., Dill, R.F., Heezen, B.C., 1968. Diapiric intrusions inforeset slopesedimentsoffMagdalenadelta, Colombia. Am.Assoc. Petrol. Geol. Bull. 52, 21972207.Snow, J.W., 1976. The climate of northern South America: Colom-bia. In: Schwerdtfeger, W. (Ed.). Climates of South and CentralAmerica, Elsevier, Amsterdam, pp. 358379.UNESCO, 1978. Atlas of World Water Balance, UNESCO Press,Paris.UNESCO, 1979. Discharge of selectedrivers of theworld, meanmonthlyandextreme discharges (19721975), Stud. Repts.Hydrol., vol. 3, UNESCO Press, Paris.UNESCO, 1985. Discharge of selected rivers of the world, vol. 3,UNESCO Press, Paris (no. 4).UNESCO, 1992. Discharge of selectedrivers of theworld, meanmonthlyandextreme discharges (19801984), Stud. Repts.Hydrol., vol. 3, UNESCO Press, Paris.Vernette, G., 1985. La plateforme continentale caraibe de Colombie(du debouche du Magdalena au Golfe de Morrosquillo). Impor-tance du diapirisme argileux sur la morphologie et la sedimenat-tion. PhD thesis, Universite Bordeaux I, Bordeaux, France.Vernette, G., Mauffret, A., Bobier, C., Briceno, L., Gayet, J., 1992.Mud diapirism, fan sedimentation and strike-slip faulting,Caribbean Colombian margin. Tectonophysics 202, 335349.Vorosmarty, C.J., Willmott,C.J., Choudhury, B.J., Schloss, A.L.,Stearns,T.K., Robeson, S.M., Dorman, T.J., 1996. Analyzingthedischargeregimeofalargetropicalriverthroughremotesensing, ground-based climatic data, and modelling. WaterResour. Res. 32, 31373150.Waylen, P.R., Caviedes, C.N., Quesada, M.E., 1996. InterannualvariabilityofmonthlyprecipitationinCostaRica. J. Climate9, 26062613.Welch, P.D., 1967. The use of Fast Fourier Transform for the esti-mation of power spectra: a method based on time averaging overshort, modiedperiodograms. IEEETrans. AudioElectroa-coust. 15, 7073.Wright, L.D., Coleman, J.M., 1973. Variations inmorphologyof major river deltasasfunctionsof oceanwaveandriverdischargeregimes. Bull. Am. Assoc. Petrol. Geol. 57, 370398.J.D. Restrepo, B. Kjerfve / Journal of Hydrology 235 (2000) 137149 149