Content uploaded by Raymond S Bradley
Author content
All content in this area was uploaded by Raymond S Bradley
Content may be subject to copyright.
Limnology, sedimentology, and hydrology of a
jökulhlaup into a meromictic High Arctic lake1
Ted Lewis, Pierre Francus, and Raymond S. Bradley
Abstract: A large ice-dammed lake drained catastrophically into Lake Tuborg, Ellesmere Island, beginning on 25 July
2003. Limnological, sedimentological, and hydrological parameters were recorded before, during, and after this event.
For several weeks prior to the jökulhlaup, water overtopped the ice-dammed lake and flowed into Lake Tuborg’s fresh-
water basin. A shallow sill separates the freshwater basin from a larger, deeper basin containing -25 PSU (practical
salinity units) salt water. The sill blocked underflows from entering the saltwater basin before the jökulhlaup. The ice-
dammed lake drained completely and catastrophically when englacial or subglacial conduits developed, and a glacier
portal formed 980 m from the Lake Tuborg shore, marking the beginning of the jökulhlaup. The level of Lake Tuborg
increased by 7.6 m in 84 h. This jökulhlaup is the largest known to have occurred in the High Arctic, and the largest
witnessed in Canada since 1947. Strata of very cold water flowed above the chemocline for about 14 km, from the sill
to the southwest end of the lake. The cold strata turbulently mixed with underlying salt water, allowing for saltwater
flocculation of suspended sediment, causing rapid settling. The saltwater layer very slightly freshened and cooled.
Close to the sill, near-surface sediments derived from the jökulhlaup are coarse and laminated; however, no erosion
occurred toward the distal end of the lake, where a fining upward unit with a coarse base was deposited.
Résumé : Un grand lac retenu par des glaces s’est drainé de manière catastrophique dans le lac Tuborg, de l’île Ellesmere,
le 25 juillet 2003. Les paramètres limnologiques, sédimentologiques et hydrologiques ont été mesurés avant, durant et
après cet événement. Durant plusieurs semaines avant le jökulhlaup, de l’eau se déversait par-dessus le barrage de glace
du lac et s’écoulait dans le bassin d’eau douce du lac Tuborg. Un seuil peu profond sépare le bassin d’eau douce d’un
bassin plus grand et plus profond contenant de l’eau à une salinité d’environ 25 USP (unités de salinité pratique).
Avant le jökulhlaup, le seuil empêchait des courants de fond d’entrer dans le bassin d’eau salée. Le lac retenu par les
glaces s’est drainé complètement et de manière catastrophique lorsque des canalisations dans ou sous la glace se sont dé-
veloppées et qu’un portail glaciaire s’est formé à 980 m des berges du lac Tuborg, marquant ainsi le début du jökulhlaup.
Le niveau du lac Tuborg s’est élevé de 7,6 m en 84 heures. Ce jökulhlaup est le plus grand connu qui se soit produit
dans le Grand Nord et le plus gros à être vu au Canada depuis 1947. Une couche d’eau très froide coulait par-dessus le
chemocline sur une distance d’environ 14 km, depuis le seuil jusqu’à l’extrémité sud-ouest du lac. La couche froide s’est
mêlée de manière turbulente avec l’eau salée sous-jacente engendrant une floculation des sédiments en suspension dans l’eau
salée et une sédimentation rapide. La couche d’eau salée est devenue légèrement plus douce et plus froide. Les sédiments de
surface provenant du jökulhlaup, sont grossiers et laminés à proximité du seuil; toutefois, il ne s’est pas produit d’érosion à
l’extrémité distale du lac, où s’est déposée une unité à base grossière et à granodécroissance vers le haut.
[Traduit par la Rédaction] Lewis et al. 806
Introduction
Jökulhlaups (Icelandic for “glacier-burst”) have never
been monitored in detail entering downstream water bodies.
The potential for erosion and sediment bypassing during
jökulhlaups is assumed to be great, and this is probably why
models describing glacier dam position and jökulhlaup fre-
quency and intensity (e.g., Clague and Evans 1994) have not
yet been tested using long sediment archives. If a site could
be identified where a characteristic jökulhlaup facies is de-
posited without producing an unconformity, an uninterrupted
record of jökulhlaup frequency could be obtained from a
sediment archive.
Lake Tuborg is a large, fiord-type lake on Ellesmere Island
in the Canadian High Arctic (Fig. 1) adjacent to the Agassiz
Ice Cap. Its bottom sediments are annually laminated (varved)
in some locations. It also contains trapped salt water (it is
meromictic). A large ice-dammed lake catastrophically drained
into Lake Tuborg in July 2003. A limnologic process study
was underway in Lake Tuborg at the time of the jökulhlaup.
Can. J. Earth Sci. 44: 791–806 (2007) doi:10.1139/E06-125 © 2007 NRC Canada
791
Received 25 July 2006. Accepted 21 November 2006. Published on the NRC Research Press Web site at http://cjes.nrc.ca on
20 July 2007.
T. Lewis2and R.S. Bradley. Climate System Research Center, Department of Geosciences, University of Massachusetts, 233
Morrill Science Center, Amherst, MA 01003–9297, USA.
P. Francus. Institut national de la recherche scientifique, Centre Eau, Terre et Environnement, 490 rue de la couronne, Québec, QC
G1K 9A9, Canada.
1Polar Continental Shelf Project Contribution 028-06.
2Corresponding author (e-mail: lewist@geo.umass.edu).
© 2007 NRC Canada
792 Can. J. Earth Sci. Vol. 44, 2007
Fig. 1. (a) Regional map of Ellesmere Island and surroundings in the Canadian High Arctic. (b) Physiography and topography near Lake Tuborg and the ice-dammed lake
(IDL). Contour interval is 500 feet (-150 m); shading is feet asl (above sea level). (c) Lake Tuborg bathymetry. Solid bathymetric contours are every 20 m. Stars are core
locations; triangles are limnologic monitoring stations. Location “i”in(b) and (c) is at 80.98°N, 75.555°W.
Conductivity–temperature–depth (CTD) casts, surface grav-
ity cores, and a lake level record were obtained in the 2001–
2003 melt seasons, so unique records of physical lake condi-
tions before, during, and after the jökulhlaup were obtained.
A major objective of this paper is to identify the type and
spatial variability of jökulhlaup-derived processes and de-
posits. These data also provide an opportunity to differenti-
ate jökulhlaup-derived lacustrine sedimentary processes and
deposits from those that are not jökulhlaup derived. This is
particularly important at Lake Tuborg because there are sev-
eral relatively small supraglacial lakes on the Agassiz Ice Cap
that drain—perhaps annually—as slush flows that quickly and
energetically transport large volumes of sediment to the lake
(Braun et al. 2000). The data also allow characterization of
boundary layer mixing processes along the sharp and strong
Lake Tuborg chemocline; these processes have previously
only been recorded in estuaries (e.g., Geyer and Smith 1987)
and laboratory experiments (e.g., Rimoldi et al. 1996). The
regionally unique hydrology and drainage mechanisms of
the 2003 jökulhlaup are also described.
Site description
Lake Tuborg was formed about 3000 years ago when An-
toinette Glacier advanced and trapped sea water in the lake
(Long 1967). The lake, located in a steep-walled 450–600 m
deep valley, is at -11 m asl (above sea level) (Fig. 1b). Soils
are very thin cryic regosols superimposed on permafrost,
and vegetation is very sparse. Lake surface area is 42 km2,
and maximum length and width are 20.9 and 3.4 km, respec-
tively (Fig. 1c). There is a large, deep, meromictic basin at
the southwest end (maximum depth = -145 m), and a smaller,
shallower freshwater basin at the northeast end of the lake
(maximum depth = -74 m; Fig. 1c). A 34 m deep sill sepa-
rates the two basins. Lake ice cover is nearly perennial.
A major stream enters Lake Tuborg at “i” (throughout the
text, “i” to “iv” refer to locations in Fig. 1b). It is entirely fed
by snowmelt; peak streamflow occurs in late June, and dis-
charge decreases greatly when watershed snow cover is ex-
hausted shortly afterward. Watersheds above “iii” and “iv”
are extensively glacierized, with higher mean elevations than
at “i”. This creates higher discharge duration and amount,
and higher sediment transport than at “i”, but the timing of
peak flow is delayed until early to mid-July when peak sum-
mer air temperature occurs (Braun et al. 2000). Small supra-
glacial lakes on the Agassiz Ice Cap quickly drained in July
1995, and discharge and suspended sediment transport briefly,
but greatly, increased (Braun et al. 2000). The ice-dammed
lake responsible for the 2003 jökulhlaup is part of the water-
shed above “ii”, but the lake captures almost all glacial melt
from the catchment while filling, so the streamflow regime at
“ii” is normally nival. CTD casts and visual observations at
the glacier terminus near “ii” showed no evidence of sub-
glacial meltwater discharge, even during peak summer melt.
The edge of the Agassiz Ice Cap parallels Lake Tuborg at
-900 m asl (Fig. 1b). The Agassiz Ice Cap generally is cold
based, and discharge is from supraglacial snow and ice melt;
however, outlet glaciers near sea level at the northeast and
southwest ends of the lake are warm based (Braun et al.
2000). The ice-dammed lake is dammed by a tributary gla-
cier of the Agassiz Ice Cap that branches into a small valley
between 300 and 460 m asl (Figs. 1b, 2). Bergs calve into
the lake at its northeast and southwest ends, and the glacier
dam at the southwest end normally blocks all outflow.
Mean annual air temperature at Eureka (Fig. 1a;10m
asl), the closest Meteorological Service of Canada (Environ-
ment Canada) weather station to Lake Tuborg, is –20 °C,
and the only months with average temperatures greater than
0 °C are June (2.3 °C), July (5.6 °C), and August (3.1 °C).
Methods
Monitoring at Lake Tuborg took place continuously from
mid May to mid August in 2001–2003. Coring and limnologic
monitoring stations are shown in Fig. 1c. Air temperature was
recorded every 15 minutes near lake level; precipitation was
collected at ground level and manually read. Lake level was
recorded every 15 minutes (cf. Reedyk et al. 1997). End of
season lake level was marked with cairns to ensure a common
lake level datum. The lake level recorder could not be used
after 21 July 2003 because of shifting ice pans and rapidly
rising lake level, so water height after this date was surveyed
with a staff and level. The measurements were made near “i”.
Water column conductivity, temperature, and density were
measured with a SBE 19 SEACAT Profiler CTD3(Sea-Bird
Electronics Inc., Bellevue, Washington, USA). Conductivity is
reported as specific conductivity inµS/cm (adjusted by 2%/°C
to 25 °C; Ludlam 1996), temperature is reported as potential
temperature (Tθ), and density is potential density (σθ) refer-
enced to 0 decibars (1 bar = 100 kPa). Brunt-Väisälä fre-
quency (N2) is a measure of density stratification and resistance
to mixing (Wüest and Lorke 2003) and is calculated with 1 m
depth bins. CTD casts were plotted in Ocean Data View
(Schlitzer 2005). A damaged connector prevented the CTD
profiler instrument from being lowered deeper than -80min
2003. Dissolved oxygen was measured with a Hydrolab
DataSonde 4a(Hach Company, Loveland, Colorado, USA).
Transmissivity is measured in percent with a Sea Tech
transmissometer (Sea Tech Inc., Corvallis, Oregon). When
transmissivity is >10%, it is transformed to suspended sedi-
ment concentration (SSC). However, transmissivity from the
saltwater layer is not transformed.4
Sediment cores were obtained with an Ekman dredge. Thin
sections were prepared by following procedures in Francus
and Asikainen (2001). Grain size was determined by scanning
electron microscopy (SEM) image analysis of thin sections
(Francus 1998).5Grain size is presented as “equivalent disc
diameter” (EDD; Francus 1998). Virtually no microorganisms
© 2007 NRC Canada
Lewis et al. 793
3See supplementary Table S1 for SEACAT thermistor and conductivity cell accuracies. Supplementary data can be purchased from the De-
pository of Unpublished Data, CISTI, National Research Council of Canada, Ottawa, ON K1A 0S2, Canada. Data are also available along
with the article on the NRC Research Press Web site.
4See supplementary Fig. S1.
5See supplementary Table S2 for SEM settings and image analysis procedure details.
were present. Mean grain size was calculated every 500 µm
along thin section photomosaics.
Bathymetric soundings (303 measurements) were obtained
in Lake Tuborg. Ice-dammed lake dimensions were deter-
mined with an aerial photograph (a-16687-47, -1 : 60 000,
taken July 1959; Fig. 2).
Results
Climate and weather
From 28 May to 10 August, each year from 2001 to 2003,
daily air temperatures at Eureka and Lake Tuborg are highly
correlated (r2= 0.79, slope = 0.84), with mean air tempera-
ture differences within 0.4 °C in all years. In 2001 and 2002,
melting degree-days at Lake Tuborg were very similar to the
1948–2002 Eureka mean melting degree-days, but 2003 Lake
Tuborg melting degree-days were 23% greater.6Much of the
anomalous warmth in 2003 occurred in July. Mean Eureka
air temperature from 3 to 18 July was 4 °C higher than the
1948–2002 mean for the same period. The year 2003 had the
10th warmest summer on record (1947–2003), with 435 melt-
ing degree-days. Rain totalling 37.5 mm fell at Eureka in
2003, the 38th percentile for the 1948–2003 record. Only
28 mm fell at Lake Tuborg during monitoring in 2003, and
no rain fell for more than a month prior to the jökulhlaup.
Lake level, discharge, and jökulhlaup observations
Lake level in all years began increasing in mid-June (Fig. 3)
shortly after mean daily air temperature at lake level in-
creased above freezing. Lake level was unusually high
beginning in early July 2003. In early to mid-July, the ice-
dammed lake overflowed between the Agassiz Ice Cap to the
south and bedrock to the north (Fig. 2). Water flowed over
an ice ledge, eroding boulder-sized pieces of ice. These ice
blocks were then transported to the outwash plain at “ii,”
where they conglomerated into ice rafts (Fig. 2). Sediment-
laden ice rafts detached from shore and floated around the
freshwater basin as lake level continued to rise (Fig. 4a). No
“leaking” (Gilbert 1971) of the ice-dammed lake was ob-
served near the glacier terminus. Lake Tuborg lake level
decreased by -0.75 m between 16 and 22 July (Fig. 3) after
air temperatures fell following their peak on 4 July 2003.7
The ice-dammed lake catastrophically burst, dramatically
raising lake level by 7.6 m from 25 July 08:00 to 28 July
20:00 (Fig. 3). To increase Lake Tuborg lake level by this
amount, 3.2 × 108m3of water is necessary, providing a mini-
mum estimate of the jökulhlaup volume (see Discussion). The
rate of lake level increase requires that average discharge into
Lake Tuborg for this period exceeded 1040 m3s–1.
The jökulhlaup volume can also be estimated assuming the
ice-dammed lake morphology is half an ellipsoid. The length,
© 2007 NRC Canada
794 Can. J. Earth Sci. Vol. 44, 2007
Fig. 2. The Agassiz Ice Cap (AIC), ice-dammed lake (IDL), and Lake Tuborg (LT) from aerial photographs a-16687-47 and a-16977-
63. Dotted lines mark channels that usually feed the northeast basin of LT, and the solid line represents the IDL overflow route. An
“ice raft” in LT on 2 August 1960 (date of the aerial photograph) is circled. Arrows at B and C/D correspond to the locations and di-
rections of photographs shown in Fig. 4. Crosses are the assumed northeast and southwest boundaries of the IDL. Scale is variable due
to parallax, but the mosaic is about 15 km × 9 km, and north is up.
6See supplementary Fig. S2
7See supplementary Fig. S2.
width, and maximum depth of the ice-dammed lake are 6 km,
1.5 km (Fig. 2), and 116 m, producing a volume of 5.5 ×
108m3. Boundaries from 1959 are broadly similar to those of
2003, based on GPS waypoints obtained at the ice-dammed
lake perimeter. Prominent “ice-rims” on the empty ice-dammed
lake walls were observed by helicopter on 28 July (Fig. 4d).
The elevation difference between the uppermost ice-rim and
the bottom of the drained ice-dammed lake was 116 m, pro-
viding the maximum depth.
Discharge issued from a newly developed glacier portal
980 m from the Lake Tuborg shore (Fig. 4b). When the
jökulhlaup ended, the portal was -50 m diameter. The outwash
plain near “ii” prograded, and Lake Tuborg is now almost
completely isolated from the northeast glacier (cf. 1959 photo
in Fig. 2). Englacially eroded ice blocks floated as bergs in
the freshwater basin during and after the event.
Maximum discharge (Qmax) from subglacially draining
jökulhlaups is related to the volume drained (Vt,×10
6m3),
where Qmax = 46(Vt)2/3 (Clague and Mathews 1973; Walder
and Costa 1996). Qmax using the Lake Tuborg lake level Vtis
2200 m3s–1, and Qmax using the Vtfrom ice-dammed lake
morphology is 3150 m3s–1 (see Discussion).
Limnology and sedimentary processes before the
jökulhlaup
The northeast, freshwater basin
Before each melt season began, the upper water column
(epilimnion) was inversely thermally stratified, and bottom
waters (hypolimnion) were relatively warm (e.g., Fig. 5,
30 May cast). The water column was nearly isohaline; bot-
tom water was -50 mg/L denser than near-surface water
(Fig. 5; 30 May cast), and N2at the lower thermocline was
4.1×10
–5 s–2, representing somewhat weak resistance to
mixing (cf. Wüest and Lorke 2003).
During the period when the ice-dammed lake was over-
flowing, the water column was significantly fresher than in
the pre-melt season (Fig. 5, 11 July cast). There was a maxi-
mum density inversion of 72 mg/L. Water samples at 10 m
and near bottom at 30 m had 47 mg/L and 218 mg/L SSC,
respectively.
One day before the jökulhlaup began (24 July 2003), two
CTD casts were performed on the sill and in the freshwater
basin (Fig. 6, stations 5 and 6). Casts were obtained 10 min-
utes apart, and stations are separated by 1 km. There was no
hypolimnion in the freshwater basin. The water column cooled
and freshened with depth, producing a density inversion of
88 mg/L. SSC was >23 mg/L through most of the water col-
umn. By contrast, on the sill, near-bottom water was signifi-
cantly warmer, had higher specific conductivity, and was
much denser than in the freshwater basin. SSC only reached
16 mg/L ina5mthick overflow (Fig. 6, station 6).
The southwest, meromictic basin
Before the jökulhlaup in the saltwater basin, specific con-
ductivity in the upper freshwater layer (mixolimnion) was
-900 µS/cm, and specific conductivity in the deeper saltwater
layer (monimolimnion) was -41 850 µS/cm (-25.01 PSU;
Fig. 7). The extremely sharp and strong chemocline creates
an abrupt -20 000 mg/L density contrast between salt and
fresh water, and -0.43 s–2 N2. Wüest and Lorke (2003) cite
0.1 s–2 as a typical maximum N2in lakes. Before the 2003
melt season began, dissolved oxygen was 10.5–12 mg/L from
the surface to 58 m, and dropped to <1 mg/L below 60.5 m.
A sharp decrease to -5% Tzconsistently occurs near the
chemocline (Figs. 7, 8, 9). Seven Van Dorn water samples
were obtained from 50–59 m in late May 2003, and SSC
was <2.5 mg/L in every sample. From the Tzcalibration,
SSC should have been extremely high8(see Discussion).
Response to slush flow events
In 2001, 2002, and 2003, small supraglacial lakes on the
Agassiz Ice Cap drained, transporting large volumes of slush
and sediment into Lake Tuborg near “iv” (cf. Braun et al.
2000). From late May to mid-July 2003, CTD casts were re-
peatedly performed close to inflow near “iv” (Fig. 8a, station 8).
Before early July, discharge was only from unglaciated terrain
at lower watershed elevations, and water column turbidity was
<2.5–4 mg/L (Fig. 8a). Beginning on 3 July 2003, discharge
and sediment transport increased dramatically, and strong
interflows formed at 25 and 48 m (Fig. 8a). The upper water
column became increasingly turbid from the bottom up from
3 to 16 July (Fig. 8a).
During peak inflow at “iv”, several casts were also per-
formed farther from the delta, to determine the extent of pro-
cesses derived from slush flows in the meromictic basin
(Fig. 8b, station 9). On 8 July, the transmissometer was
overranged below 30 m (Fig. 8b), but a near-bottom Van
Dorn sample contained 70 mg/L SSC. The next day at the
© 2007 NRC Canada
Lewis et al. 795
Fig. 3. Lake Tuborg 2001–2003 lake levels. Solid lines are auto-
matically logged stilling well data. Circle symbols are lake
heights determined with a staff and level.
8See supplementary Fig. S1.
same location, the water column was significantly less turbid
between 20 and 45 m, and beneath the chemocline (Fig. 8b).
Inflow at “iv” significantly dropped beginning on 13 July,
and continued to wane through the remainder of the melt
season. SSC was extremely high at station 8 (near inflow at
“iv”) until 12 July, then the lower water column began to
freshen (Fig. 8a). The water column cleared from the bottom
up, and freshening continued until profiling ended on
© 2007 NRC Canada
796 Can. J. Earth Sci. Vol. 44, 2007
Fig. 4. Photographs from Lake Tuborg and the ice-dammed lake (IDL) in 2003. (a) Ice raft; arrows point to a large clast and an ice
boulder coated in fine-grained sediment. (b) The jökulhlaup on 28 July 2003, with the portal circled. (c) Full IDL on 18 July.
(d) Largely empty IDL on 28 July. Note the “ice rims” below the arrow. Locations for (b), (c), and (d) are shown in Fig. 2 as B and
C/D. Photos are courtesy of Anders Romundset.
Fig. 5. Freshwater basin CTD casts from before, during, and after the jökulhlaup in 2003. Potential temperature (Tθ), specific conductivity
(SpC), and potential density (σθ).
16 July (Fig. 8a). Specific conductivity below the chemocline
decreased by -100 µS/cm near inflow at “iv” following the
2003 slush flow events.
Limnology during and after the jökulhlaup
During and after the jökulhlaup in the freshwater basin,
near-bottom water samples were very turbid. At station 3 on
26 July, near bottom SSC was 77 mg/L; at station 4 on 28
July, SSC was 1093 mg/L.
After the jökulhlaup in the saltwater basin, a strong over-
flow was present in the epilimnion, extending -9 km from
the beginning of a transect near the sill (Figs. 9a;1c). Close
to the sill, turbid plumes extended as far as -3 km above and
below the chemocline. A very distinct “cold stratum” was
present between -35 m and the chemocline (Fig. 9b; cf.
Phelps 1996). The epilimnion warmed with distance from
the sill.
Pre- and post jökulhlaup temperature and specific con-
ductivity in the meromictic basin are compared in Fig. 10.
The “cold stratum” was not present as late as 16 July 2003
(Fig. 10a). Although the cold stratum was less thermally
dense than the overlying, warmer, epilimnion (Fig. 10a),
the thermal density deficit was overcome by saltier, more
dense water in the lower mixolimnion compared with pre-
jökulhlaup conditions (Fig. 10b). When the post-jökulhlaup
lower mixolimnion is visualized in a vertical section, spe-
cific conductivity is seen to slightly decrease with distance
from the sill (Fig. 11b).
The post-jökulhlaup monimolimnion specific conductiv-
ity was -550 µS/cm lower, and temperature was -0.04 °C
lower compared with pre-jökulhlaup conditions recorded in
June 2001 (Figs. 10c, 10d). As late as 16 July 2003, the
monimolimnion was -2.57 °C, and specific conductivity
was -41 800 µS/cm (Figs. 10c, 10d).
Downlake temperature and specific conductivity variation
in the monimolimnion along the -10 km saltwater basin
transect were extraordinarily low less than two weeks after
the jökulhlaup ended (<88 µS/cm and <0.003 °C).
© 2007 NRC Canada
Lewis et al. 797
Fig. 6. CTD casts from the day before the jökulhlaup began (24 July 2003) in the freshwater basin (station 5) and on the sill (station 6).
Potential temperature (Tθ), specific conductivity (SpC), potential density (σθ), and suspended sediment concentration (SSC).
Fig. 7. CTD cast from the meromictic basin (station 13) before the jökulhlaup (10 June 2001): potential temperature (Tθ), specific
conductivity (SpC), potential density (σθ), Brunt-Väisälä frequency (N2), and beam transmissivity (Tz; note the reversed x-axis).
Dissolved oxygen in the mixolimnion was -1 mg/L higher
in the cold stratum compared with casts from the same loca-
tion earlier in the melt season. The monimolimnion was not
significantly oxygenated by the jökulhlaup.
Near-surface sediments
Before the jökulhlaup, at the end of the period when water
was overflowing from the ice-dammed lake (21 July 2003), a
surface core was obtained at station 4 in the freshwater ba-
sin. At the base of the core, five centimetre-scale units with
coarse bases grade upward (Fig. 12a, V1–V5). Sediment
above -80 mm is somewhat coarser grained and is less no-
ticeably laminated than sediment below 80 mm. However,
three units fine upward, and inter-unit grain size coarsens
upwards (Fig. 12a, F1–F3).
After the jökulhlaup, on 5 August, another surface core was
obtained at the same location (Fig. 12b). Laminae are dis-
tinctly angled below 95 mm, fine upward, and are relatively
coarse. Several dark very fine-grained lenses are present be-
low -120 mm (Fig. 12b, RU), and one has apparently folded
internal structure (Fig. 12b, RU-F). Sediment is nearly mas-
sive above -95 mm, but very gradually fines upward.
Surface cores were also obtained on the freshwater and
saltwater sides of the sill after the jökulhlaup (Figs. 13a,13b,
stations 5 and 7). Both cores are from roughly the same water
depth (62–63 m), and coring stations are 2 km apart. On the
freshwater side, sediments are very fine-grained and dense:
repeated coring with a weighted Ekman dredge failed to re-
cover more than -33 mm (Fig. 13a). In the core from the salt-
water side, there are laminae below -85 mm, but no clay
caps. Texture coarsens upward above 85 mm, but abruptly
fines at about 30 mm, above which sediment is massive and
slightly fines upward (Fig. 13b).
At a much more distal location, four cores were obtained
within -50 m of each other in 2003 in -135 m water depth
(Fig. 14, station 12). All cores, except K, were obtained
before the jökulhlaup. The lowermost 60 mm of the stratig-
raphy (160–220 mm) is largely massive, fine-grained, and
slightly fines upward. Sediments above about 160 mm are
somewhat coarser, irregularly laminated, and of variable grain
size and thickness. Midway through core K, there is a sharp
contact, above which sediments are much more coarse and
fine upward.
Discussion
Jökulhlaup hydrology
Overtopping of the ice-dammed lake likely contributed
greatly to the anomalously high lake levels before the
jökulhlaup in early to mid-July 2003 (Fig. 3). The lake level
© 2007 NRC Canada
798 Can. J. Earth Sci. Vol. 44, 2007
Fig. 8. Limnologic response to the 2003 slush flows. (a) Gridded suspended sediment concentration (SSC) at station 8 near inflow at “iv.”
Contour interval is 2 mg/L at ≤9 mg/L and 4 mg/L at >9 mg/L. (b) Transmissivity (Tz) and SSC farther from inflow at station 9 on 30 June,
8 July, and 9 July 2003. Tzis not transformed to SSC below the chemocline, or when <10%. Tzdecreases to the right.
decrease after mid-July occurred after air temperature9and
glacial melt declined, so less overtopping of the ice-dammed
lake likely occurred. In 2003, rainfall runoff did not signifi-
cantly contribute to ice-dammed lake filling or overtopping,
nor did it trigger the jökulhlaup.
The jökulhlaup was likely triggered by ice-dam flotation,
which theoretically occurs when ice-dammed lake depth ex-
ceeds 90% of the ice-dam height; however, dam overtopping
occurred prior to the jökulhlaup, probably because of dry-
based glacier bed adhesion (Roberts 2005). The event likely
ended when lake level fell below outlet conduits, the engla-
cial tunnel system was sealed by mechanical blockage, or
cryostatic pressure sealed the conduits (Roberts 2005).
Of 125 ice-dammed lakes studied near Expedition Fiord,
Axel Heiberg Island (Fig. 1a), most drained incompletely by
overtopping their dams (Maag 1969). The only other ice-
dam flotation jökulhlaup observed in the High Arctic ended
by ice-dam resettling, and the lake did not completely drain
(Blachut and McCann 1981).
The July 2003 jökulhlaup was the largest recorded in Can-
ada since about 1947 (cf. Clague and Evans 1994; Geertsema
and Clague 2005), and the actual volume drained was likely
more than the two jökulhlaup volume estimates. Volume cal-
culation using the level of Lake Tuborg does not account for
discharge out of Lake Tuborg, which was extreme during the
event. Discharge into Lake Tuborg from sources other than
the jökulhlaup is also not accounted for. However, watershed
snowpack was largely exhausted for about a month prior to
the jökulhlaup, and air temperature fell after 19 July, so dis-
charge at “i,” “iii,” and “iv” was very low. The morphologic
© 2007 NRC Canada
Lewis et al. 799
Fig. 9. Post-jökulhlaup (9 August 2003) vertical sections from stations 9–14 (dotted line in Fig. 1c) in the upper 80 m of the meromictic
basin. (a) transmissivity, Tz, and (b) potential temperature, Tθ. Contour interval is 10% for Tz,and0.1°CforTθ. Casts are vertical dashed
lines. Isolines are very closely spaced at the chemocline temperature gradient.
9See supplementary Fig. S2.
volume estimate is also likely low, since large floating ice
wedges (Maag 1969; Blachut and McCann 1981) exist at both
ends of the ice-dammed lake, and its length was estimated
conservatively (Fig. 2).
The depth of the ice-dammed lake is precise, since ice-
rims on the empty lake walls clearly showed maximum lake
level (Fig. 4d). Ice-rims likely formed as the lake drained,
when ice was draped on wave- or ice-cut terraces created in
ice-dammed lake filling cycles (Maag 1969).
Underestimating the jökulhlaup volume would result in an
underestimation of maximum discharge. However, if the glacier
near the ice-dammed lake is cold based, the calculated maxi-
mum discharge might be too high, since the Clague–Mathews
relationship was determined from warm based jökulhlaups
(Walder and Costa 1996).
Processes and deposits in the northeast, freshwater
basin
Varves typically accumulate in the freshwater basin. In
summer, heightened air temperature raises the Agassiz Ice
Cap freezing level, increases glacial discharge and sediment
flux to Lake Tuborg (Braun et al. 2000), and turbid sediment
plumes disperse sediment through the lake, forming rela-
tively thick and coarse laminae. In winter, fluvial input stops,
and clay-sized particles settle very slowly out of suspension,
forming a thin clay layer that defines an annual couplet. Wa-
ter depth at the freshwater basin coring location is deeper than
wave base (Håkanson and Jansson 1983), and the temperature
gradient between the hypolimnion and metalimnion typically
resists mixing (Fig. 5, 30 May cast), allowing undisturbed ac-
cumulation of varve couplets. This interpretation is consistent
with varve-like structures at the bottom of the pre-jökulhlaup
station 4 core (Fig 12a) and cesium-137 dated percussion
cores from a similar location (Smith et al. 2004).
When water was overflowing from the ice-dammed lake in
mid-July, the deepest parts of the northeast basin were filled
with cold, fresh, and turbid water (Figs. 5, 6). Bottom water
was mixed, removing the weak hypolimnion that was present
in May (Fig. 5). This cold, fresh water could only have been
from the overflowing ice-dammed lake, since no major streams
besides “ii” enter the freshwater basin (Fig. 1b). Water in
the ice-dammed lake was likely near 0 °C, since it is in con-
tact with, and fed almost directly by, the Agassiz Ice Cap, is
partially berg and ice-shelf covered, is perennially ice cov-
ered, and is about 300 m higher than Lake Tuborg (cf. Maag
1969; Gilbert 1971; Blachut and McCann 1981). Dissolved
load and conductivity were also likely low in the ice-dammed
lake, since (1) its largest watersheds are highly glacierized;
(2) glacial meltwater reaching the lake is very briefly in con-
tact with the ground; and (3) it is above marine limit.
Underflows were present in the freshwater basin two weeks
before the jökulhlaup. Their formation was facilitated by the
removal of the thermocline. The 11 and 24 July casts are in-
versely thermally and salinity stratified, producing strong
density inversions (Figs. 5, 6) that could only have been
compensated for by suspended sediment. This interpretation
is supported by highly turbid near-bottom water samples
from the freshwater basin while the ice-dammed lake was
being overtopped.
Pre-jökulhlaup underflow deposition is recorded in the up-
per part of the freshwater basin surface core (Fig. 12a).
Fining upward cycles above 80 mm reflect periods of wan-
ing sediment transport and deposition, and were possibly de-
posited diurnally. Intra-cycle coarsening upward reflects an
increase in energy with time, as discharge from the over-
flowing ice-dammed lake likely increased in early to mid-
July (Fig. 3). Amazingly, more sediment was deposited in
the deepest part of the freshwater basin in the days to weeks
leading up to the jökulhlaup than was deposited in the four
preceding years (Fig. 12a). Ice-rafted debris rainout also oc-
© 2007 NRC Canada
800 Can. J. Earth Sci. Vol. 44, 2007
Fig. 10. Temperature and specific conductivity before and after
the jökulhlaup in the saltwater basin. Potential temperature (Tθ;
(a,c)) and specific conductivity (SpC; b,d) in the meromictic
basin before (grey shading) and after (black shading) the
jökulhlaup. Shaded regions encompass maximum and minimum
values for each series; solid lines are means. The grey-shaded pre-
jökulhlaup series consists of 24 casts from 8 to 24 June 2001,
and the black-shaded post-jökulhlaup series consists of seven
casts (one from 7 August, and six from 9 August 2003). Post-
jökulhlaup casts were completed after lake level regressed, and
lake level was very similar in both series (Fig. 3). The thick
dashed line is the last cast in the meromictic basin prior to the
jökulhlaup, completed on 16 July 2003. Note that abscissa scales
on sub-panels showing monimolimnion parameters (c,d) are very
different from abscissa scales showing mixolimnion parameters
(a,b) to better illustrate subtle post-jökulhlaup cooling and
freshening of the monimolimnion.
curred (Fig. 4a). Importantly, the sill separating the saltwater
and freshwater basins confined underflows to the freshwater
basin prior to the jökulhlaup (Fig. 6, station 6 cast).
Underflows in the freshwater basin during and after the
jökulhlaup may initially have been erosive. An unknown
amount of sediment separates the pre- and post-jökulhlaup
cores at station 4 (Figs. 12a, 12b). Fine-grained lenses at the
base of the post-jökulhlaup core are likely rip-up of semi-
consolidated sediment; angled laminae may be cross beds;
and texture is coarser than pre-jökulhlaup sediment at the
same site (Fig. 12) pointing toward deposition from bottom
currents. Closer to the sill, jökulhlaup-derived erosive under-
flows exposed dense fine-grained sediments (Fig. 13a, sta-
tion 5), probably as flow was constricted, channelized, and
accelerated near the southwest end of the freshwater basin
(Gee et al. 2001).
Processes and deposits in the southwest, saltwater basin
The monimolimnion is warmed by small amounts of solar
radiation that reaches the chemocline through the ice cover
(Ludlam 1996; Van Hove et al. 2006). The transmissivity de-
cline at the chemocline (Figs. 7, 8b,9a) is probably caused
by dissolved sulfur associated with a microbial community
or an iron-oxide precipitate (Belzile et al. 2001).
Varve formation would be promoted in the meromictic
basin by the anoxic monimolimnion, the high N2at the
pycnocline, and high circulation rates (Fig. 14, MPJ) relative
to many other High Arctic varved lakes. However, it is pos-
sible that all fine-grained sediment flocculates and is depos-
ited before winter, preventing the regular formation of clay-
caps (Gilbert 2000).
During and after the jökulhlaup, cold ice-dammed lake
water flowed through the freshwater basin, overtopped the
34 m sill, then flowed along the chemocline and reached the
southwest end of the lake (Figs. 9b, 10a). Flow on boundary
layers promotes the development of internal waves that can
eventually oversteepen, forming Kelvin-Helmholtz billows.
Billows form when the gradient Richardson number (Ri=
N2/(du/dz)2) is <0.25 (du/dz is velocity shear; McCool and
Parsons 2004). Billows rapidly collapse due to internal insta-
bility, resulting in a thickened boundary layer (Geyer and
Smith 1987). This interpretation is consistent with the post-
jökulhlaup chemocline thickening (Fig. 10b), and the decrease
in conductivity with distance slightly above the chemocline
(Fig. 11b). Flow at the chemocline would have waned with
distance, and distal billows would have had smaller length
scales. In the Fraser River estuary, chemocline thickening
has been recorded under similar physical conditions, and
Kelvin-Helmholtz billow length scales decreased distally
(Geyer and Smith 1987).
In the meromictic basin, extremely high shear is required
for Rito drop below 0.25 because of the exceptionally high
N2at the chemocline. This is why the chemocline thickening
was subtle (Fig. 10b), and monimolimnion cooling and fresh-
ening was minimal (Figs. 10c,10d).
The Lake Tuborg chemocline remains anomalously sharp
compared with most other High Arctic meromictic lakes
(Ludlam 1996; Van Hove et al. 2006) because flows along
the chemocline during jökulhlaups disturb vertical diffusion
by molecular diffusion and wind-generated mixing (Toth and
Lerman 1975; Van Hove et al. 2006).
The depth of the Lake Tuborg chemocline is also regionally
anomalous, and was thought to be controlled by the ice-contact
depth (Ludlam 1996), but shallow bathymetry at both ends of
the lake discounts this possibility (Fig. 1c). Rather, chemocline
depth is a function of the sill depth, shear, N2, and turbidity.
© 2007 NRC Canada
Lewis et al. 801
Fig. 11. Specific conductivity (SpC) vertical sections from after the jökulhlaup on 9 August 2003 (stations 9–14; dotted line in
Fig. 1c) at two depth ranges. Contour interval is 50 µS/cm in (a), and 250 µS/cm in (b). Note the greater vertical exaggeration for the
lower section. Vertical dashed lines are CTD casts.
Interflows progressed much less far into the meromictic ba-
sin than overflows (Fig. 9a). Kelvin-Helmholtz billows would
have mixed turbid fresh water with salt water, allowing
fine-grained particles to flocculate and drop from suspension.
Jökulhlaup-derived sedimentary processes near the sill in
the meromictic basin are inferred from the station 7 core
(Fig. 13b). The bottom slope is relatively steep, and the
chemocline is only 7 m above the lake bottom at this site.
Several processes could have deposited the thin, coarse laminae
in the core: (1) Downward propagating internal waves could
have redistributed sediment. Little energy is required to do
this, since the monimolimnion has near zero N2(Fig. 7;
Smyth, personal communication). (2) A slump could have
been triggered either by internal waves on the chemocline
hitting the sill (Rimoldi et al. 1996) or by freshly deposited
sediment oversteepening the sill. (3) Settling-driven convec-
tion. Hyperpycnal flows could have been generated as parti-
cles accumulated at the chemocline until their concentration
exceeded the density of the monimolimnion (Hoyal et al.
1999; Parsons et al. 2001; McCool and Parsons 2004). It is
noted that double-diffusive convection (Parsons et al. 2001)
was not possible at the chemocline because the cold stratum
was superimposed on the warm monimolimnion—a thermally
stable situation at temperatures less than maximum water den-
sity. (4) “River-generated” turbidity currents with interstitial
fresh water. When turbidity currents were insufficiently thick
to overtop the sill, they would have suddenly been restricted
to the freshwater basin, and saltwater basin deposition would
have been solely from interflows and overflows (Chikita et al.
1996). This is a plausible explanation for the abrupt decrease
in grain size above 30 mm depth (Fig. 13b). However, clas-
sic river-generated turbidity currents with interstitial fresh
water require extremely high SSC to overcome density defi-
cits created by ambient salt water, and are globally very rare
(Gilbert 2000; Mulder et al. 2003). At Lake Tuborg, turbid-
ity current SSC would have to exceed 20 000 mg/L for SSC
to exceed this density deficit (Fig. 7). Although the most
turbid near-bottom water sample in the fresh water basin
during the jökulhlaup only contained 1100 mg/L, fluvial
SSC in jökulhlaups has exceeded 200 000 mg/L (Mulder et
al. 2003).
The process(es) responsible for the laminae in the station
7 core did not progress into the deep, distal parts of the salt-
water basin. Interflows were not spatially extensive in the
© 2007 NRC Canada
802 Can. J. Earth Sci. Vol. 44, 2007
Fig. 12. Ekman cores obtained from the same location before and after the jökulhlaup. Coring location is station 4, depth is 74 m.
(a) Thin sections and grain size from a core obtained on 21 July 2003. Sediment that was likely deposited during the period of ice-
dammed lake overflow is shaded. V1–V5, clay-caps; F1–F3, bottoms of fining-upward cycles. (b) Thin sections and grain size from a
core obtained on 5 August 2003. RU, rip-up; RU-F; folded rip-up. Images are flatbed scans of thin sections in cross-polarized light,
with contrast and brightness enhanced to better illustrate structures. EDD, equivalent disk diameter (Francus 1998).
saltwater basin after the jökulhlaup (Fig. 9). The extent of
underflows is unknown, but near-bottom processes are in-
ferred from surface core K (Fig. 14). The coarse, mostly
massive, fining upward cap (Fig. 14, J03) was deposited
during and after the jökulhlaup, as it is not present in cores
F, A, or B. The cap was deposited non-erosively, and under-
lying sediments were not disturbed (Fig. 14). The -90mbe
-
tween the chemocline and lake bottom would minimize the
potential for erosion by downward propagating internal
waves. Slump and hyperpycnal flow progression to the cor-
ing site is unlikely because of (1) the great distances be-
tween the coring location, sill, and jökulhlaup inflow; (2) the
1° average bed slope between the sill and saltwater basin;
and (3) the high monimolimnion salinity and density.
The nearly massive structure of the K cap points to depo-
sition by sediment rainout from hypopycnal flow. Sediment
was likely rapidly deposited at the site from the upper
monimolimnion and lower mixolimnion by flocculated sed-
iment that settled faster than rates predicted by Stokes Law.
Supraglacial lake drainings above “iv”
Repeated slush flow events are capable of very slightly
freshening the monimolimnion during peak glacial melt in
early July (Fig. 10d; 16 July 2003 cast). However, sediment
plumes associated with these events are not as extensive as
jökulhlaup-derived plumes (cf. Figs. 8, 9a). Supraglacially
derived plumes are not capable of producing the distinctly
thick, nearly massive deposit in the post-jökulhlaup core at
station 12 (Fig. 14, J03), nor of producing the distinct post-
jökulhlaup freshening, cooling, and mixing that occurred after
the jökulhlaup (Figs. 9, 10, 11).
Supraglacial lakes above “iv” are small compared with the
ice-dammed lake. The water volume discharged in the entire
45-day 1995 monitoring season (Braun et al. 2000) was 60
and 105 times less than the two ice-dammed lake estimates
of the jökulhlaup volume, but only about 20% of the 1995
discharge occurred during slush flow events. Qmax in 1995
was -16 m3s!1near “iv,” compared with 2200–3150 m3s–1
from the jökulhlaup.
Previous jökulhlaups
The ice-dammed lake drained in 1993 (R.M. Koerner, per-
sonal communication, 1996). Detailed measurements were
next made at Lake Tuborg in 1995, when a cold stratum re-
mained in the saltwater basin (Phelps 1996). Observations in
1995 included kettle holes, a glacier portal near “ii,” and a
© 2007 NRC Canada
Lewis et al. 803
Fig. 13. Ekman cores from both sides of the sill obtained after the jökulhlaup (11 August 2003). Both cores were taken in 62–63 m water
depth. (a) Thin sections and grain size of a surface core obtained on the freshwater side of the sill at station 5. (b) Thin sections and
grain size of a surface core obtained on the saltwater side of the sill at station 7. Images are flatbed scans of thin sections in cross-
polarized light, with contrast and brightness enhanced to better illustrate structures. EDD, equivalent disk diameter (Francus 1998).
© 2007 NRC Canada
raised shoreline on marine mud terraces on the north shore
of the freshwater basin (M.J. Retelle, personal communica-
tion, 2005). It is likely that the lower fining upward se-
quence in the stratigraphy (Fig. 14, PJ) is the upper part a
deposit from the 1993 jökulhlaup.
A jökulhlaup also occurred between 1960 and 1963. Air
photos from August 1960 show a full, overflowing ice-
dammed lake, and an ice-raft near “ii” (Fig. 2). Bergs were
observed at Lake Tuborg in 1963 (Hattersley-Smith and
Serson 1964). There was no cold stratum in 1963 (Hattersley-
Smith and Serson 1964), so the jökulhlaup likely occurred
closer to 1960 than 1963.
Conclusions
A large and rare jökulhlaup drained into Lake Tuborg in
July 2003 while a detailed lake process study was being con-
ducted. The jökulhlaup drainage style, and the effects on the
limnology and sedimentology of Lake Tuborg have been
described.
Before the jökulhlaup, a weak thermally derived hypo-
limnion was removed in Lake Tuborg’s proximal freshwater
basin. This facilitated the formation of hyperpycnal flows.
During and after the jökulhlaup, underflows eroded the
lake bottom in parts of the freshwater basin, exposing
804 Can. J. Earth Sci. Vol. 44, 2007
Fig. 14. Stratigraphy of Ekman cores and grain size from a -135 m deep, distal location in the meromictic basin (station 12). All cores, ex-
cept K, were obtained before the jökulhlaup in 2003. J03, sedimentation from the 2003 jökulhlaup; MPJ, minimum 2003 sedimentation prior
to the jökulhlaup; PJ, likely the top of the previous jökulhlaup deposit. The sediment surface in A and F was purposely lost to retrieve cored
sediments deeper than -13 cm (the length of the subsampling tubes). The last core (except K) obtained was B, on 24 June 2003. Images are
flatbed scans of thin sections under cross-polarized light, with contrast and brightness enhanced to better illustrate structures. EDD, equivalent
disk diameter (Francus 1998).
dense, laminated sediments, and creating an unconformity.
However, some areas of the freshwater basin experienced
high sedimentation rates, and cross-beds and fine-grained
rip-ups were deposited.
Alternatively, in the saltwater basin, underflows were limited
to steep and proximal areas close to the sill. More distally,
jökulhlaup-derived sedimentary processes were not erosive,
and a unique thick, coarse, fining upward facies was depos-
ited, probably because of the low bottom slope and the great
depth between the chemocline and lake bottom. Underflow
run-out distances were limited by rapid deposition from salt-
water flocculation.
The monimolimnion has remained remarkably salty for
-3 ka, and through at least three jökulhlaups (1960–1963,
1993, 2003). Mixing of the monimolimnion has been limited
by the extremely high N2at the chemocline.
Monitoring has also allowed the sedimentary signal and
limnologic consequences of jökulhlaups to be disentangled
from those of slush flows from supraglacial lakes (Braun
et al. 2000). It should therefore be possible to unambigu-
ously identify jökulhlaup deposits in the long core record,
and produce a millennial-scale undisturbed reconstruction of
jökulhlaup frequency, which would test long-term models of
jökulhlaup frequency and magnitude (Clague and Evans
1994).
Acknowledgments
National Science Foundation (NSF) grant ATM-9708071,
NSF Doctoral Dissertation Research Improvement Award
0221376, Geological Society of America graduate student
grants, an Arctic Institute of North America grant-in-aid, and the
Gloria A. Radke prize from the University of Massachusetts
supported this research. The Polar Continental Shelf Project
(PCSP) and VECO Polar Resources, Littleton, Colorado
provided outstanding logistical support. PCSP was especially
responsive to rapidly evolving conditions in the field during
the jökulhlaup. Lesleigh Anderson, James Bradbury, David
Mazzucchi, Joe Rogers, Anders Romundset, and Chloë Stuart
provided assistance in the field. Robert Gilbert and Bill
Smyth provided valuable advice on mixing processes. Com-
ments by Garry Clarke and an anonymous reviewer are
greatly appreciated.
References
Belzile, C., Vincent, W.F., Gibson, J.A.E., and Van Hove, P. 2001.
Bio-optical characteristics of the snow, ice, and water column of
a perennially ice-covered lake in the High Arctic. Canadian
Journal of Fisheries and Aquatic Science, 58: 2405–2418.
Blachut, S.P., and McCann, S.B. 1981. The behavior of a polar ice-
dammed lake, Ellesmere island, N.W.T., Canada. Arctic and Al-
pine Research, 13: 63–74.
Braun, C., Hardy, D.R., and Bradley, R.S. 2000. Hydrological and
meteorological observations at Lake Tuborg, Ellesmere Island,
Nunavut, Canada. Polar Geography, 24: 83–97.
Chikita, K.A., Smith, N.D., Yonemitsu, N., and Perez Arlucea,
M. 1996. Dynamics of sediment-laden underflows passing over
a subaqueous sill; glacier-fed Peyto Lake, Alberta, Canada.
Sedimentology, 43: 865–875.
Clague, J.J., and Evans, S.G. 1994. Formation and failure of natu-
ral dams in the Canadian Cordillera. Geological Survey of Can-
ada, Bulletin 464, Natural Resources Canada, Ottawa, Ont.
Clague, J.J., and Mathews, W.H. 1973. The magnitude of jökulhlaups.
Journal of Glaciology, 12: 501–504.
Francus, P. 1998. An image analysis technique to measure grain-size
variation in thin sections of soft clastic sediments. Sedimentary
Geology, 121: 289–298.
Francus, P., and Asikainen, C.A. 2001. Sub-sampling unconsolidated
sediments: A solution for the preparation of undisturbed thin-
sections from clay-rich sediments. Journal of Paleolimnology, 26:
323–326.
Gee, M.J.R., Masson, D.G., Watts, A.B., and Mitchell, N.C. 2001.
Passage of debris flows and turbidity currents through a topo-
graphic constriction: Seafloor erosion and deflection of flow
pathways. Sedimentology, 48: 1389–1409.
Geertsema, M., and Clague, J.J. 2005. Jökulhlaups at Tulsequah
Glacier, northwestern British Columbia, Canada. The Holocene,
15: 310–316.
Geyer, W.R., and Smith, J.D. 1987. Shear instability in a highly strati-
fied estuary. Journal of Physical Oceanography, 17: 1668–1679.
Gilbert, R. 1971. Observations on ice-dammed Summit Lake, Brit-
ish Columbia. Journal of Glaciology, 10: 351–356.
Gilbert, R. 2000. Environmental assessment from the sedimentary
record of high-latitude fiords. Geomorphology, 32: 295–314.
Håkanson, L., and Jansson, M. 1983. Principles of lake sediment-
ology. Springer-Verlag, New York, N.Y.
Hattersley-Smith, G., and Serson, H. 1964. Stratified water of a
glacial lake in northern Ellesmere Island. Arctic, 17: 108–111.
Hoyal, D.C.J.D., Bursik, M.I., and Atkinson, J.F. 1999. Settling-
driven convection: A mechanism of sedimentation from strati-
fied fluids. Journal of Geophysical Research, 104: 7953–7966.
Long, A. 1967. Age of trapped sea-water at the bottom of Lake
Tuborg, Ellesmere Island, N.W.T. Transactions, American Geo-
physical Union, 48: 136.
Ludlam, S.D. 1996. The comparative limnology of High Arctic,
coastal, meromictic lakes. Journal of Paleolimnology, 16: 111–131.
Maag, H. 1969. Ice dammed lakes and marginal glacial drainage
on Axel Heiberg Island, Axel Heiberg Island research reports.
McGill University, Montréal, Que.
McCool, W.W., and Parsons, J.D. 2004. Sedimentation from buoy-
ant fine-grained suspensions. Continental Shelf Research, 24:
1129–1142.
Mulder, T., Syvitski, J.P.M., Migeon, S., Faugeres, J., and Savoye,
B. 2003. Marine hyperpycnal flows: Initiation, behavior and re-
lated deposits. A review. Marine and Petroleum Geology, 20:
861–882.
Parsons, J.D., Bush, J.W.M., and Syvitski, J.P.M. 2001. Hyperpycnal
plume formation from riverine outflows with small sediment con-
centrations. Sedimentology, 48: 465–478.
Phelps, K.J. 1996. Laminated lacustrine sediments from glacial
Lake Tuborg, northern Ellesmere Island, Canada. Undergraduate
thesis, Department of Geology, Bates College, Lewiston, Me.
Reedyk, S., Cooke, S., Burchett, G., and Prepas, E.E. 1997. Design
for an inexpensive continuous digital output water level recorder.
Water Resources Research, 33: 1523–1526.
Rimoldi, B., Alexander, J., and Morris, S. 1996. Experimental tur-
bidity currents entering density stratified water: Analogues for
turbidites in Mediterranean hypersaline basins. Sedimentology,
43: 527–540.
Roberts, M.J. 2005. Jökulhlaups: a reassessment of floodwater flow
through glaciers. Reviews of Geophysics, 43: RG1002,, 21 p.,
doi:10.1029/2003RG000147.
© 2007 NRC Canada
Lewis et al. 805
Schlitzer, R. 2005. Ocean data view [online]. Available from
www.awi-bremerhaven.de/GEO/ODV [accessed 7 October 2006].
Smith, S.V., Bradley, R.S., and Abbott, M.B. 2004. A 300 year re-
cord of environmental change from Lake Tuborg, Ellesmere Is-
land, Nunavut, Canada. Journal of Paleolimnology, 32: 137–148.
Toth, D.J., and Lerman, A. 1975. Stratified lake and oceanic
brines: Salt movement and time limits of existence. Limnology
and Oceanography, 20: 715–728.
Van Hove, P., Belzile, C., Gibson, J.A., and Vincent, W.F. 2006.
Coupled landscape–lake evolution in High Arctic Canada. Cana-
dian Journal of Earth Sciences, 43: 533–546.
Walder, J.S., and Costa, J.E. 1996. Outburst floods from glacier-
dammed lakes: The effect of mode of lake drainage on flood
magnitude. Earth Surface Processes and Landforms, 21: 701–723.
Wüest, A., and Lorke, A. 2003. Small-scale hydrodynamics in lakes.
Annual Review of Fluid Mechanics, 35: 373–412.
© 2007 NRC Canada
806 Can. J. Earth Sci. Vol. 44, 2007
