JournalofMarineSystems31(2002)245–259 www.elsevier.com/locate/jmarsys Algerian Eddies lifetime can near 3 years I. Puillat, I. Taupier-Letage*, C. Millot LOB/CNRSUMR6535,AntennedeToulon,BP330,F-83507LaSeyneSurMer,France Received26June2000;accepted31May2001 Abstract TheAlgerianCurrent(AC)isunstableandgeneratesmesoscalemeandersandeddies.Onlyanticycloniceddiescandevelop f andreachdiametersover200kmwithverticalextentsdowntothebottom( 3000m).AlgerianEddies(AEs)firstpropagate eastwardalongtheAlgerianslopeatfewkilometersperday.InthevicinityoftheChannelofSardinia,afewAEsdetachfrom the Algerian slope and propagate along the Sardinian one. It was hypothesized that AEs then followed a counter-clockwise circuitintheeasternpartofthebasin.Maximumrecordedlifetimeswereknowntoexceed9months.Withintheframeworkof the1-yearEddiesandLeddiesInterdisciplinaryStudyoffAlgeria(ELISA)experiment(1997–1998),weexhaustivelytracked f two AEs, using mainly an 3-year time series of NOAA/AVHRR satellite images. We show that AEs lifetimes can near 3 years,exceeding33monthsatleast.Wealsoconfirmthelong-livedAEspreferentialcircuitintheeasternpartoftheAlgerian Basin,andspecify that itmay include several loops (at leastthree). D2002Elsevier ScienceB.V.All rights reserved. Keywords:AlgerianCurrent;Mesoscale;Eddies;Remotesensing;NOAA/AVHRR 1. Introduction between the crest of the meander and the continental slope. Transient (few days) surface cyclonic circula- The Algerian Basin is a key-area for the general tions (CC) are also frequently observed on satellite circulation of water masses in the Western Mediterra- images upstream from the meander crest, and have nean (see Millot, 1999 for a review). The Algerian been evidenced in situ by drifting buoys trajectories Current(AC),namedastheflowofModifiedAtlantic (Font et al., 1998a) and hydrology (Moran et al., Water (MAW), along the Algerian slope, is unstable. 2000).However,onlyanticycloniceddiescandevelop, The general structure of an AC instability, as far as and given in this paper are the generic name of MAW is concerned, is depicted from the available in Algerian Eddies (AEs). The generation of AEs is f f situ and laboratory experiments as rather complex observedfrom 0(cid:1)to 8(cid:1)E.AEstrajectoriesbegin (Obaton et al., 2000). Basically, instead of flowing with an alongslope-downstream propagation at few steadily alongslope, the AC meanders. Part of the kilometers per day, while within their embedding f MAW then recirculates inside the meander, so that a meander, the water typically flows at 50 km/day. surface anticyclonic eddy is created, embedded AEs diameters vary during their course between 50 f and 150 km up to 250 km. AEs lifetimes range from few weeks to several months. The maxima * Corresponding author. Tel.: +33-4-94-30-49-13; fax: +33-4- reported up to now are at least 6 months, with in situ 94-87-93-47. E-mailaddress:[email protected](I.Taupier-Letage). observations (Millot et al., 1997), and at least 9 0924-7963/02/$-seefrontmatterD2002ElsevierScienceB.V.Allrightsreserved. PII:S0924-7963(01)00056-2 246 I.Puillatetal./JournalofMarineSystems31(2002)245–259 Fig.1.TrajectoriesofAlgerianEddies:(a)96-1,(b)97-1,in1996(dottedline),1997(solidline),and1998(dashedline).Numbersreferto imagesonPlate1.Bathymetryinm. I.Puillatetal./JournalofMarineSystems31(2002)245–259 247 months, with satellite images (Taupier-Letage and 1998, involving the 1-year deployment of a nine- Millot, 1988). mooring network and four main cruises. ELISA’s The AEs alongslope-downstream propagation usu- aims were to study (1) the general circulation of ally ends in the Channel of Sardinia, where AEs the water masses, (2) the origin, structure and tra- dramatically interact with the bathymetry (see Fig. jectories of AEs, (3) the biological response associ- 1). Most of them (the less energetic?) collapse and ated to mesoscale dynamical phenomena, and (4) the release MAW that continues to flow through the biological consequences of the mesoscale activity on channel (Sammari et al., 1999). A few AEs per year the functioning of the Algerian Basin. It was thus of (the most energetic? called ‘‘events’’ by Millot et al., the utmost importance to know the history and 1997) follow the deep isobaths northward across the trajectory of the AEs sampled. The aim of this paper channel and detach from the Algerian slope, which is to present the results of the thorough NOAA/ impliesadeepstructure.Itisnowconfirmedthatthese AVHRR satellite images analysis from early 1996 AEs can be anticyclonic coherent structures down to to late 1998 that we conducted, in order to achieve f thebottom( 3000m,Ruizetal.,2002).Additional AEs tracking. The methodology is briefly described complexity arises, however, from recent observations inSection2.ThemainresultsaboutAEslifetimeand showing that this vertical coherence down to the preferential trajectory are detailed in Section 3, and bottom is not permanent (ELISA data, unpublished). we take this opportunity to further document and AEs detached from the Algerian slope then drift characterize AEs. In Section 4, we point out ques- northward along the Sardinian one, as confirmed by tions raised by these results. Vignudelli (1997), Iudicone et al. (1998), and Bouzi- f nac et al. (1998). Before reaching 40(cid:1)N, AEs detach from the Sardinian slope and propagate in 2. Method the open basin (where an AE was sampled for the first time by Katz, 1972). Some AEs come back Themostefficient(andcost-effective)waytotrack eventually close to the Algerian slope, where they AEs over long term is to use satellite data such as interact with their parent current. The pooling of images from NOAA/AVHRR thermal infrared chan- f several-month-long trajectories of different AEs on nels,astheyprovideaspatio-temporalcover( basin f different yearsledFudaetal.(2000) tohypothesizea and day) adequate to study such mesoscale phe- preferential counter-clockwise circuit in the eastern nomena1 (e.g. Arnone and La Violette, 1986; Le partoftheAlgerianBasin.Now,interactionsbetween Vourch et al., 1992). The complementary use of alti- theACandAEs,and/orbetweenAEsthemselves,can meter data (from ERS and TOPEX/POSEIDON, e.g. substantially modify this general scheme. This leads, Ayoub et al., 1998; Iudicone et al., 1998) helps to fill e.g. to offshoreward deviation of the AC path or of the gaps when cloud cover is important, and when upstream AEs, to the interruption of the AEs along- thermal gradients weaken, as the temperature differ- slope-downstreampropagation,uptoAEsdestruction ence between MAW and resident water reverses by (Taupier-Letage and Millot, 1988). springtime andfall. Therelativecomplexityof themesoscale phenom- Previousstudieshaveshownthecoherencebetween ena and the important variability associated with AEs AEs signatures on satellite images and contempora- explain why the Algerian Basin, a key-place for the neous in situ observations, definitely validating the circulation of the water masses in the Western useofsatellitesuperficialsignaturestotrackAEs(e.g. Mediterranean, is paradoxically poorly known for Millot,1991;Millotetal.,1997;BenzohraandMillot, such a close area. This dictated, within the frame- 1995a,b; Font et al., 1998a,b; Bouzinac et al., 1999; work of the European program MAST-3 MAss Sammari et al., 1999; Fuda et al., 2000). In order to Transfer and Ecosystem Response (MATER), the Eddies and Leddies Interdisciplinary Study off Alge- ria experiment (ELISA) in the eastern part of the 1 Ocean color satellites such as SeaWiFS are also extremely basin (Taupier-Letage et al., 2001). Intensive field efficient, however, image processing is more complex and work was carried out between July 1997 and July availabilityismoreproblematic. 248 I.Puillatetal./JournalofMarineSystems31(2002)245–259 Plate1.(1–35):NOAA/AVHRRimagestimeseriesshowingtheAEs96-1and97-1successivepositions.Contrastisspecificallyadjustedfor eachimagesothatgreyscalesarenottobecompared.Temperatureincreasesfromwhitetoblack. I.Puillatetal./JournalofMarineSystems31(2002)245–259 249 identifyAEsthroughouttimeandpapers,wewillusea spondtotheranktheAEsappeared.However,asitis notation close to that used for the Gulf Stream rings, unlikely that any exhaustive study of all contempora- associating the year the AE appeared with an incre- neous AEs will be performed, only those AEs picked mentingnumber.Inanidealsituation,numberscorre- upforspecificfeatureswillbenumbered. 250 I.Puillatetal./JournalofMarineSystems31(2002)245–259 Plate1.(1-35)(continued). WeusedNOAA/AVHRRSeaSurfaceTemperature Anomalies (SLA) maps from January 1998 to early (SST) daily composites from February 1996 to late 1999, as well as with the ELISA currentmeter time December 1998 from the ISIS DLR data base, and series on the nine moorings from mid July 1997 to NOAA/AVHRR channel 4 (brightness temperature) early July 1998. The variability of the AEs thermal images (three to six passes per day) from late June signatures,addedtotheseasonaltemperature reversal 1997tolateDecember1998.Asnoautomaticmethod of the water masses (see Section 3.6), prevents from is available so far to track AEs automatically (most keepingaconstantcolortotemperaturepalette.Hence, often,there isno thermal gradientassociated with the contrastwasadjustedspecificallyforeachimage,and AC,andtheAEsignatureisonlypartiallyvisible;see colors arenot tobe compared between images. details Taupier-Letage et al., 1998), position and Due (i) to the amount of satellite data (more than diameter estimates were deduced from the visual 2000 AVHRR images and 350 SLA objectives anal- analysis of AEs thermal signatures. AEs tracking yses)and(ii)tothedurationofthesurvey,werestrict f wasbackedwithTOPEX/POSEIDON/ERSSeaLevel illustration to 1 image/month (Plate 1), and pro- I.Puillatetal./JournalofMarineSystems31(2002)245–259 251 Plate1.(1-35)(continued). 252 I.Puillatetal./JournalofMarineSystems31(2002)245–259 Plate1.(1-35)(continued). vide monthly (at least) positions with trajectory sche- (Plate 1:1), 96-1 was already located offshore, which matics in Fig. 1. means that it was independent from an AC meander, and hence, most probably several months old. It has been tracked till December 1998 (Fig. 1a). As con- 3. Observations firmed by both altimetric (not shown) and thermal vanishing signatures, it was decaying, appearing as Two AEs, called 96-1 and 97-1, dominated the warmer water swirls South of Ibiza (Plate 1:35). This eddy field during the ELISA period and have been yields a lifetime exceeding 33 months. 97-1 has been thoroughly tracked during most of their lifetimes. trackedfromlateMarch1997,whenitwasafewdays OtherAEshavebeentemporarilytracked,forinstance old (Plate 1:14) in the AC, till the end of December when in the vicinity of, or interacting with, 96-1 and 1998 along the Algerian slope in the vicinity of the 97-1, but will not be detailed here. ChannelofSardinia(Plate1:35,Fig.1b).Thisyieldsa As discussed in Section 3.1, we think that the lifetime exceeding 21 months (tracking was interrup- image time series presented in Plate 1 is self-explan- ted only because no more data were available to us). atory, so that a tedious chronological description of Among the few other AEs we tracked, lifetimes AEs positions (Fig. 1) can be avoided. Now, we take ranged from 3 to 13 months. benefit from this time series to further document different phases and aspects of AEs’ life. 3.2. AC instability 3.1. Lifetimes OnPlate1:15,97-1displayedtheclassicalfeatures of a recent AC instability, with a well-developed We could definitely track the AE 96-1 from early surface cyclonic circulation (CC). About 1 month March1996(earlierpositionscouldnotbedetermined later, the cyclonic circulation had decayed (Plate withconfidence,sincecloudcoverwasimportantand 1:16). The continuity between the AC meander flow no altimetric data were available to us). At that time (leaving the coast near 2(cid:1)–2(cid:1)30VE and coming back I.Puillatetal./JournalofMarineSystems31(2002)245–259 253 Plate1.(21)(a)NOAA/AVHRRimage.(b)SeaLevelAnomaly(SLA)map(courtesyofP.-Y.LeTraon).SolidlinesindicatepositiveSLA, correspondingtoanticyclones. alongslopenear f3(cid:1)30VE)andtherecirculatingpart mostvisiblepartofanACinstability—whennotthe that defined 97-1 appeared extremely clear. About 3 only one—is usually the embedded eddy (e.g. 98-2 months later (Plate 1:19), the meander size had in Plate 1:29). Considering that open-sea AEs can increased. The meander no longer came back straight comebacksubsequentlyalongtheAlgerianslope,itis alongslopedownstreamfrom97-1,duetointeractions thus most often impossible, using thermal signatures with 96-1. Thermal signature continuity suggested onshorttermonly,todiscriminatebetweenembedded that part of the AC water was skirting 96-1 offshore AEs and open-sea ones. f before coming back shoreward near 7(cid:1)E, that is downstream from 96-1, in a paddle-wheel-like effect. 3.3. Trajectories AnanalogoussituationhappenedonPlate1:26,when partofthemeanderflowembedding97-1wasrelayed Pooling all observed trajectories yields what we northward by 96-1. regard as the AEs preferential path. AE generation Individualizing the AC meander water from the occursalongtheAlgerianslope.Inmostcases,itisin basin resident one, however, is generally not easy, the western part of the basin, such as 97-1 near 0(cid:1) especially as mixing on edges can weaken thermal (Plate 1:14), but it can also be further east, as in the difference. Moreover, as water is upwelled where the case of 98-2 generated early May 1998 near 5(cid:1)– meander leaves the slope, cooler inside-spiraling iso- 5(cid:1)30VE. Then, AEs propagate eastward alongslope f thermsclearlymaterializetherecirculation,sothatthe (3–5 km/day), as shown by 97-1 positions on 1(cid:1)E 254 I.Puillatetal./JournalofMarineSystems31(2002)245–259 Plate1.(1-35)(continued). f inMay1997(Plate1:15)andon 3(cid:1)Eabout1month This was the case of the 3-month-apart Plate 1:26 and a half later (Plate 1:16). AEs can also remain and 29: the two AEs aligned on a North–South axis withoutpropagatingformonths,as96-1betweenearly were 96-1 and 97-1 in the former, and 97-1 and 98-2 July and late October 1997 (Plate 1:16–20; Fig. 1a), in the latter. and97-1betweenearlyJulyandearlyDecember1997 Now,duetointeractionswiththeACorwithother (Plate 1:17–22; Fig. 1b). The use of the SLA maps AEs, departures from this general scheme are (Plate1:21-b)duringthisperiodenabledinmostcases observed, for instance early detachment from the to track AEs even during cloudy periods. At the Algerian slope. This was the case of the AE 98-1 entranceoftheChannelofSardinia,afewAEsdetach (not shown) sampled during the ALGERS-98 cam- from the Algerian slope. Then, they propagate along paign (Ruiz et al., 2002), which described a shorter the Sardinian slope (96-1 on Plate 1:23–25, 97-1 on counter-clockwise circuit in the western part of the f Plate1:27–29).Beforereaching 40(cid:1)N,AEsdetach basin.As well, theloopingschemecan stop:96-1 for from the Sardinian slope and drift in the open basin example went drifting to the northwest and reached (96-1onPlate1:25–26,97-1onPlate1:28–29).From the Balearic slope by summer 1998 (Plate 1:26), f Fudaetal.(2000)and these 3-year observations, it where its trajectory seemed then closely constrained seems that AEs preferentially come back to the by the bathymetry. f Algerian slope, and preferentially near 5–6(cid:1)E (Fig. 1). AEs thus describe a counter-clockwise 3.4. AEs diameter and drift speed circuit in the eastern part of the basin. This circuit possibly includes several loops: 96-1 described at AEs diameter varies during their course, as shown least three loops, 97-1 described at least one.2 Due to by 97-1 between Plate 1:16 (f110 km), 19 (f170 f f such a path, different eddy fields can appear similar. km), 25 ( 110 km), and 26 ( 150 km). The entrance of the Channel of Sardinia seems to be a criticalplaceinthisregard.Thisregionisabarrierfor 2 We have indications that 97-1might have beenundertaking largeanddeepAEs,whichoftenremainthereforafew anotherloop,asisolatedimagesshowacoherentAEsignaturenear 7(cid:1)30VE on December27, 1998, and near f8(cid:1)15VE on February months (Fig. 1a). Interactions with the bathymetry 21,1999. (northward-leading) and with the AC (eastward-lead-
Description: