Seismic stratigraphy, buried beach ridges and contourite ... · Seismic stratigraphy, buried beach...

Seismic stratigraphy, buried beach ridges and contourite drifts:the Late Quaternary history of the closed Lago Cardiel basin,Argentina (49�S)ADRIAN GILLI1*, FLAVIO S. ANSELMETTI*, DANIEL ARIZTEGUI� , MILAN BERES� ,JUDITH A. MCKENZIE* and VERA MARKGRAF�*Geological Institute, Swiss Federal Institute of Technology ETHZ, Sonneggstrasse 5, 8092 Zurich,Switzerland�Institute F.A. Forel & Department of Geology and Paleontology, University of Geneva, Rue desMaraichers 13, 1205 Geneva, Switzerland�Institute of Arctic and Alpine Research, University of Colorado, Bolder, CO 80309–0450, USA

ABSTRACT

The results of a seismic stratigraphic analysis of a closed lake basin, Lago

Cardiel, in southernmost South America are reported. Very few high-

resolution, continental records spanning the Late Quaternary have been

obtained from this region. Seismic sequence stratigraphic analysis allows a

reconstruction of lake level variations. Two major hiatuses of unknown age

occurred during the early evolution of the basin with the deposition of an

alluvial fan in a restricted area in the intervening time period. Following the

development of a relatively shallow lake during the late Pleistocene and a

short desiccation pulse around 11 220 14C yr BP, a transgression of over 135 m

occurred at the beginning of the Holocene. The transgression was associated

with the formation of beach ridges preserved in the lake stratigraphy on the

floor of the modern Lago Cardiel at four different elevations. The preservation

of largely unreworked beach ridges indicates a stepwise rise in the lake level.

There is no seismic evidence of a major lowering of the lake below modern

level during the entire Holocene. Deposition since the mid-Holocene is marked

by strong lateral differences in sediment accumulation with a depocentre

slightly to the north of the basin midpoint and a pronounced mounded

distribution. Seismic reflection geometries, as well as sedimentological

characteristics indicate a lacustrine contourite drift covering an area of 80–

100 km2. As Lago Cardiel is under the influence of westerly winds, these most

likely drove lake circulation. The identification of drowned beach ridges and

of contourite drifts illustrates that high-resolution seismic stratigraphy is not

only a powerful tool in reconstructing past lake level elevations for closed lake

basins, but it can also provide information about the rate of lake level changes

and the presence and strength of lake currents.

Keywords Contourite drifts, high-resolution seismic, lake level fluctuations,Patagonia, shorelines, wind-driven currents.

INTRODUCTION

Numerous studies (e.g. Abbott et al., 2000; Vers-churen et al., 2000; Baker et al., 2001) haveutilized lake levels in closed basins for thereconstruction of past regional hydroclimaticconditions. Lake levels in closed basins are

1Present address: Department of Geological Sciences,University of Florida, Gainesville, FL 32611, USA(E-mail: [email protected])

Sedimentology (2005) 52, 1–23 doi: 10.1111/j.1365-3091.2004.00677.x

� 2004 International Association of Sedimentologists 1

sensitive and respond quickly to changes in theregional precipitation/evaporation ratio, rising intimes of wetter climatic conditions and droppingduring drier periods. Therefore, the establishmentof long, complete and well-dated lake levelrecords is crucial for describing past hydrologicconditions and provides a major input to palaeo-climate reconstructions. A broad variety of pal-aeolimnological techniques can be applied toreconstruct past lake level fluctuations on differ-ent time scales (e.g. Richardson, 1969; Street-Perrott & Harrison, 1985; Harrison & Digerfeldt,1993; Ariztegui et al., 2000). For the recent pastback to a maximum of a few hundred years ago,instrumental records (e.g. gauge measurements orsatellite images) or historical observations areavailable to assess lake levels and lake shorepositions (e.g. Piovano et al., 2002). On geologicaltimescales, information about lake level elevationcan be extracted from geomorphologic featuresand/or from stratigraphic and lithologic studies oflake sediments. Geomorphologic evidence suchas shorelines, delta complexes, spits and wave-cut cliffs can be exposed in outcrops above themodern lake level. When calibrated with a robustchronology, these relatively easily accessiblesedimentary deposits can be used to reconstructthe timing and extent of past lake level high-stands (e.g. Stine & Stine, 1990; Thompson, 1992;Adams & Wesnousky, 1998). However, the iden-tification of lake level lowstands is essential toestablish a complete record of past lake levelfluctuations. Detailed information about subaqu-eous geology can be extracted from sedimentarycores and seismic reflection profiles. Estimationsof water depth can be made from sediment coresby exploring the sensitive relationship betweenwater depth and general sediment composition(Harrison & Digerfeldt, 1993). By using a combi-nation of several measured proxies, a coherentreconstruction of past lake levels can be made.Seismic stratigraphy, on the other hand, uses thegeometry of seismic sequences, onlap relation-ships and the occurrence of erosional surfacesand incised channels to pinpoint the past eleva-tion of the lake level (e.g. Scholz & Rosendahl,1988; Valero-Garces et al., 1996; Seltzer et al.,1998; Ariztegui et al., 2000; Johnson et al., 2000).Lago Cardiel is a closed lake basin located in

the southernmost part of South America, whichis, with the exception of Antarctica, the onlymajor continental landmass reaching to such highsouthern latitudes. Consequently, palaeoenviron-mental archives from this unique geographicallocation are crucial for the understanding of

climate evolution in the Southern Hemispherein that they provide a link between low latitudeclimate archives and ice core records from Ant-arctica. Until recently, few multiproxy continen-tal archives spanning the Late Quaternary havebeen obtained from this region and therefore thestudy of the lacustrine record of Lago Cardielprovides an important constraint on the LateQuaternary palaeoenvironmental evolution ofsouthern South America.

STUDY AREA

Lago Cardiel is an endorheic basin lying on thePatagonian Plateau (Argentina) between the An-dean Cordillera and the Atlantic coast at alatitude of 49�S (Fig. 1A). The heart-shaped basinhas a diameter of approximately 20 km with amodern lake area of about 370 km2 and a maxi-mum water depth of 76 m (Fig. 1C). The sur-rounding terrain consists of deformedCretaceous/Tertiary shales and flat-lying Tertiaryvolcanics (Feruglio, 1950; Heinsheimer, 1959;Ramos, 1982, 1989). Glacial geomorphologicalstudies have clearly documented that the entirecatchment of Lago Cardiel was unaffected byAndean glaciers during the last glaciation(Rabassa & Clapperton, 1990; Wenzens, 2002,2004) and therefore the lake is totally disconnec-ted from any glacial or melt water input. The totalcatchment area is about 4500 km2 with the RıoCardiel as the principal, perennial inflowing river(Fig. 1B).The lake and its catchment area lie in the

orographic rain shadow of the Patagonian Andes(see Hoffmann, 1975; McCulloch et al., 2000)with a mean annual precipitation of about150 mm near the lake increasing to a maximumof 500 mm in the catchment area to the west andnorth-west (1300–1700 m elevation), where theRıo Cardiel originates. As a result of low precipi-tation near the lake area only a few smaller,permanent and ephemeral streams originatearound the modern lake. The climate of this areais dominated by persistent and strong westerlywinds (Fig. 1A, inset), which are especially pro-nounced during the austral summer (October toFebruary; Prohaska, 1976). Mean monthly tem-peratures recorded in Gobernador Gregores (forlocation see Fig. 1B), the nearest climate-stationto Lago Cardiel and c. 80 km away, vary between0Æ8 �C in June and 14Æ1 �C in December. The meanannual temperature is about 8Æ5 �C (http://www.r-hydronet.sr.unh.edu).

2 A. Gilli et al.

� 2004 International Association of Sedimentologists, Sedimentology, 52, 1–23

In outcrop studies, Galloway et al. (1988) andStine & Stine (1990) used bulk radiocarbonmethods to date palaeo-shoreline deposits toassess Lago Cardiel’s past lake level highstands.Two lake level highstands have been proposedfor the late Pleistocene that reached up to+ 75 m above present level, but the timing andthe exact extent of these highstands are poorlyconstrained. The early Holocene is dominatedby a major and well-dated lake level highstandthat culminated around 9500 14C yr BP at anelevation of + 55 m (Stine & Stine, 1990). Asubstantial mid-Holocene highstand (around5130 14C yr BP) at + 21Æ5 m and four minorlake level highstands (+ 10 to + 12 m) in the last2000 years have been further described by Stine

& Stine (1990). Geochemical and palaeoecologi-cal analysis of two cores taken with a square-rod Livingstone piston corer near the modernshore confirm the high lake level throughout theearly Holocene as well as fluctuating lake levelconditions during late Holocene times (Markgrafet al., 2003). The lowermost spill point of theclosed depression in which Lago Cardiel sits isa gravel plain in the north-east, which probablyrepresents an ancient drainage outlet of LagoCardiel into the Rıo Chico (Fig. 1B). Its heightof more than 250 m above the modern lakelevel and the occurrence of varnished gravelssuggest that the lake did not drain by thisspill point in the recent past (Galloway et al.,1988).

50°

54°

66¡72¡

~ 300 km

Lago Cardiel49° S / 71°W Gdor.Gregores

Río Gallegos

NNE

E

SE

S

SW

W

NW

10203040

A

49°00'S

48°55'S

48°50'S

71°05'W71°10'W71°15'W71°20'W

Río Card ie l

RíoBayo

0 5km

N

40 m

50 m

60 m

30 m

20 m

10 m

70 m

Bathymetry

C

49°00'S

48°55'S

48°50'S

71°05'W71°10'W71°15'W71°20'W

Seismic network & figure locations

Río Card ie l

0 5km

N

Fig. 8Fig. 11

Fig. 7

Fig.

2

Fig. 9

Fig. 3C

Fig. 3A

Fig. 3B

DR

íoBayo

Bahía Pescaderia

20 km

GobernadorGregores

Lago Cardiel

LagoStrobel

Río Chico72°W 71°W

49°SGravel plain

700

1000

1500

400

B

Fig. 1. (A) Map of the southern tip of South America showing the location of Lago Cardiel. The relative frequency(%) of the annual distribution of the wind direction (1951–60) for Gobernador Gregores is given in the inset(Prohaska, 1976). Thirty-two per cent of the time are calm conditions and do not contribute to the graph. (B)Simplified topographical map of the catchment area of Lago Cardiel. A gravel plain north-east of the lake mayindicate an ancient spillover into the Rıo Chico. (C) Bathymetric map of Lago Cardiel. Modern maximum water depthis 76 m (1999). (D) Seismic grid with the location of the seismic profiles and the corresponding figure numbers.

Seismic stratigraphy of Lago Cardiel, Argentina 3


TWT (ms)

80604030

Sublake level depth (m)

50

2000

m

NE

100

120

140

160

70 90 110

SW

TWT (ms)

80604030

Sublake level depth (m)

50

2000

m

VI

V

NE

100

120

140

160

70 90 110

SW

IIIIII

IV

V?

?

Fig.2.Uninterpreted

(above)and

interpreted

(below)su

mmary

seismic

profile

(SW

–NE)acquired

with

aboomersystem

coveringall

defined

seismic

se-

quences.

Romannumbers

representtheseismic

sequencesandsm

allarrowsindicate

reflectionterm

inations.

Thedepth

isgivenin

milliseconds(m

s)tw

o-w

ay

traveltime(TW

T)andconsequentlyconvertedto

sublakeleveldepth

(m)basedonaP-w

avevelocityof1450ms)

1forwaterandsedim

ent.

4 A. Gilli et al.


METHODS

Twelve high-resolution seismic lines with a com-bined length of over 140 km were obtained inOctober/November 1999 using a 3Æ5 kHz single-channel profiling system (Fig. 1D). To increase theseismic penetration, three additional lines of� 40 km were acquired with a stronger 1–12 kHzGEOPULSETM boomer source. In a second seismicfield campaign inMarch 2002, an additional 65 kmof 12-channel seismic profiles acquired with a1 in3 airgun system (400 Hz dominant frequency)were collected to aid the interpretation of thedeeper seismic sequences. All the seismic profileswere digitally recorded in SEG-Y format using anon-differential global positioning system (GPS).The seismic data were processed in the programspw

TM. Processing included digital subtraction ofconstant shallow noise, water bottom mute, band-pass filtering, and an automatic gain control with awindow length of 180 ms (plus CDP-sorting, NMOcorrection, stack and spikingdeconvolution for theairgun data). The time-depth conversion is basedon a P-wave velocity of 1450 m s)1 in water andsediment meaning that 10 milliseconds (ms) intwo-way travel-time (TWT) are equal to 7Æ25 m indepth. The seismic interpretation and the horizonmanagement were done using kingdom suite

TM

software. Visualizations of the basin-wide sedi-mentary infill (topography and isopach maps) andvolume calculations of the interpreted sequenceswere made with gocad

TM 2.0.3.To verify and to date the findings of the seismic

sequence analysis, 10 piston cores with a lengthof up to 10 m were recovered using the ETH-Kullenberg coring system. Preliminary results ofthe cores and their correlation with the seismicprofiles and the available dates have been previ-ously published (Gilli et al., 2001). All presentedradiocarbon ages are uncalibrated. To betterunderstand the distribution of the suspendedparticles in the modern lake system, a satelliteimage (Landsat 7) was analysed to obtain asnapshot of the lake circulation.

DESCRIPTION OF SEISMICSTRATIGRAPHIC SEQUENCES

An acoustic penetration of over 50 m with the3Æ5 kHz seismic system and even deeper withthe stronger boomer and airgun sources allowedthe imaging and discrimination of six majorseismic sequences, labelled in roman numbers(I-VI from youngest to oldest; Fig. 2; Gilli et al.,

2001). The seismic data show no indication offaults or any other neotectonic activity in theyoungest five sequences.

Sequence VI

The oldest seismic-stratigraphic sequence,Sequence VI, is present in all seismic profilesand represents acoustic basement. The seismicpenetration with the 3Æ5 kHz and the strongerboomer device was limited to the uppermost 10 mand 15–20 m of this sequence respectively. Thereflections are broad-spaced and folded on a kmscale. They exhibit an irregular internal deforma-tion on a smaller scale (10 s of metres, Fig. 2).Sequence VI is overlain either by Sequence V, IVor III with an angular unconformity. The top ofSequence VI is an irregular surface and cannot becontinuously traced throughout the basin due tothe limited penetration of the seismic signal in thearea where Sequence V is present (e.g. Fig. 2, leftside). Locally channels are incised into SequenceVI to a sublake level depth of between 65 and110 ms along the western slope of the basin(Fig. 3A–C). The channels are V-shaped with adepth of incision of 6–13 ms (� 4Æ5 and 9Æ5 m) anda maximum width of approximately 200–250 m.

Sequence V

Sequence V is only observed in the seismicprofiles along the western margin of the lake(Fig. 4A). It forms an isolated body that clearlyoverlies Sequence VI (Figs 2 and 5). The base ofSequence V is only partly imaged in the 3Æ5 kHzseismic data, but the deeper penetrating airgunprofiles permit the extent (Fig. 4A), the topogra-phy prior to Sequence V deposition (Fig. 4B) andthe thickness of this sequence (Fig. 4C) to betentatively mapped. The basin topography priorto the deposition of Sequence V had a prominentdepression six km north-east of the point wherethe Rıo Cardiel enters the lake with a maximumsublake level depth of 154 ms TWT (Fig. 4B).This is the deepest area of the upper boundary ofSequence VI over the entire basin. In the samearea, Sequence V reaches a maximum verticalthickness of 46 ms (� 33Æ4 m). In the seismicallyimaged part of Sequence V acquired with theboomer system, a few easterly dipping prominentreflections divide the otherwise almost transpar-ent or gently eastward dipping layered sequenceinto smaller prograding units (Fig. 5A). In aseismic section collected perpendicular to thisW–E oriented prograding structure (Fig. 5B), all



the reflections are subhorizontal. The sequence isagain divided by a set of high-amplitude reflec-tions and their subhorizontal orientation supportsthe prograding structure for Sequence V. A seis-mic profile acquired with an airgun system(Fig. 5C) shows the character of Sequence V inthe northern isopach thick. The lower part of thesequence here is made up of an almost transparentfacies, whereas in the upper part a few localizedreflections are imaged. The prograding units ofSequence V are not resolved on this profile,possibly due to the lower resolution of the airgunsystem. The upper boundary of this sequence isvery similar to the previously described top ofSequence VI with an irregular and in placesstrongly incised surface (e.g. Fig. 5B and C).

Sequence IV

The extent of Sequence IV is limited to the centralpart of the basin below a sublake level depth ofapproximately 110 ms (Figs 2 and 6D). Owing toa pair of depressions in the palaeotopographyformed by the underlying sequences (VI and V;Fig. 6H), Sequence IV has two areas of anomalousthickness (Fig. 6D). The maximum sequencethickness is 26 ms and 27 ms TWT in thenorthern and southern depocentres respectively.A similar internal seismic pattern is seen in bothdepocentres. Sequence IV can be divided into twodistinct seismic facies. A lower part below� 130 ms TWT consists of medium to low-amplitude reflections, which are sometimes dis-continuous. The upper part is mostly transparentin the 3Æ5 kHz seismic profiles with only a fewdistinct reflections. In the boomer survey,the upper part of Sequence IV consists oflow-amplitude reflections (Figs 2, 5A and B).

With the exception of a few reflections thatinternally onlap older Sequence IV sediments ata depth of around 130 ms TWT (see circles inFig. 5A and B), all reflections laterally onlapSequence VI, or, where present, Sequence V.The geometry of the marginal lateral onlapschanges vertically within Sequence IV. Thereflections have a straight and horizontal onlapbelow � 135 ms TWT, whereas the onlaps aboveare slightly bent upward. As a consequence, thevertical difference of a reflection between itslateral onlap and its elevation in the basinamounts to a few metres (illustrated by twodotted reflections in Fig. 5A). The top of Se-quence IV is marked by several irregular, some-times chaotic high-amplitude reflections. Ingeneral, this upper sequence boundary is morechaotic and discontinuous towards the lake mar-gin compared to in the central lake area (Fig. 7),and it consistently onlaps Sequence VI or Vbetween 108 ms and 111 ms TWT.A restricted sedimentary body is present in the

seismic profiles of the north-eastern bay (BahıaPescaderia, see Fig. 1D for location) below asublake level depth of 79 ms TWT. This body isonly partly imaged revealing a low-amplitude totransparent seismic facies and a minimum thick-ness of 13 ms TWT (� 9Æ4 m). Its occurrencebelow Sequence III indicates a similar stratigraph-ic position as Sequence IV in the main basin, eventhough this restricted body was deposited up to30 ms TWT shallower than Sequence IV in themain basin.

Sequence III

Sequence III drapes the existing palaeotopo-graphy formed by the previously deposited

200 m 200 m 200 m

5 m

s ~

3.6

m

5 m

s ~

3.6

m

65 ms 47.1 m

69 ms 50 m

110 ms 79.8 m

5 m

s ~

3.6

m

VI

VI

VI

III III III

A CB

~ ~

Fig. 3. Three examples of incisedchannels at the western side of thebasin (see Fig. 1D for exact loca-tions).

6 A. Gilli et al.


sequences. It can be traced laterally over theentire basin up to the modern shoreline (Fig. 8).The only exception is the steep northern slope

(Fig. 1C), where the youngest three sequences areabsent. The maximum thickness of this sequenceis almost 10 ms (about 7 m) in the south-westerncentral basin. It thins towards the modern shore-line (Fig. 6C). The seismic facies of this sequenceconsists mainly of faint internal reflections. Theseare weaker, or absent in the lower quarter of thesequence (Fig. 7). In contrast, high-amplitudereflections are locally present in the upper partof Sequence III in the western central area (Fig. 2,on the left side). On the slope, Sequence III ismostly transparent and therefore onlaps of low-amplitude reflections at the base of Sequence IIIare rare. A prominent concordant double reflec-tion caps the sequence.

Sequence II

Sequence II conformably overlies Sequence IIIover the entire modern lake basin. The thicknessof this sequence varies between 13 ms TWT (i.e.less than 10 m) at the depocentre and less than afew metres in the shallow area near the shore(Fig. 2). An isopach map of Sequence II (Fig. 6B)shows a mounded sediment distribution patternwith a depocentre slightly north of the basin’smidpoint. This change in the depositional patterninitiated after the lowest third of this sequence.The mid and upper portion of Sequence IIconsists of continuous, narrow-spaced and well-defined low-amplitude reflections, whereas in thelower third, the reflections are only faint orsometimes even absent. The reflections of theupper two-thirds of Sequence II, as in SequenceIII, show increased amplitudes in the westerncentral area (Fig. 2, left side). In this area, thetopmost reflections terminate at the upper se-quence boundary possibly forming a pseudo-toplap, where the laminations become too thinto be seismically imaged (Fig. 7). An erosionaltruncation at the top of Sequence III in this areacannot be excluded. As a result of the preferentialaccumulation of sediments in the centre of thebasin, the deepest part of the lake moved to thewest (Fig. 6E).

Sequence I

The sediment distribution pattern of Sequence Imimics and amplifies the pattern observed inSequence II. The mounded depositional pattern isstrongly enhanced, but the location of the depo-centre of Sequence II and I are identical (seeFig. 6A and B). The thickness of Sequence I variesstrongly from almost 26 ms (about 18Æ8 m) near

Fig. 4. Lateral extent (A), basal topography (B) and thethickness (C) of Sequence V. The location of seismicprofiles shown in Fig. 5A–C are indicated in (A). Thevalues for the topography and thickness maps are in msTWT.



the depocentre to a very condensed or evenabsent section in more distal areas. A prominentmoat at the foot of the steep northern slopeterminates the northern extent of Sequence I witha convex-upward geometry (Fig. 9). Because ofthe low sequence thickness associated with themoat, it is easily traced in the isopach map(Fig. 6A) extending along the northern shore fromthe west to the east even penetrating partly intothe north-eastern bay (Bahıa Pescaderia). Theseismic facies resemble Sequence II, with regular,thin-spaced and continuous low-amplitude

reflections. Again like Sequence II, some reflec-tions increase their amplitude towards the moatand show a wavy appearance (Fig. 9). Within themoat, local scouring causes erosional truncationsand the infill of small depressions is evident fromthe pattern of reflection onlaps. In the area westand south-west of the depocentre, some reflec-tions appear to downlap onto the I/II sequenceboundary (Fig. 7), but it remains unclear whetherthis is real or apparent (i.e. pseudo-downlap;analogous to the pseudo-toplap at the topof Sequence II). However, this indicates a

500 m

S N

B120

TW

T (

ms)

150

140

130

VI

V

IV

?

500 m

W E

AIII

90

Sub

lake

leve

l dep

th (

m)

100

110

120

TW

T (

ms)

150

140

130

500 m

W E

A

VI

V

IV

V

? ??

90

100

110

500 m

S N

BIII

120

TW

T (

ms)

150

140

130

500 m

W E

C100

110

160

170

500 m

W E

C

80

Sub

lake

leve

l dep

th (

m)

90

100

110

120VI

V

IV

IIIII

airgunbuble pulse

Fig. 5. Uninterpreted (left) and interpreted (right) seismic profiles of Sequence V (for location see Fig. 4A). (A)Boomer profile parallel and (B) perpendicular to the progradation direction of this sequence. Note that in (A) thechange in the bending upward of the onlap within Sequence IV is displayed with two examples (dotted lines). Aninternal onlap of Sequence IV occurred around 130 ms TWT (ellipse in A and B) and coincides with a general changein the seismic facies of this sequence. (C) Airgun profile through the deepest and thickest part of Sequence V.

8 A. Gilli et al.


Fig. 6. Isopach maps for (A) Se-quence I (B) Sequence II (C) Se-quence III, and (D) Sequence IV inms TWT. Contour interval is 5 msTWT (equals � 3Æ6 m). The volumeof each sequence is given in thelower left corner of each map.Bathymetric maps for (E) the top ofSequence II (F) the top of SequenceIII (G) the base of Sequence III, and(H) the base of Sequence IV in msTWT. Contour interval is 10 msTWT (equals � 7Æ25 m). Cross in thesouth-east corner of each map indi-cates 49�00¢S/71�05¢W.



progradation of the Sequence I mounded sedi-ment body towards the west and south-west(Figs 2 and 7). Preferential deposition in thecentre of the lake contributed to the increasinglyasymmetric bottom morphology, with the deepestpart of the lake basin now in the west (Fig. 1C).Near the modern, gently dipping, shore at theeastern side of the lake, Sequence I, as well asSequence II, are strongly eroded down to a waterdepth of approximately � 10 ms TWT (� 7 m)showing a saw-tooth-like morphology (Fig. 8).Subsequently in greater water depth, between10 ms to 30 ms TWT, a small bulge of accumu-lated sediment is present. The well-definedreflections in this accumulation zone terminateat the upper and lower end of the bulge (Fig. 8).

BEACH RIDGES

Spectacular sets of exposed beach ridges encircleLago Cardiel above the modern lake level. Theyare particularly well developed along the gentlysloping east coast of the lake and preserveinformation concerning past lake level high-

stands. The identification of subaqueous beachridges in seismic profiles can add to the record ofpast lake level elevations and therefore act as anindependent constraint on former lake levellowstands complementing arguments based onthe geometrical relationship of seismic sequences(i.e. onlaps of seismic sequences and/or thefluvial incisions into older seismic sequences)and the sedimentological analysis of cores.

Submerged beach ridge description

Numerous beach ridges can be recognized on theCardiel seismic data. They are all positioned atthe top of Sequence VI, or where it is present, thetop of Sequence V, and they formed during thedeposition of sequences IV and III. For example, aridge feature occurs consistently at the samestratigraphical position above the lateral onlapof the IV/III sequence boundary (Fig. 10), and thisis interpreted as a beach ridge or ridges thatbecame buried during a transgression at thebeginning of Sequence III. The beach ridges varyin size, shape and internal structure. The crestsrange between a sublake level depth of 102 and

V III

II

I

IV

500 m

500 m

70

Sub

lake

leve

l dep

th (

m)

80

W E

90

100

120

TW

T (

ms)

110

90

100

120

TW

T (

ms)

110

75

85

70

Sub

lake

leve

l dep

th (

m)

80

75

85

CAR 99-

7P

Fig. 7. Unconformity (toplap of Sequence II/downlap of Sequence I) at the Sequence boundary II/I in the north-western part of the lake displayed in an uninterpreted (above) and interpreted (below) seismic section. The un-conformity is possibly of pseudo-character, as the layers become too thin to be seismically imaged. Sequence I andpart of Sequence II prograded towards the west.

10 A. Gilli et al.


110 ms TWT (74–80 m). Some of the beach ridgeshave a composite appearance showing up to threeculminations presumably representing severalphases of sediment deposition (Fig. 10, seismicline 8E and 11S). Their width varies between� 100 and � 300 m, and the composite beachridges can be up to 700 m wide. The general lackof imaged internal structures and the absence ofan obvious basal surface prevent an exact deter-mination of their heights. Nevertheless, estima-tions for the vertical height range between � 1 and� 4 m. The shape of the beach ridges is clearlyasymmetrical showing a gently dipping lakewardside with a slope angle of normally less than 1�and only in rare cases exceeding an angle of 3�.The landward side of the beach ridge is generallysteeper than the lakeward side reaching a maxi-mum slope angle of 6�. Some of the beach ridgesdo not show any inclined landward side (Fig. 10,e.g. seismic line 14S). In one case, the depression

on the landward side of the beach ridge is filledby sediment (Fig. 10, seismic line 11SW). Twobeach ridges show an almost flat ridge crest(Fig. 10, seismic line 7NW and 14S). Only thecomposite beach ridges show internal structuresrelated to possible ridge amalgamation.Beach ridges are not restricted to the lateral

onlap of the VI/III sequence boundary. On thegently dipping east slope of the Cardiel basin, aflight of well-preserved beach ridges with ridgesat different sublake level depths is present(Fig. 11). The analysis of the entire seismicdataset revealed the multiple presence ofdrowned beach ridges at a sublake level depthsof around 135 ms/98 m, 120 ms/87 m, 107 ms/77 m (above the lateral onlap of Sequence bound-ary IV/III), 90 ms/65 m, 73 ms/53 m, 57 ms/41 mand 28 ms/20 m. The geographical distribution ofeach beach ridge within the Cardiel basin ismapped out in Fig. 12.

multiple reflectionsVIIII

III

10

TW

T (

ms)

40

30

20

500 m

W

0

Sub

lake

leve

l dep

th (

m)10

20

E

0

50

30

10

TW

T (

ms)

40

30

20

500 m

W

0

Sub

lake

leve

l dep

th (

m)10

20

E

0

50

30

? beach ridgesmay associate with other beach ridge formation around ~ 28 ms TWT

wave erosion

Fig. 8. Uninterpreted (above) andinterpreted (below) seismic profileof the gently dipping slope at theeast shore of the basin. Sequence IIIcan be traced up to the modernshoreline, whereas the youngest twosequences are thinned owing towave action. The reworked sedi-ment is consequently deposited in adepth of 10–30 ms TWT.



Beach ridge formation

The dimensions of the buried beach ridges aresimilar in height, but considerably wider than theexposed beach ridges found above themodern lakelevel that have widths of only a few 10–100 m.Earlier studies (Thompson, 1992; Adams & Wes-nousky, 1998; Ibbeken & Warnke, 2000; Komatsuet al., 2001) have documented shorelines and

beach ridges in ancient lacustrine settings with asimilar dimension and shape as the exposedCardiel ridges, i.e. with widths of a few hundredmetres. Studies in the marine environment haveshown the existenceof beach ridgeswith awidth ofover 500 m (e.g. Sanders & Kumar, 1975; Kellyet al., 1999). The large diversity in themorphology(size, shape, crest elevation) of the described beachridges (Fig. 10) is the result of various factors

70

TW

T (

ms)

110

100

80

Sub

lake

leve

l dep

th (

m)

500 m

90

120

60

70

80

NW SE IV

III

II

ID

C

B A1600 m

ABC

1.8

1.6

1.4

1.2

1

0.8

0.6

0.4

0.2

0 D

Cor

e de

pth

(m)

CAR 02-3CAR 02-4CAR 02-5CAR 02-6

Seismic profile

Petrophysical core profiles

1 1.2Den. (g/cm3) MS (10-5*SI)

0 604020 1 1.2Den. (g/cm3) MS (10-5*SI)

0 604020 1 1.2Den. (g/cm3) MS (10-5*SI)

0 6040201 1.4Den. (g/cm3) MS (10-5*SI)

0 2001.8 100

Fig. 9. Seismic profile at the northern border of the lake (above) with the petrophysical core properties (below) atfour different sites (A to D). The seismic profile shows a typical contourite drift geometry with a strong thinning ofSequences I and II towards the channel (Site D) at the northern slope break. The density, as well as the magneticsusceptibility (MS) of short cores increases from the depocentre of the contourite drift (near Site A) towards thechannel (Site D). Dashed line kept at a fixed value to visualize the increase in density and magnetic susceptibility;note the changed axis for the petrophysical logs at Site D.

12 A. Gilli et al.


Fig.10.Serialseismic

profilesofbeachridgesattheonlapoftheSequenceboundary

IV/III(i.e.around107msTW

T).Theyare

arrangedalongthebeachridge

mainly

form

edin

thesouthern

andwestern

partofthelake.Theseismic

profile

numberis

indicatedin

thelowerrightcornerofeachsection.14Sand13W

were

acquiredwithaboomersystem.



including the local slope angle, the amount andcharacteristics of the sediment available for trans-port, the availability of accommodation, the waveregime, possible early diagenesis (i.e. calcitecementation forming beach rock) and the lengthof time the lake level resides at a particularelevation (Elliott, 1986; Adams & Wesnousky,1998). During storms, waves run up the ridge andcarry sediment from the front of the ridge over thecrest, depositing it on the landward side of thebarrier. This process, called barrier rollover, hasbeen inferred in outcrop studies of lacustrinedeposits (Adams & Wesnousky, 1998) and isresponsible for the typical asymmetric shape ofthe Cardiel beach ridges with a gently dippingslope lakeward and a steeper landward slope. Italso explains the difference in the elevation of thebeach ridge crests. For example, the crest elevationof beach ridges at the lateral onlap of the IV/IIIsequence boundary varies between 102 and110 ms TWT, which is equivalent to a differencein elevation of about 5Æ8 m. In contrast, the crestelevations of the beach ridges around 120 ms TWT(Fig. 13) differ only within a small range of 2 msTWT (equal 1Æ4 m). This smaller variability in the

crest elevation is probably the result of lower waveenergy reaching the shore due to the smaller fetcharea of the earlier lake. Previous studies in ancient(Adams &Wesnousky, 1998), as well as in modernlacustrine systems (Atwood, 1994) have identifiedsimilar ridge crest elevation variability. Thisfurther suggests that the height of the beach ridgecrests above the sill-water level is not constant(Orford et al., 1991; Adams & Wesnousky, 1998).The mapped lateral distribution of the beach

ridges (Fig. 12) shows that some of them aremissing on certain seismic lines. Lake floor topo-graphy is an important feature controlling beachridge formation. Adams & Wesnousky (1998)reported that beach ridges are most easily devel-oped on shorelines with an overall slope angle ofless than 4�. This may explain the discontinuity ofthe beach ridges in the Lago Cardiel basin.The formation of beach ridges under regres-

sional and transgressional conditions has beenwidely studied in the marine environment (seeElliott, 1986 and Selley, 1996 for reviews). Duringmarine regression, a beach ridge is left inland andexposed to subaerial erosion. The sediment ispartly carried back to the sea and re-deposited on

75

100

120

TW

T (

ms)

500 m 40

Sub

lake

leve

l dep

th (

m)

50

50

60

70

80

W E

IIIII

I

IV

75

100

120

TW

T (

ms)

500 m 40

Sub

lake

leve

l dep

th (

m)

50

50

60

70

80

W EVIbeach ridge

around 107 ms TWT

beach ridgearound 90 ms TWT



multiple reflections

change in the orientation of the profile

Fig. 11. Series of beach ridges on the gently dipping eastern slope of the lake. The well-preserved beach ridgesindicate an ‘in-place drowning’ of the ridges due to a stepwise lake level rise at the beginning of Sequence III. Notethe change in the orientation of the profile (see Fig. 1D for the exact location).

14 A. Gilli et al.


the beach face. The processes under transgres-sional conditions are more complicated. Twomechanisms are proposed for the landwardmigration of beach ridges during a transgression.The first mechanism has been termed ‘shorefaceretreat’ and involves the continuous landwardmigration of the beach ridge during a transgres-sion (Swift, 1968), which makes it difficult topreserve the individual beach ridges. The secondmechanism assumes ‘in-place drowning’ of thebeach ridges and the formation of a new ridge atmore landward positions (Sanders & Kumar,1975). The in-place drowning scenario results inthe original beach ridge being left almost un-touched. These mechanisms represent two endmembers and any combinations are possible.Rapid water level rises and a high sedimentsupply favour the in-place drowning of the beachridges whereas a slow water level rise and lowsediment supply favour ridge migration (Elliott,1986). In Lago Cardiel, the beach ridges aregenerally well preserved. This fact makes it ratherunlikely that the beach ridges were formed duringan early regression of the lake, as they would bereworked during the subsequent transgression atthe beginning of Sequence III. Furthermore, asignificant period of subaerial exposure of the

beach ridges after their formation under regres-sional conditions would lead to incision anddissection, as have been described for the highestand oldest beach ridges encircling Lago Cardiel(Stine & Stine, 1990). Therefore, the well-pre-served beach ridges beneath the modern lakewere probably formed under transgressive condi-tions, with a stepwise rise in lake level causingin-place drowning.The time taken to form fully developed beach

ridges is highly variable, but several studies inlacustrine (e.g. Adams & Wesnousky, 1998) andmarine environments (for summary see Carter,1988; p. 121) suggest that relatively fast formationof beach ridges (less than a year) is possible. Nodirect measurements are available to constrainthe period for the formation of the beach ridges inthe Lago Cardiel basin. Taking into account thelarge dimensions of these ridges, their formationcan be estimated within the range of severalmonths to a few years. Rapid formation of thebeach ridges would imply that only a thinsediment layer was deposited during the ridgeformation, which might explain the almost com-plete absence of onlapping reflections basinwardof the shoreline. In addition, rapid beach ridgeformation may have prevented the presence of

49°00'S

48°55'S

48°50'S

71°05'W71°10'W71°15'W71°20'W

Río Card ie l

Río Bayo

0 5km

N

57 m

s

73 m

s

28 m

s

135 ms

120 ms

90 m

s10

7 m

s

90 ms73 m

s

57 ms

28 ms

28 ms57 ms

135 ms

90 ms

90 m

s

107 ms

?

Fig. 12. Lateral distribution of se-ven drowned beach ridges. Meansublake level depth of the beachridge crests are given in ms TWT.They occur at the top of SequenceVI, or where present, at the top ofSequence V.



reflectors within Sequence III corresponding to alake level stillstand that can be linked to thedocumented beach ridges.

LACUSTRINE DRIFT

In contrast to the marine environment, only a fewexamples of contourite drift deposits have beendescribed in lacustrine settings (Lake Superior:Johnson et al., 1980; East African Rift lakes: John-son, 1996; Lake Baikal: Ceramicola et al., 2001;Lake Geneva: Girardclos et al., 2003). In LagoCardiel, Sequence I as well as the upper part ofSequence II are characterized by a moundedsediment distribution pattern reflecting large lat-eral differences in sedimentation rate within thebasin (Fig. 6A and B). The thickest part of thissediment mound reaches a maximum thickness ofabout 35 ms (� 25 m) in the northern central basinand covers an area of about 80–100 km2. Thisdeposit has striking similarities at three differentspatial scales with seismic criteria for recognizingcontourite drifts in marine environments (Faug-eres et al., 1999; Rebesco & Stow, 2001). These

criteria are: (1) On a large scale, the drift morphol-ogy can be described as a ‘mounded elongateddrift’ separated from the lake margin in the northby a distinct moat channel at the slope break. Incontrast to many marine contourite drifts, there isnowidespread discontinuity associatedwith high-amplitude reflections at the base orwithin the driftimplying a gradual start and a quasi-stable pres-ence of the currents since initiation. (2) On thescale of the seismic sequences (i.e. medium spatialscale), typical marine drift characteristics such asthe upward-convex geometry of the sequenceboundaries are found near the northern channel(Fig. 9). A lateralmigration of the drift systemwithsigmoidal progradational reflections is especiallypronounced towards thewest (Figs 2 and 7). (3)Ona smaller scale, the seismic facies of the Cardieldrift show low to medium-amplitude reflectionswith high lateral continuity and small sedimentwaves near the northern channel (Fig. 9) resem-bling the seismic facies of marine contouritedeposits. These seismic characteristics on differ-ent spatial scales point towards the presence of acontourite drift in Lago Cardiel. To further confirmthe current-controlled sedimentation pattern, four

250 m

114

120

130

TW

T (

ms)

85

Sub

lake

leve

l dep

th (

m)

90

Sub

lake

leve

l dep

th (

m)

13NE10S14S7SE 250 m 250 m 250 m

114

120

130

TW

T (

ms)

8E9aNE 12SE250 m250 m250 m

85

Sub

lake

leve

l dep

th (

m)

90

Fig. 13. Seismic profiles of beach ridges at a sublake level depth of � 120 ms TWT. Their dimensions are signifi-cantly smaller than the ones at the onlap of the Sequence boundary IV/III (i.e. Figure 10) as they were formed arounda smaller and shallower lake. The seismic profile number is indicated in the lower right corner of each section. 14Sand 13NE were acquired with a boomer system.

16 A. Gilli et al.


short gravity coreswith lengths between 1Æ3 m and1Æ6 m were recovered along one of the seismicprofiles (Fig. 9). This transect connects the depo-centre of the drift (Site A) with the axis of thenorthern channel (Site D) at the edge of themound.Petrophysical analysis (gamma-ray attenuationbulk density and magnetic susceptibility) of theshort gravity cores using a multisensor core logger(MSCL) revealed distinct changes in the densityand magnetic susceptibility profiles along thistransect (Fig. 9). The mean density of the coreslocated on the mound itself (sites A, B and C)shows only a slight lateral increase towards thechannel, whereas the density profile in the chan-nel floor (Site D) has dramatically higher values. Inaddition, the frequency and the absolute numbersof distinct layers with higher density increasetowards the channel from the sites A to C. Thedensity profile in the channel floor itself showsthicker and abrupt intervals with values up to2 g cm)3. These density changes are closely rela-ted to variations in grain size, whereas coarse-grained sediments exhibit higher density valuesdue to their lower porosity and lower watercontent (Weber et al., 1997 and references there-in). Themagnetic susceptibility profiles show verysimilar trends to the density profiles (Fig. 9). Thehigher magnetic susceptibility values towards andwithin the channel are likely associated withferromagnetic minerals such as magnetite, whichhave significantly higher densities than, for exam-ple, quartz or feldspar. Consequently, ferromag-netic grains are concentrated in areas of highbottom transport energy (Pirrung et al., 2002). Thecommon trend of higher density and magneticsusceptibility values in the northern channelindicate the occurrence of strong bottom currentactivity near the northern slope break that produceseveral centimetre-thick coarse layers. The eleva-ted density of these distinct layers near thenorthern channel is supported by the seismicprofiles, which show increasing amplitudes ofthe reflections in this area (Fig. 9). All thesearguments support the seismic interpretation thatthe mounded sedimentary body is a lacustrinecontourite drift.

Origin of the lake circulation

Contourite drifts result from current-drivendeposition. The currents are driven by forcessuch as wind, tides, waves and/or thermohalinecirculation. The location of Lago Cardiel withinthe westerly windbelt system suggests that thecurrents responsible for producing the contour-

ites were probably primarily wind-driven. Ana-lysis of the prevailing wind directions forGobernador Gregores (Fig. 1A, inset) shows thatthe relative annual frequency for wind originatingfrom the western sector (NW –W – SW) is over50% (analysed period: 1951–60; Prohaska, 1976).These predominant westerly winds produce awind stress on the lake surface that causes thesurface water to move. There are no direct currentmeasurements available for Lago Cardiel, but acomparison with studies in Lake Kinneret (Sea ofGalilee; northern Israel), a similar lacustrinesetting with a unidirectional and persistent windfield reveals some clues about the potential flowpattern in Lago Cardiel. Lake Kinneret exhibits alarge-scale counterclockwise circulation patternduring westerly wind periods in summer (Ser-ruya, 1975). Simulations with a three-dimen-sional model for Lake Kinneret (Pan et al., 2002)have confirmed that the wind stress field in thearea of the lake is the dominant factor controllingthe formation, as well as the direction, of thislake-scale gyre. By analogy, the flow pattern inLago Cardiel probably also consists of one largegyre with unknown vorticity due to the lack oflake current observations and data on the hori-zontal wind field. Basin-scale gyres are known tocharacterize even large lakes, and even thosewithout particularly rounded shapes (Schwab &Beletsky, 2003). A satellite image of Lago Cardiel(Landsat 7; 18 October, 1998; Fig. 14) documentsthe presence of a strong alongshore current by theclear deflection of the incoming suspensionplume of the Rıo Cardiel to the south (Fig. 14B)during the snowmelt in austral spring. The per-sistence of this current is supported by theabsence of any kind of prograding delta complexin the seismic profiles in the front of the mouth ofthe modern Rıo Cardiel. As the result of the stronglongshore current, coarse riverine sediment isdeposited south-east of the river mouth, forcingthe latter to migrate towards the north-west(Fig. 14B). The extent of longshore transportsuggests the presence of a persistent counter-clockwise gyre within the lake. The combinedeffect of such a gyre and the Coriolis force wouldresult in transport of fine sediment particlestowards the lake centre, where the highest sedi-mentation rates and evidence of contourite moun-ding are observed (i.e. Sequence I & II; Fig. 6Aand B). The satellite image, seismic, sedimento-logical and geomorphological evidence thereforeindicate the modern and past presence of strongcurrent activity in Lago Cardiel. The driving forcefor this lake current is most likely persistent



westerly winds. This is in good agreement withthe palaeoclimatic evolution of this area with theonset of the modern wind regime in the mid-Holocene (Markgraf, 1993).

HISTORY OF LAGO CARDIEL

The interpretation of the lateral distribution,shape, and internal characteristics of the seismic

sequences, as well as the presence of shorelinefeatures (e.g. the beach ridges) allows a recon-struction of past lake level positions and, togetherwith the timing of the onset of contourite driftdeposits, reveals further information about thepalaeoenvironmental evolution of the area. Theestablishment of a coherent age-model forthe deposition of the sequences and especiallyfor the sequence boundaries constrains the lakelevel history; 14C-AMS dates on organic remains

Fig. 14. (A) Satellite image of LagoCardiel (Landsat 7; 18Æ10. 1998).Subaerial exposed shorelines (blackarrows) are especially well definedon the gently dipping east shore. (B)Close-up of the suspension plume atthe mouth of the Rıo Cardiel. Notethe north-western migration of theinlet due to deposition of coarsesediment material by the strongalongshore current in the south-east.The size of this asymmetric deltapoints clearly towards a persistentactivity of a lake current at the rivermouth.

18 A. Gilli et al.


and dated tephra layers can be linked to theseismic stratigraphy by correlation with the petr-ophysical profiles of the sedimentary cores (Gilliet al., 2001).

Sequence VI & V: early evolution of the lakebasin

Because of the limited seismic coverage ofSequence VI and the absence of any sedimento-logical constraints, the origin of Sequence VIremains unclear. It consists of either olderlacustrine sediments or, more likely, the Cretac-eous-Tertiary claystones that make up the bed-rock surrounding the basin. The angularunconformity and the high-amplitude characterof the irregular top of Sequence VI (Fig. 2)indicate a major hiatus, which was most probablyassociated with a complete desiccation of theentire basin. From its distribution, shape andinternal structures, Sequence V is interpreted as aformer alluvial fan unit. The occurrence of thissequence is restricted to the western side of thebasin (Fig. 4A) coinciding with the inflow of RıoCardiel. It is puzzling that no lake sediments werefound in the basin centre corresponding to theformation of the alluvial fan that makes upSequence V. Under the assumption of a closedlake basin, aeolian erosion would be the onlymechanism available to export fine lake sedi-ments out of this depression. The lack of any ageconstraints for the alluvial fan deposits preventsany detailed interpretation of the origin andsignificance of Sequence V. A second majorhiatus of unknown duration is indicated by theirregular upper boundary of Sequence V.

Sequence IV: the establishment of a shallowlake

With Sequence IV, the lacustrine sedimentationof the modern Lago Cardiel began. The flat onlapgeometry of the reflections in the lower part ofSequence IV is almost horizontal indicating ashallow lake with only a few metres of water. Inthe upper part of Sequence IV, the drapedgeometry of the onlapping reflections indicatesthat the lake was deeper during this period(Fig. 5A, upper dotted line). The pronouncedhigh-amplitude reflections in the lower part maycorrespond to repeated lake level lowerings oreven total desiccations of the basin. Because ofthe absence of erosional features, these periods ofnear-desiccation must have been short. The inter-nal onlap around 130 ms TWT is related to a

minor lake level fall in the middle of SequenceIV. The formation of two rather small beach ridgesets at a depth of 135 ms and 120 ms TWTindicates a stillstand of the lake level positionwithin Sequence IV. The lake level stabilizedagain at the top of Sequence IV forming the best-developed set of drowned beach ridges in LagoCardiel at a depth of around 107 ms TWT(Fig. 10). The irregular IV/III sequence boundary,which is slightly more pronounced near theshore, points to a minor lake level lowering atthe end of Sequence IV. Core material is onlyavailable from core CAR 99–7P, which penetrated2Æ5 m into the Sequence IV (Fig. 7). A 14C-AMSdate obtained from a piece of wood at theSequence boundary IV/III revealed an age of11 220 ± 85 14C yr BP. Together with a seconddate from the base of core CAR 99–7P, a minimumage of around 20 800 yr BP for the onset ofSequence VI deposition has been estimated underthe assumption of a constant sedimentation rate(Gilli et al., 2001). However, given the internalevidence for lake level stillstands and lowerings,the deposition of Sequence IV probably begansignificantly earlier.A restricted shallow lake that was separated

from the main lake formed in the Bahıa Pescade-ria with a possible overflow at a level of 79 msTWT. The limited documentation of this localsedimentary unit deposited in a restricted part ofthe lake makes it impossible to determine theexact extent of this sedimentary body and istherefore neglected in the volume calculationsdiscussed below.

Sequence III: early Holocene transgression

The regional hydrological balance changed dra-matically at the beginning of Sequence III.Sequence III is found throughout the entire basinup to the modern shoreline (Fig. 8) implying alarge transgression with a lake level rise of almost80 m. This lake level rise shows a strong stepwisecharacter. A series of lake level stillstands pro-duced a set of beach ridges at sublake level depthsof around 90 ms/65 m, 73 ms/53 m, 57 ms/42 mand 28 ms/20 m. The excellent preservation ofmost beach ridges suggests a rapid stepwise risein lake level after ridge formation, resulting inin-place drowning of these shoreline features.The overall duration of the transgression isconstrained by a 14C date on a piece of woodembedded in lacustrine sediments (10 230 ± 6514C yr BP) obtained from a Livingstone-typepiston core (CAR 99–02 L) taken near the modern



shore (Markgraf et al., 2003). This is a minimumage for lacustrine conditions near the modernshoreline implying that this transgression with avertical height of almost 80 m occurred within afew hundred years. This transgression presuma-bly continued and eventually formed the previ-ously described shoreline at + 55 m above themodern level around 9800–9500 14C yr BP (Stine& Stine, 1990; Gilli et al., 2001) indicating a totaltransgression of 135 m across the Pleistocene/Holocene boundary. The transparent seismiccharacter of the lower part of Sequence III isprobably the result of the elevated clastic inputthat led to rapid deposition of the lacustrinedeposits. The maximum thickness for this relat-ively uniform sequence occurs in front of the RıoCardiel inlet (Fig. 6C). According to the lake levelhighstand record of Stine & Stine (1990), the lakelevel receded to the modern position until theend of Sequence III. The Sequence boundary III/IIconsists of a tephra layer, several cm thick, fromHudson volcano that erupted at 6700 14C yr BP(Markgraf et al., 2003). The high impedance con-trast of the tephra produces the observed high-amplitude reflection.

Sequences II and I: onset of contourite driftdeposition

The distribution of these two sequences up tothe modern shoreline indicates that the lakelevel never receded significantly below themodern lake level during this interval. Thisinterpretation is further supported by the analy-sis of a sediment core taken at the modernshoreline that indicates a lake level as high orhigher than today (Markgraf et al., 2003). Thelake level reconstruction by Stine & Stine (1990)proposed a mid-Holocene highstand of + 21Æ5 maround 5130 14C yr BP and four minor excur-sions up to +10 m in the last 2000 years. Thealmost transparent seismic facies in the lowerpart of Sequence II may represent the mid-Holocene highstand, though this is not suppor-ted by the analysis of the near shore sedimentcore (Markgraf et al., 2003). The four minor lakelevel fluctuations in the late Holocene werepossibly too small to be resolved on the seismicprofiles. Nevertheless, fundamental changes oc-curred in the general lateral distribution of thesediment after the deposition of the lower por-tion of Sequence II. The formation of contouritedrifts with a pronounced mounded depocentrein the northern central area of the lake indicatesthe presence of strong bottom currents. The

onset of the contourite drifts is an indicator ofpast wind conditions at a latitude of 49�S andreveals that the westerlies over the Lago Cardielarea had strengthened after the mid–Holocene.Sequence boundary II/I is marked by anothertephra layer, which corresponds to the eruptionof one of the volcanoes of the Northern AndeanAustral Volcanic Zone (NAVZ) around3010 ± 45 14C yr BP. The strong thinning ofSequence II and I near the present-day shoreline(Fig. 8) is the result of re-deposition of uncon-solidated beach material by wave action. Erosionof beach sediments usually starts at wave base,which can easily be as deep as 5 m (Sly, 1994and references therein), especially during majorstorm periods. This is in good agreement withevidence for beach erosion down to 7 m (Fig. 8).The saw-tooth topography is the result of somehorizons having a slightly higher resistance tothe wave-induced shore erosion. The erodedmaterial is transported offshore by rolling andsliding processes and subsequently deposited inan offshore accumulation zone that is welldeveloped as a bulge on the east slope of LagoCardiel (Fig. 8).

Estimation of the Late Quaternary sedimentflux to the lake

The volume of the youngest four sequences canbe reconstructed from the seismic data (Fig. 6A–D, lower left corner). Given the dated sequenceboundaries, the average annual sedimentationvolume (m3/ year) can be determined for se-quences IV to I (Fig. 15). For Sequence IV, theanalysis reveals an average annual sedimentationvolume of � 61 700 m3/ year under the assump-tion that Sequence IV started around 20 800 yrBP. In the more probable case of an earlier start ofSequence IV, this value would significantlydecrease. The average annual sedimentation vol-ume for Sequence III is � 236 000 m3/ yr. How-ever, much of the sediment was deposited abovethe modern shoreline during the early Holocenelake level highstand and this component is notincluded in this estimate. Thus, the sedimentaryinput during Sequence III could have been signi-ficantly higher. During Sequence II and I, theaverage annual sedimentation volume increasedstrongly from � 302 200 m3/ yr to � 611 400 m3/yr. This large increase is in part explained by thelower compaction of the younger sediments, aswell as the recycling of easily erodable lacustrinesediments deposited during the lake level high-stand of Sequence III.

20 A. Gilli et al.


CONCLUSIONS

The character and the geometrical relationship ofthe seismic reflections (e.g. onlaps), the sedimentdistribution and the presence of shoreline fea-tures (e.g. in-place drowned beach ridges) allowthe reconstruction of the Lago Cardiel lake levelcurve through time. Six major seismic sequencescan be distinguished and traced in the subsur-face of Lago Cardiel. The stratigraphically lowestsequence, Sequence IV, is overlain in some areasby an inferred alluvial fan complex comprisingSequence V. The age of these sequences isunknown. The angular unconformity at the topof Sequence VI, as well as the strong andirregular reflection on top of Sequence V, indi-cates two major hiatuses presumably associatedwith periods of desiccation during drier climateconditions of possible long duration. The mod-ern lake was established with the deposition ofSequence IV, which spans at least the LastGlacial Maximum until the end of the Pleisto-cene (i.e. 11 220 14C yr BP) and includes a shortdesiccation episode. A major transgression up tothe modern shoreline indicates a substantialchange in the regional hydrological balanceduring the late Pleistocene/early Holocene. Setsof beach ridges at depths of 90 ms, 73 ms, 57 msand 28 ms TWT indicate that this fast transgres-sion of almost 80 m within several hundredyears was pulsed. Beach ridges provide ideallake level indicators, especially during fasttransgressions, when the time is too short toestablish clear onlapping reflections. The excel-lent preservation of the shoreline features pointstowards a stepwise rise in lake level leading toin-place drowning of beach ridges. This trans-gression probably culminated in an early Holo-cene lake level highstand of + 55 m (Stine &

Stine, 1990; Markgraf et al., 2003) resulting in anoverall vertical transgression of 135 m. There is noevidence of erosion near the shore in the threeyoungest sequences indicating that the lake levelnever receded significantly below the modernshoreline, which is in good agreement with inde-pendent sedimentological and palaeoecologicaldata (Markgraf et al., 2003). Within Sequence II,the depositional pattern changed from a moreuniform distribution to one that focused depos-ition in the north-central part of the basin. Thischange in the mid-Holocene produced a moun-ded-shaped sediment distribution with a maxi-mum thickness of 25 m. The mound is interpretedas having been deposited by bottom currents.This distribution was most likely produced as aresult of a change in the pattern and strength ofthe lake current-system induced by strong andpersistent westerly winds that commenced in themid-Holocene.

ACKNOWLEDGEMENTS

We are indebted to Jorge Moreteau, whose helpand ideas together with the tireless effort of hiscrew made the field campaigns on the windy LagoCardiel a success. We are especially grateful toJean Captain for transport and field operations ofthe GEOPULSETM seismic unit. For assistanceand essential help in the field, we thank PlattBradbury, Robert Hofmann and Antoinette Ludin.Alfred Wuest (EAWAG Kastanienbaum) isthanked for the stimulating discussions aboutthe lake current and Michael Maxelon for thehelp with the gocad

TM program. The helpfulcomments of N. Wattus and an anonymousreviewer, as well as of the editor P. Haughtonare gratefully acknowledged.This study is part of the PATO/PaLaTra project,

which is cofunded by the U.S. National ScienceFoundation grants NSF-EAR-9709145, NSF-ATM-008267 and NSF-ATM-0081279 to V. Mark-graf/K. Kelts; and by the Swiss National ScienceFoundation grants N�21–50862Æ97 and N�20–61704Æ00/1 to the ETH Zurich and University ofGeneva limnogeology groups.

REFERENCES

Abbott, M.B., Finney, P.B., Edwards, M.E. and Kelts, K.R.(2000) Lake-level reconstructions and paleohydrology of

Birch Lake, Central Alaska, based on seismic reflection

profiles and core transects. Quaternary Res. 53, 154–166.

0

100000

200000

300000

400000

500000

600000

700000

25000 20000 15000 10000 5000 0av. a

nn. s

edim

ent v

olum

e (m

3 /yr

)

Age (14C yr. BP)

?IV III III

Fig. 15. Average annual sedimentation volumes for thesequences IV to I.



Adams, K.D. and Wesnousky, S.G. (1998) Shoreline proces-

ses and the age of the Lake Lahontan highstand in the

Jessup Embayment, Nevada. Geol. Soc. Am. Bull. 110,1318–1332.

Ariztegui, D., Anselmetti, F.S., Kelts, K., Seltzer, G.O. and

D’Agostino, K. (2000) Identifying paleoenvironmental

change across South and North America using high-resolu-

tion seismic stratigraphy in lakes. In: Interhemispheric Cli-mate Linkages (Ed. V. Markgraf). Academic Press, London.

Atwood, G. (1994) Geomorphology applied to flooding prob-

lems of closed-basin lakes, specifically Great Salt Lake,

Utah. Geomorphology, 10, 197–219.Baker, P.A., Seltzer, G.O., Fritz, S.C., Dunbar, R.B., Grove,

M.J., Tapia, P.M., Gross, S.L., Rowe, H.D. and Broda, J.P.(2001) The history of South American tropical precipitation

for the past 25,000 years. Science, 291, 640–643.Carter, R.W. (1988) Coastal Environments – an Introduction to

the Physical, Ecological and Cultural Systems of Coastlines.

XV, 617 S. Academic Press, London.

Ceramicola, S., Rebesco, M., De Batist, M. and Khlystov, O.(2001) Seismic evidence of small-scale lacustrine drifts in

Lake Baikal (Russia). Mar. Geophys. Res. 22, 445–464.Elliott, T. (1986) Siliciclastic Shorelines. In: Sedimentary

Environments and Facies (Ed. H. G. Reading), 2nd edn.

Blackwell, Oxford.

Faugeres, J.-C., Stow, D.A.V., Imbert, P. and Viana, A. (1999)Seismic features diagnostic of contourite drifts. Mar. Geol.

162, 1–38.Feruglio, E. (1950) Descripcion Geologica de la Patagonia, III.

431 pp. Direccion Nacional de Yacimientos Fiscales, Bue-

nos Aires.

Galloway, R.W., Markgraf, V. and Bradbury, J.P. (1988) Dat-ing shorelines of lakes in Patagonia, Argentina. J. S. Am.Earth Sci. 1, 195–198.

Gilli, A., Anselmetti, F.S., Ariztegui, D., Bradbury, J.P., Kelts,K.R., Markgraf, V. and McKenzie, J.A. (2001) Tracking

abrupt climate change in the Southern hemisphere: a seis-

mic stratigraphic study of Lago Cardiel, Argentina (49�S).Terra Nova, 13, 443–448.

Girardclos, S., Baster, I., Wildi, W., Pugin, A. and Rachoud-Schneider, A.-M. (2003) Bottom-current and wind-pattern

changes as indicated by Late Glacial and Holocene sedi-

ments from western Lake Geneva (Switzerland). Eclogae

geol. Helv. 96 (Suppl. 1), S39–S48.Harrison, S.P. and Digerfeldt, G. (1993) European lakes as

palaeohydrological and palaeoclimatic indicators. Quater-

nary Sci. Rev. 12, 233–248.Heinsheimer, J.J. (1959) El Lago Cardiel. Anales Academia

Argentina Geografia, 3, 86–132.Hoffmann, J.A.J. (1975) Atlas Climatico de America del Sur.

World Meteorological Organization, Geneva and UNESCO,

Paris.

Ibbeken, H. and Warnke, D.A. (2000) The Hanaupah-Fan

shoreline deposit at Tule Spring, a gravelly shoreline de-

posit of Pleistocene Lake Manly, Death Valley, California,

USA J. Paleolimnol. 23, 439–447.Johnson, T.C. (1996) Sedimentary processes and signals of

past climate change in the large lakes of the East African Rift

Valley. In: The Limnology, Climatology and Paleoclimatol-ogy of the East African Lakes (Eds T. C. Johnson and E. O.

Odada), pp. XII, 664 S. Gordon & Breach, Amsterdam.

Johnson, T.C., Carlson, T.W. and Evans, J.E. (1980) Con-

tourites in Lake Superior. Geology, 8, 437–441.Johnson, T.C., Kelts, K. and Odada, E. (2000) The Holocene

history of Lake Victoria. Ambio, 29, 2–11.

Kelly, W.M., Albanese, J.R., Coch, N.K. and Harsch, A.A.(1999) Seismic reflection and vibracoring studies of the

continental shelf offshore central and western Long Island,

New York. Mar. Georesour. Geotec. 17, 141–153.Komatsu, G., Brantingham, P.J., Olsen, J.W. and Baker, V.R.

(2001) Paleoshoreline geomorphology of Boon Tsagaan

Nuur, Tsagaan Nuur and Orog Nuur; the Valley of Lakes,

Mongolia. Geomorphology, 39, 83–98.Markgraf, V. (1993) Paleoenvironments and paleoclimates in

Tierra del Fuego and southernmost Patagonia, South

America. Palaeogeogr. Palaeoclimatol. Paleoecol. 102, 53–68.

Markgraf, V., Bradbury, J.P., Schwalb, A., Burns, S.J., Stern,C., Ariztegui, D., Gilli, A., Anselmetti, F.S., Stine, S. andMaidana, N. (2003) Holocene palaeoclimates of southern

Patagonia: limnological and environmental history of Lago

Cardiel, Argentina. Holocene, 13, 581–591.McCulloch, R.D., Bentley, M.J., Purves, R.S., Hulton, N.R.J.,

Sugden, D.E. and Clapperton, C.M. (2000) Climatic infer-

ences from glacial and palaeoecological evidence at the last

glacial termination, southern South America. J. Quaternary

Sci. 15, 409–417.Orford, J.D., Carter, R.W.G. and Jennings, S.C. (1991) Coarse

clastic barrier environments; evolution and implications for

Quaternary sea level interpretation. Quatern. Int. 9, 87–104.Pan, H., Avissar, R. and Haidvogel, D.B. (2002) Summer cir-

culation and temperature structure of Lake Kinneret.

J. Phys. Oceanogr. 32, 295–313.Piovano, E.L., Ariztegui, D. and Damatto Moreira, S. (2002)

Recent environmental changes in Laguna Mar. Chiquita

(central Argentina): a sedimentary model for a highly vari-

able saline lake. Sedimentology, 49, 1371–1384.Pirrung, M., Futterer, D., Grobe, H., Matthiessen, J. and Ni-

essen, F. (2002) Magnetic susceptibility and ice-rafted deb-

ris in surface sediments of the Nordic Seas: implications for

Isotope Stage 3 oscillations. Geo-Mar. Lett. 22, 1–11.Prohaska, F. (1976) The climate of Argentina, Paraguay and

Uruguay. In: Climates of Central and South America (Ed.

W. Schwerdfeger), 12, pp. 13–112. Elsevier, Amsterdam.

Rabassa, J. and Clapperton, C.M. (1990) Quaternary glacia-

tions of the Southern Andes. Quaternary Sci. Rev. 9, 153–174.

Ramos, V.A. (1982) Geologia de la region del Lago Cardiel,

Provincia de Santa Cruz. Asociacion Geologica Argentina,

Revista, 37, 23–49.Ramos, V.A. (1989) Andean foothills structures in Northern

Magallanes Basin, Argentina. AAPG Bull. 73, 887–903.Rebesco, M. and Stow, D. (2001) Seismic expression of

contourites and related deposits: a preface. Mar. Geophys.

Res. 22, 303–308.Richardson, J.L. (1969) Former lake-level fluctuations, their

recognition and interpretation. Mitt. Internat. Verein.Limnol. 17, 78–93.

Sanders, J.E. and Kumar, N. (1975) Evidence of shoreface re-

treat and in-place ‘drowning’ during Holocene submergence

of barriers, shelf off Fire Island, New York. Geol. Soc. Am.Bull. 86, 65–76.

Scholz, C.A. and Rosendahl, B.R. (1988) Low lake stands in

lakes Malawi and Tanganyika, East Africa, delineated with

multifold seismic data. Science, 240, 1645–1648.Schwab, D.J. and Beletsky, D. (2003) Relative effects of wind

stress curl, topography, and stratification on large-scale

circulation in Lake Michigan. J. Geophys Res. Oceans, 108,CZ, Article No. 3044.

22 A. Gilli et al.


Selley, R.C. (1996) Ancient sedimentary environments and

their sub-surface diagnosis. XIV, 300 S. Chapmann Hall,

London.

Seltzer, G.O., Baker, P., Cross, S., Dunbar, R. and Fritz, S.(1998) High-resolution seismic reflection profiles from Lake

Titicaca, Peru-Bolivia: Evidence for Holocene aridity in the

tropical Andes. Geology, 26, 167–170.Serruya, C. (1975) Wind, water temperature and motions in

Lake Kinneret: general pattern. Verh. Internat. Verein.

Limnol. 19, 73–87.Sly, P.G. (1994) Sedimentary processes in lakes. In: Sediment

Transport and Depositional Processes (Ed. K. Pye). Black-

well, Oxford.

Stine, S. and Stine, M. (1990) A record from Lake Cardiel of

climate in southern South America. Nature, 345, 705–708.Street-Perrott, F.A. and Harrison, S.P. (1985) Lake level and

climate reconstruction. In: Paleoclimate Analysis and

Modeling (Ed. by A. D. Hecht), pp. 291–340. Wiley, New

York.

Swift, D.J.P. (1968) Coastal erosion and transgressive strati-

graphy. J. Geol. 76, 444–456.Thompson, T.A. (1992) Beach-ridge development and lake-

level variation in southern Lake Michigan. Sediment. Geol.

80, 305–318.

Valero-Garces, B.L., Grosjean, M., Schwalb, A., Geyh, M.,Messerli, B. and Kelts, K. (1996) Limnogeology of Laguna

Miscanti; evidence for mid to late Holocene moisture

changes in the Atacama Altiplano (Northern Chile).

J. Paleolimno. 16, 1–21.Verschuren, D., Laird, K.R. and Cumming, B.F. (2000) Rainfall

and drought in equatorial east Africa during the past 1,100

years. Nature, 403, 410–414.Weber, M.E., Niessen, F., Kuhn, G. and Wiedicke, M. (1997)

Calibration and application of marine sedimentary physical

properties using a multi-sensor core logger. Mar. Geol. 136,151–172.

Wenzens, G. (2002) The influence of tectonically derived relief

and climate on the extent of the last Glaciation east of the

Patagonian ice fields (Argentina, Chile). Tectonophysics,345, 329–344.

Wenzens, G. (2004) Comment on ‘modelling the inception of

the Patagonian icesheet’. Quat. Int. 112, 105–109.

Manuscript received 1 May 2003;revision accepted 17 August 2004.



Seismic stratigraphy, buried beach ridges and contourite ... · Seismic stratigraphy, buried beach...

Documents

Transcript of Seismic stratigraphy, buried beach ridges and contourite ... · Seismic stratigraphy, buried beach...