15
Chapter 2 - Literature Review
2.1
Introduction
The inference of hydroclimatic conditions from lacustrine sediments has
been attempted with varying techniques and success by numerous
researchers (e.g. Gilbert, 1975; Perkins and Sims, 1983; Leonard, 1985;
Hardy et al., 1996).
Lakes are end-members of their catchments, so a
detailed account of lacustrine sedimentation is not the only information
required to improve paleoenvironmental interpretations.
Within a given
watershed, temporal and spatial variability in climate, weather, geology,
physiography and hydrology are vital controllers of lacustrine sedimentation.
A better understanding of the processes governing these environmental
variables will lead to more confident and accurate paleoenvironmental
assessments (Bradley et al., 1996; Gilbert et al., 1997).
This
review
begins
by
briefly
describing
the
bedrock
geology,
physiography, glaciation, modern climate, vegetation, terrestrial landforms
and paraglacial activity in the High Arctic. These variables are emphasized
where they are important to the physical lacustrine environment.
Next,
differences between nival and proglacial hydrology are discussed. The role of
the “rival signal” and the impact of glaciers are discussed in relation to
lacustrine and fluvial hydrology and suspended sediment concentration
(SSC). Limnological processes such as stratification, underflow, interflow,
overflow and homopycnal flow are investigated, and the potential importance
of each type to arctic proglacial sedimentation is assessed.
16
Studies of arctic paleolimnology are comparatively rare, as are arctic
glacial, fluvial, and limnological process studies. The majority of published
reports describe arctic watersheds with little glacial cover and low organic and
clastic sedimentation rates (Doran, 1993; Davidge, 1994; Abbott and Stafford,
1996; Bradley et al., 1996).
Therefore, relevant studies from temperate,
montane and alpine watersheds are reviewed as well.
2.2
Geology and physiography
The Canadian Arctic can be divided into three physiographic regions: the
Canadian Shield, the Arctic Platform and the Innuitian Tectonic Province.
Canadian Shield rocks outcrop on Western Ellesmere and Devon islands,
Baffin Island, Boothia Peninsula, Prince of Wales Island, and central Victoria
Island.
These are Proterozoic sedimentary, volcanic and metamorphic
lithologies (Trettin, 1991). Shield rocks have been tectonically inactive for
long periods of time, and usually form rolling highlands and uplands (Dawes
and Christie, 1991).
Overlying the Shield, the Arctic Platform includes much of the central High
Arctic, including eastern Devon, northwest Baffin, Somerset, Victoria and
Banks Islands. The platform is composed of Phanerozoic sedimentary rocks
of mostly marine origin, that have not been folded or faulted on a regional
scale, and usually form plains and plateaus (Dawes and Christie, 1991;
Trettin, 1991).
In the Innuitian Tectonic Province, accumulations of sedimentary rocks,
periods of volcanism, and tectonic activity have caused subsidence, folding
17
and faulting.
This province is composed of rocks of the Sverdrup Basin,
which outcrop on many of the low-lying central High Arctic Islands.
Orogenies such as the Ellesmerian and Eurekan have folded and thrust up
Innuitian rocks, creating highlands on Ellesmere and Axel Heiberg Islands,
with mountains higher than 2200 m ASL (Trettin, 1991). Thirty-four percent of
the Innuitian Province is mountainous (Maxwell, 1980).
2.3
Glaciation
Until recently, the extent of glacial coverage during the Last Glacial
Maximum (LGM) was contested. It was unknown whether a continuous ice
mass existed (called the Innuitian Ice Sheet), or whether glaciers extended
only short distances beyond their present borders (called the Franklinian Ice
Complex; Blake, 1970; England, 1976; Hodgson, 1989). Based on the dating
of moraines, meltwater channels and erratics in the Nares Strait region, and
glaciological evidence from Devon Island, the Innuitian Ice model is currently
favoured (Dyke, 1999; England, 1999).
The technique of dating molluscs, bowhead whales (Balaena mysticetus),
and driftwood logs on raised beaches has been used to reconstruct Holocene
sea level changes following deglaciation in the Arctic Archipelago.
Deglaciation began in eastern coastal areas ca. 10 ka BP (14C-years; Dyke,
1998), while a maximum emergence of approximately 140 m has occurred in
the Bathurst Island and northern Eureka sound areas (Hodgson, 1989).
Although the pattern of uplift can be modeled under both the Franklinian Ice
Complex and Innuitian Ice Sheet models, it is currently assumed the area of
18
maximum uplift represents the area that was underlain by the thickest part of
the Innuitian Ice Sheet (Dyke, 1998).
Remnants of LGM ice have been cored through the modern Devon and
Agassiz ice caps; though poorly dated, basal ice is assumed to be older than
100 ka (Hodgson, 1989). The paleoclimatic record at Devon Island Ice Cap
(DIIC) has been reconstructed for the last 5.3 ka using δ18O (Paterson et al.,
1977), and melt layers for the last 100 years (Koerner, 1977), providing a
unique record of High Arctic climate change. The δ18O profile is influenced
not only by changes in climate, but also by changing atmospheric circulation,
sea ice extent, and changing glacial mass balance (Paterson et al., 1977;
Bradley,
1990).
Nevertheless,
heavy
isotope
depletion
at
the
Holocene/Pleistocene boundary, enrichment during the mid-Holocene climatic
optimum, and an enrichment during the Little Ice Age all point to the likelihood
that climate change is recorded in δ18O in ice cores recovered from the
summit of the DIIC.
2.4
Climate and vegetation
In general, the length of the ablation season in the Arctic is extremely
short, with the shortest seasons in coastal areas, particularly near the Arctic
Ocean, Baffin Bay and Amundsen Gulf. In the Eureka Sound intermontane
area, the ablation season is extended by 15-20 days relative to other regions
of the High Arctic, because it is in the lee of the Princess Margaret Range of
Axel Heiberg Island (Edlund and Alt, 1989).
19
Arctic circulation is dominated by a persistent cold-cored westerly
circumpolar vortex, which is centered over the central arctic islands in winter,
creating constant cold temperatures.
In the Arctic Archipelago during
summer, air mass influence is usually more marine than arctic, but
temperatures remain cool, except during late June, July and early August
(Maxwell, 1980).
The Arctic Archipelago is classified as an arctic desert or semi-desert
(Bradley, 1990). The primary control on precipitation is the frequency and
intensity of cyclonic activity, with secondary controls being topography and
mean atmospheric flow (Edlund and Alt, 1989). Thus, temperature patterns
do not coincide with precipitation patterns. Precipitation is almost always less
than 17 cm/a, and most precipitation occurs in a small number of synoptic
events (Bradley, 1990). Annual precipitation maximums occur in August and
November.
The interpretation of regional-scale climatic information in the High Arctic
is constrained because of the low density and short operational histories of
meteorological stations. Systematic collection of climatic data in the Arctic
Archipelago did not begin until 1947 (Maxwell, 1980). The few stations that
do exist tend to be located in low elevation coastal areas (Hardy and Bradley,
1997).
Vegetation patterns coincide with mean cloud and temperature regimes.
Warmer, intermontane areas such as Eureka Sound have the highest plant
diversities, while areas influenced by air masses from the Arctic Ocean in the
20
north, central, and northwest regions have shorter growing seasons and lower
biodiversities.
Plateaus and highlands in the southern islands have
intermediate biodiversities (Edlund and Alt, 1989).
2.5
Terrestrial sedimentation in the Arctic
2.5.1
Mechanical and chemical erosion and deposition
Frost shattering is the dominant form of mechanical weathering in much of
the Arctic, and sediment created by this process can sometimes be identified
by sedimentary texture (Gilbert and Church, 1983).
Chemical weathering is important in softer carbonate terrain, but is usually
less important than at temperate latitudes. In granitic terrain, oxidation does
occur, but chemical weathering is minimal, and is limited to magnetite, augite
and biotite (Church, 1972).
Chemical deposition in arctic lakes is important when inputs of clastic
sediment are low, and when lakes are situated in small watersheds. Iron-rich
bands are likely created when iron is deposited, then enters the reduced zone
in the subsurface. Next, the iron migrates to the redox boundary, and is
precipitated when it is oxidized in the upper sediments (Gilbert and Church,
1983).
2.5.2
Aeolian activity
In most arctic basins, aeolian deflation is an important proglacial process
only during winter, when winds are high and precipitation is low (Svendsen et
al., 1989; McKenna Neuman, 1990).
Transport of fine, unconsolidated
sediment by wind is facilitated by impacts from snow. Transport is limited
21
during summer because of lower wind velocity, a flooded sandur, and high
lake levels. A lag develops in the source area when light summer winds
separate fine grains from coarse (McKenna Neuman, 1990).
Where aeolian transport occurs in winter, an interstratifiied niveo-aeolian
deposit would be expected (Lewkowicz and Young, 1991). This situation
would be common in areas of rolling physiography, high katabatic winds, high
sublimation, and/or low snowfall. Alternatively, in calm, flat areas where snow
cover is continuous, and is not removed by winter wind scour, aeolian activity
may not begin until the early melt period. In these cases, deposits of aeolian
sand appear as a veneer on top of the snowpack (Lewkowicz and Young,
1991).
2.5.3
Periglacial landforms and processes
Arctic periglacial landforms include talus and rock debris slopes,
blockfields, frost-heaved terrain, patterned ground, frost cracks, ground ice
phenomena and gelifluction slopes (Church, 1972). Since many lakes are
located in valley bottoms, sediment is often delivered to lakes by processes
associated with these landforms.
Colour Lake, Axel Heiberg Island, is in an unglacierized basin that
receives significant sediment inputs from gelifluction (Doran, 1993).
Gelifluction was estimated to contribute 15 – 30 % of the total sediment
budget of the lake, although the actual value is probably slightly higher, since
the influence of gelifluction on fluvial sedimentation was not quantified. On
slopes surrounding Colour Lake (5 – 30o), soil moved downslope between 7
22
and 14 mm/a. Gelifluction would be expected to be more important in arctic
watersheds with relatively steep slopes, abundant clay and sand, and a thick
and saturated active layer.
Gravity flows on the steep slopes of Baffin Island fiords occur because of
frost wedging, heavy rain, persistent water flow, thermal change, wind, snow
and ice fall, vegetation wedging, chemical weathering, and possibly tectonic
activity (Church et al., 1979).
Processes of gravity flow include rockfalls,
rockslides, debris flows, and creep, forming a variety of landforms as a result.
These include talus sheets and cones, debris cones and alluvial cones
(Church et al., 1979). On Baffin Island, the majority of rockfalls occur early in
the melt season, and on south-facing slopes as a result of springtime snow
movement and freeze-thaw cycles.
2.5.4
Paraglacial activity and the “geologic norm”
Much of the periglacial erosion in the Arctic occurs when previously
deposited sediment is reworked by a secondary process (Church, 1972).
Numerous landforms of unconsolidated material are remnants of Pleistocene
glaciation, and provide an important source of remobilized sediment. These
landforms can include drumlins, kames, outwash fans, and moraines, which
are exposed to modern mechanical, chemical, fluvial, and glacial processes
of erosion. It has been hypothesized that because of the prevalence of these
“old” deposits, contemporary sediment yields have no relation to the present
rate of sediment production. The concept of a “geologic norm” has been
introduced to explain conditions of sediment availability.
Conditions of
23
sedimentation in the later Holocene are heightened with respect to the norm
because of the availability of older deposits for “upstream” erosion and
“downstream” deposition (Church and Ryder, 1972). This has been termed
the “paraglacial cycle” (Church and Slaymaker, 1989). Variables affecting the
geologic norm include basin size, geology, climate, vegetation and regional
uplift. Large basins that have been unglacierized for a long time, with an
abundance of easily eroded rocks, short melt seasons, sparse vegetation,
and small rates of isostatic rebound would be expected to be near their
geologic norms (Church and Ryder, 1972).
Often, a geologic norm may be difficult to characterize, since processes
constantly change over the short and long term. For example, the avulsion of
a river, which then erodes proglacial deposits, is a quasi-random event that
would cause a heightened short-term denudation rate. This could lead to
misinterpretation of a paleoclimatic signal from lake sediments.
At Nicolay Lake, Cornwall Island, there is evidence of a “paraglacial
sediment wave” that moved downwards through the catchment, and limited
the supply of late glacial sediment until the late Holocene (Lamoureux,
1999a).
2.6
Fluvial and lacustrine hydrology
Characterizing arctic hydrologic regimes as nival or proglacial based only
on the presence or absence of glaciers in a catchment can lead to
misinterpretation of important hydrometeorological processes. For example,
although the Lake C2 watershed on northern Ellesmere Island is
24
approximately 9% glacier covered, the glaciers are cold-based, and the lake
exhibits a nival regime (Hardy, 1996). Also, the Lewis River, which is fed by
the Barnes Ice Cap on central Baffin Island, displayed glacial characteristics
in 1963, while in 1964 hydrologic processes were dominated by snowmelt
and precipitation. This can be explained by high 1963 temperatures, versus
low temperatures and persistent cloud and storms in 1964 (Church, 1972).
2.6.1
Arctic nival hydrology
The idealized High Arctic nival regime is characterized by the majority of
runoff occurring over a period of weeks to months, usually in late June or
early July. Runoff is sustained by snowmelt, and once all snow has melted,
daily lake level fluctuations cease. Lake outlets are sometimes dammed by
snow, raising lake levels well above their summer means. When snow dams
break, a large proportion of the annual flow leaves the basin in only a day or
two.
For example, in 1974 in unglacierized Umingmak Creek on Devon
Island, 95% of the annual discharge (Q) occurred in only two weeks (Marsh
and Woo, 1981).
During summer, lake levels are raised only by rainfall
events and persistent lateral flow from the sides of the basin.
In a study of two nival creeks on the Fosheim Peninsula, Ellesmere
Island, the melt season was divided into four periods: early melt, main melt,
recession and recharge (Lewkowicz and Wolfe, 1994). Significant interannual
differences in Q occurred because of dissimilar winter snow accumulation;
however, the majority of the annual flow always occurred during the main melt
season. For example, in the winter of 1990, much more snow fell than in
25
1991, which delayed and reduced the annual Q peak. In both years, diurnal
fluctuations stopped shortly after peak flow (Lewkowicz and Wolfe, 1994).
Inter-annual variability in the characteristics of discharge is common for both
nival and proglacial regimes, which is of concern given the limited number of
multi-year studies (e.g., Church and Ryder, 1972; Marsh and Woo, 1981;
Lewkowicz and Wolfe, 1994; Woo and Young, 1997).
2.6.2
Arctic proglacial hydrology
Proglacial lakes and rivers can initially possess nival melt signals, but after
all snow has melted, glacial melt usually becomes increasingly important
(Marsh and Woo, 1981). In the idealized arctic proglacial regime, ice and
permanent snow cover delay discharge peaks until late July or August
(Church, 1972). When glacial mass balance is negative, or a glacier is warmbased, the glacial signal is more pronounced than for cold-based glaciers, or
glaciers with positive mass balances.
Interannual changes in mass balance are mirrored by changes in Q;
however, glacial storage of water over several seasons is possible, which
obscures mass balance estimates based on annual flow (Lawson, 1993).
Also, changes in processes of englacial water movement through the ablation
season mean that relatively short-term fluctuations in Q (e.g. weekly or
monthly) are usually not representative of climatic conditions.
In general, glacierized catchments have a moderating effect on annual
streamflows, with increasing glacial coverage increasing this effect.
In a
study of watersheds in the Cascade Mountains, Washington, the effect of
26
varying percent glacial cover has been investigated (Fountain and Tangborn,
1985). It was found that Q peaks from highly glacierized catchments have
their annual peaks coincident with the annual temperature maximum. Also,
the percentage of annual Q that occurred during the summer temperature
peak increased with increasing glacial cover (Fountain and Tangborn, 1985).
Although dependent on individual watershed characteristics, it appears
that Q in most glacierized catchments is dominated by temperature rather
than precipitation (e.g. Leemann and Niessen, 1994a).
This is likely
particularly true in the Arctic, where rainfall is extremely low. Generally, as
the percent glacial coverage of a given watershed decreases, the correlation
between precipitation and runoff increases (Lawson, 1993). Winter snowfall
is also important, since snow cover can delay the timing of ice melt, reducing
total annual glacial melt and runoff (Lawson, 1993).
Unlike nival watersheds, the classic proglacial regime is characterized by
diurnal fluctuations in Q that continue throughout the melt season. This is
true even in the Arctic, where diurnal changes in solar radiation are small
(Church, 1972, Church and Gilbert 1975, Marsh and Woo, 1981). In the early
melt season, englacial and subglacial conduits are poorly developed, albedo
on the glacier is high, and meltwater is retained by snow. Daily Q peaks can
therefore be delayed by several hours, and are commonly more muted than
later in the season (Church and Ryder, 1972; Bennett and Glasser, 1996).
27
2.7
Suspended sediment and discharge in proglacial and nival
regimes
In general, rivers and lakes in glacierized watersheds transport much
more particulate load than nival watersheds (Lawson, 1993).
In glacierized
catchments, the relationship between SSC and Q is not a simple one, since
intraseasonal changes in Q and sediment delivery are important. Generally
SSC is low in winter, highest in the early melt season, and gradually
decreases throughout the summer (Lawson, 1993). This may be a result of
the transport of relatively abundant glacial and proglacial sediment in the
early part of the melt season, which is produced by year round subglacial
erosion. Later in the melt season, deposits of easily transported sediment
would be exhausted, and sediment starvation occurs (Lawson, 1993). By
early to mid August, sediment concentrations begin to fall during the autumn
Q decline (Gilbert, 1975).
At nival Hot Weather Creek, Ellesmere Island, intraseasonal differences in
sediment availability and transport mean the link between SSC and Q is not
simple (Lewkowicz and Wolfe, 1994). For a given Q, SSC was high early in
the melt season, when sediment supply was not limited. By comparison, later
in the season, SSC was comparatively low, since sediment sources were
somewhat limited.
Therefore, SSC concentrations were less related to
transport capacity than sediment supply (Lewkowicz and Wolfe, 1994).
The existence of rhythmites in cores from glacial lake Bondhusvatn,
Norway, has led to a qualitative model for SSC/Q relationships (Østrem and
28
Olsen, 1987). As mentioned above, sediment concentration in glacial lakes is
highest early in the glacial melt season, and decreases through July and
August. Østrem and Olsen (1987) continue by theorizing that this is true even
when later season hydrograph peaks are nearly equal to early season flows.
However, if a later flow exceeds the early peak, SSC may be as high as or
higher than in the early season (Østrem and Olsen, 1987). This model was
proposed by examining historical water level records in relation to interannual
laminae, but is complicated by lacustrine and fluvial processes.
A more
rigorous test would involve a high-resolution process study.
The rising limbs of discharge hydrographs are more typically associated
with higher SSC than falling limbs. Peaks in SSC can precede Q peaks by
several hours (Østrem, 1969; Church and Gilbert, 1975). The diurnal cycle of
SSC variation is usually one of clockwise hysteresis (Church and Gilbert,
1975; Østrem, 1975; Lewkowicz and Wolfe, 1994). This effect has been
linked to two factors: first, sediment deposited after a daily peak would be
resuspended on the rising limb the following day; secondly, small sedimentladen gullies that usually have little flow contribute sediment only during high
Q (Lewkowicz and Wolfe, 1994). From the study of two Swiss glaciers, it has
been determined that short-term variations in SSQ are a function of sediment
availability, whereas seasonal patterns in sediment Q are related to the
development of glacial drainage networks (Lawson, 1993).
At Lake C2, northern Ellesmere Island, a relatively stable lag was found
between Q and solar energy flux, which produced a diurnal Q signal. A non-
29
linear link was also found between suspended sediment concentration and Q.
No lag existed for this correlation. In addition, a good correlation was found
at Lake C2 between SSC, Q, and temperature in 1991 and 1992 (Hardy,
1996).
SSC seems to rise more quickly at ice-proximal areas, where slumping
and resuspension of recent deposits are common (Church, 1972). However,
at low flows, downstream sites have the highest SSC.
It is characteristic for proglacial rivers to possess discrete SSC pulses
resulting from extreme events (Østrem, 1975). Another “noise producer” in
SSC/Q relationships is the periodic lack of sediment from either under the
glacier or in proglacial rivers (Woo and McCann, 1994).
It may only be
possible to produce meaningful relationships where an unlimited and constant
supply of sandy material exists, and unpredictable events are rare (Church,
1972).
In some catchments, quasi-stochastic events such as jőkulhlaups, the
creation of englacial storage areas, and fluvial incision and storage of
unconsolidated deposits may disrupt simple SSC/Q relationships. For
example, in 1974 an englacial cavity breached in the Athabasca Glacier,
causing dramatically increased SSC and generation of a turbidity current in
the downstream lake.
Shaw, 1981).
However, increased Q did not occur (Gilbert and
30
2.8
Lacustrine processes in arctic lakes
Processes of sedimentation in an arctic proglacial lake have never been
investigated. Discussion of likely processes is therefore extended to arctic
lakes with nival regimes, as well as temperate proglacial lakes.
2.8.1
Stratification
Most arctic lakes are classified as cold-monomictic: they never exceed
temperatures of 4oC, and experience only one period of circulation per year
(Wetzel, 1983).
Temperatures stay below the temperature of maximum
density because of limited energy inputs, and the prevalence of permafrost
(except where taliks exist under larger lakes). When an arctic lake overturns
twice per year it is classified as a dimictic lake, as occurs at Char Lake,
Cornwallis Island (Wetzel, 1983).
Amixis describes lakes that do not
circulate, but this occurs rarely in the Arctic, and is likely restricted to areas
with perennial ice cover (Wetzel, 1983). Ten glacial and non-glacial lakes
north of Cumberland sound on Baffin Island were all cold polymictic (continual
circulation as the water column remains close to 4oC; Gilbert and Church,
1983). Only one lake developed thermal stratification, probably because it is
deep and well protected from winds (Gilbert and Church, 1983).
Meromictic lakes are also relatively common in the Arctic.
Ectogenic
meromixis occurs when isostatic uplift isolates coastal regions possessing
basins that contain salt water. Cryogenic meromixis occurs when salt water
is expelled from groundwater in the formation of permafrost. Meromixis can
affect sedimentation by creating a density barrier called a chemocline, along
31
which sediment can travel before deposition.
This can create an even
accumulation of clastic sediment over the lake bottom (Retelle and Child,
1996). Alternatively, sediment can be deposited rapidly when flocculation
occurs in highly saline water, once sediment is entrained in the current-free
monimolimnion.
Anoxic conditions in the monimolimnion preclude the
possibility of bioturbation.
2.8.2
The role of lake ice
When basin ice breaks up, it can redistribute shoreline materials, and
increase evaporation rate when open water develops (Woo, 1980; Doran,
1993). In lakes such as C2, where breakup is rare, this process is minimized.
However, even in lakes with near perennial ice cover, ice-free moats can
form, where reworking of sediment by wave action occurs (Doran, 1993). Ice
push occurs when free-floating ice pans “bulldoze” littoral zone sediments on
beaches. All direct sediment/ice interaction in this type of lake is limited to the
littoral zone.
At Stanwell-Fletcher Lake, Somerset Island, the near perennial ice cover
limits suspension of sediment by wind.
There is a distinct delineation
between fluvial sands deposited at the lake margin, and silts and clays
deposited from suspension in the deepest portions of the lake (Coakley and
Rust, 1968). Inferred varves are limited to nearshore sands, which would be
disturbed by wind-generated waves. Alternatively, at Nicolay Lake, persistent
lake ice promotes calm conditions for clay deposition in winter, and promotes
the formation of varves (Lamoureux, 1999a). Ice also insulates lake water; so
32
warmer, denser river water tends to sink, promote mixing, and oxygenates
bottom sediments (Coakley and Rust, 1968).
The Dry Valley Lakes of Antarctica have perennial ice cover.
Sediment/ice interaction is therefore limited to aeolian deposition and direct
glacial deposition (Doran et al., 1994).
Sediment accumulates in
preferentially melted depressions, and is transported through the ice cover
(Squyres, 1991).
Ice cover provides a surface for long distance transport of aeolian material
by saltation and creep (Gilbert, 1990), whereas during times of open water
aeolian material would be deposited directly in front of the sediment source.
2.8.3
Overflow
Overflows are sometimes considered relatively rare and unimportant
distributors of sediment in proglacial lakes (Smith and Ashley, 1985). They
only occur when river water is less dense than the lake water they flow into.
Since proglacial lakes usually receive high SSC during the main melt period,
overflows are limited to times of low inflow, or strong stratification. They may
be more frequent in basins receiving very fine sediment or those with low
detrital input (Smith and Ashley, 1985). They are the most effective process
for transporting suspended material directly to the outlet without deposition,
especially in lakes with short residence times (Lemmen et al., 1988). This
model may be true for the majority of proglacial lakes; however, Hector Lake
in the Rocky Mountains of Alberta is an example of a glacially influenced lake
that is dominated by overflow and interflow (Smith, 1978). The lake receives
33
low enough SSC that underflows are rare, but also receives enough sediment
for deposition of annual and sub-annual laminae.
The overflows at Hector Lake are disturbed by katabatic winds that blow
in a down-lake direction off the Wapta Icefields (Smith, 1978). Overflows are
also likely to be influenced by the Coriolis effect, deflecting sediment to the
right in the northern hemisphere (Smith and Ashley, 1985). Flow modification
by the Coriolis effect (as well as winds) can produce strong “cross-lake”
variation in SSC.
In proglacial temperate lakes, high SSC may make
overflows rare (Smith and Ashley, 1985); however, overflows may be more
common when SSC is low in Arctic lakes, and where ice cover creates weak
inverse stratification.
In 1991, 1992, and 1993, Lake C2 received discrete Q and SSC pulses,
followed by periods of inflow inactivity (Retelle and Child, 1996). In late June,
at the beginning of each melt season, both Q and SSC were low, giving
inflowing water a low density, promoting the formation of plumes of sediment
that moved towards the outlet as overflows in the epilimnion. Overflows were
recorded as reduced Secchi depth and transmissivity. SSC decreased from
proximal to distal areas, and was slightly deflected to the right by the Coriolis
effect.
Later in the season when melting and SSC increased, interflows
became the more common process of sediment distribution (Retelle and
Child, 1996).
At South Sawtooth Lake, Ellesmere Island, overflows recorded as
reduced transmissivity occurred in the early melt season as meltwater from
34
the inlet was cooler, and had a lower density than ambient lakewater. Later in
the melt season, as suspended sediment increased, interflows and
underflows were more common (Patridge, 1999).
Overflows seem to be good integrators and recorders of air temperature
(Smith, 1978).
2.8.4
Interflow
If a sediment plume is slightly less dense than water in the hypolimnion,
the flow can spread throughout the lake at an intermediate depth. This is an
effective process of sediment distribution, especially when the flow spreads
out and decelerates, depositing its coarser material as it moves (Smith and
Ashley, 1985). Deposits from interflows thin from proximal to distal locations.
Coarser sediment tends to be deposited rapidly near the stream mouth, while
silts and clays are carried in the epilimnion (overflow) or along the thermocline
(interflow; Smith and Ashley, 1985).
A plunge line is sometimes observed
when turbid water descends off a delta face to the thermocline (or below to
produce an underflow). A less turbid countercurrent produces a sharp visual
contrast at the plunge point.
Because of their lack of strong stratification, interflows in arctic lakes
might be expected to be rare; however, in meromictic lake C2, an interflow
was recorded in the 1992 melt season (Retelle and Child, 1996). This is not
surprising considering the strong density gradient near the chemocline. The
interflow was recorded in the latter part of the season, when Q and SSC were
at their highest (Retelle and Child, 1996).
35
In addition to occurring in summer when hypolimnion temperature is less
than river temperature, interflows can also occur in early autumn.
This
situation is described for Kamloops Lake when cool river water depresses the
thermocline, and deeper sediment plumes move along this density barrier
(Carmack, 1979). Interflows can also be deflected to the right by the Coriolis
effect, and can be slightly modified by wind-generated currents and seiches
(Smith et al., 1982). Depending on lake water density and SSC, interflows
can readily change to overflows and underflows (Smith and Ashley, 1985).
Temporal variation in interflow and underflow SSC can create intra-annual
rhythmites.
Usually, deposits are thin, regularly bedded, and contain no
current-generated structures. Interflows and overflows typically have lower
SSC (5-30 mg/L) than underflows (Smith and Ashley, 1985).
2.8.5
Underflow
Underflows are the dominant process for sediment dispersal and
deposition in many proglacial lakes (Smith and Ashley, 1985). An exhaustive
discussion of underflows is beyond the scope of this review; therefore, their
classification, causes, timing, movement and velocity are emphasized here.
A distinction has been made between turbidity currents created by slope
failure, and those generated directly from inflow (Pharo and Carmack, 1979).
Although the structure of deposits for the two types of underflow is similar, the
processes of formation are different for each.
River-generated underflows may be recognized in some circumstances by
increases in bottom water temperature, as warm river water plunges along
36
the delta slope to the deeper portions of the lake. However, the cause of the
density difference between underflow water and ambient water in the
hypolimnion is almost always controlled by SSC rather than dissolved load or
temperature (Smith and Ashley, 1985).
For example, underflows of Rhone River water were distinguished from
delta slumping by monitoring near-bottom water temperature at lake Geneva.
The underflows were associated with increases in temperature of up to 3oC,
but most significantly, occurred during peaks of Rhone Q (Lambert and
Giovanoli,
1988).
This
underflow/discharge
relationship
was
also
demonstrated at Peyto (Smith et al., 1982) and Lillooet lakes (Gilbert, 1975),
where mean sedimentation rates co-varied with Q.
Lambert et al. (1976) found that the occurrence of underflows and their
velocities varied widely over short time scales, but were in phase with lake
stage over longer periods. Alternatively, the shorter-term pulsating currents,
which occurred on the order of 15 minutes, were independent of stage
fluctuations, and were likely created by shooting or supercritical flow.
The strength of a turbidity currents is important, since it affects the ability
of the current to transport, deposit, and erode sediment. A wide variety of
underflow velocities and dispersal distances have been recorded. In a glacial
lake in British Colombia, extremely energetic delta-proximal turbidity currents
approached 1 m/s on an early August afternoon in 1978 (Weirich, 1986).
According to a Hjulström diagram (Boggs, 1987), if this flow velocity were
common, an erosional subaqueous channel would be expected. Particularly
37
strong turbidity currents were also detected in Lake Geneva, Switzerland, in
August 1985 as recorded 1-2 km from the delta. In 78 days of measurement,
31 down-channel events were recorded, five with velocities greater than 50
cm/s (Lambert and Giovanoli, 1988). Currents of a few to a few tens of
centimetres per second appear to be more common velocities (Smith and
Ashley, 1985). Strong currents at Lake Geneva have eroded a subaqueous
channel with levees that continue for 9 km from the delta. Though current
velocities were not measured, a delta-proximal subaqueous trench has also
been found at Lake Wakatipu, New Zealand (Pickrill and Irwin, 1982).
Underflows decelerate slowly and deposit entrained sediment when they
travel over the flat lake bottom and mix with ambient water.
At ice-contact Sunwapta Lake, only one turbidity current was recorded
during two years of study, likely because of short residence time and
uniformly high lake and river water SSC. The single turbidity current was not
associated with increased Q from the glacier, but did produce extremely high
SSC at the lake bottom (measured at 11 g/L). This turbidity current was
recorded up to 1 m above the lake bottom, reached a maximum velocity of 32
cm/s, and lasted for several hours. This contrasted with more continuous
wind-generated lake currents, which varied in velocity from 0-3 cm/s (Gilbert
and Shaw, 1981).
Turbidites can potentially confuse varve counts by producing intra-annual
laminae. For example, at Lake Walensee, Switzerland, a 1912 datum was
used to show that underflow-generated laminae were intra-annual. More than
38
twice the number of laminated couplets were found than would be expected if
the layers were varves (Lambert and Hsu, 1979).
Turbidites have been
removed from varve chronologies at Lake C2 by using their unique
sedimentological characteristics (e.g. Lamoureux, 1996), since they are
generally associated with quasi-random events such as delta slope failure.
However, because underflows are also associated with peaks in Q, their
deposits, though often chaotic, can clearly have paleoenvironmental
significance in terms of glacial discharge, snowmelt and rainfall.
Except where underflow deposits “pile up” at the end of a lake, they tend
to thin distally (Gilbert and Desloges, 1987).
Underflows would be expected to be important limnological processes in
arctic proglacial lakes with high SSC inputs, though no direct evidence of their
existence has been published. Indeed, turbidity currents would be expected
to be common in many arctic lakes, since stratification is rare.
2.8.6
Homopycnal flow
Homopycnal flow is probably relatively rare in non-ice-contact lakes. This
type of flow is likely limited to shallow, freely circulating lakes with little
variation in SSC, and where inflow sediment concentration is similar to
ambient lake SSC (Smith and Ashley, 1985). However, at glacial Tasikutaaq
Lake, Baffin Island, homopycnal flow was the dominant type of flow in 1983,
which is consistent with the lack of thermal structure and low sediment input
to the lake (Lemmen et al., 1988).
39
2.8.7
Winter flow
Winter flow from glaciers is traditionally assumed to be too low to transport
large amounts of sediment to proglacial lakes.
However, based on
contemporary discharge measurements and laminated sedimentary records,
this model has been challenged (Shaw et al., 1978). It is now thought that
turbidity currents can occur in winter, given the right conditions. Cyclonic
winter storms associated with high temperatures and winter rain is one
mechanism responsible for glaciohydraulic activity in winter. Other possible
processes of turbidity current formation in winter are the failure of delta
foresets when lake level is low, or when lake ice abrades an oversteepened
delta face, creating slumps. Also, when upvalley ice-dammed lakes or ponds
exist, winter dam failures can create brief, but large Q and SSC. Failure of
the lake outlet could also produce rapid changes in lake hydrology (Shaw et
al., 1978).
Turbidites within winter clay deposits are common in varved
sequences from temperate glacially fed lakes. The importance of winter flow
is unknown in the cooler, drier High Arctic, but above freezing temperatures
commonly occur in the eastern Arctic (Gilbert and McKenna Neuman, 1988).
2.9
Conclusion
It is evident that simple generalizations are not universally applicable to
a type of lake, or its watershed. For example, a glacially fed arctic lake can
receive a dominantly nival meltwater signal one year, and have a glacial
signature the next.
Clockwise hysteresis is a useful concept for
understanding seasonal and diurnal changes in SSC, but is disrupted by
40
quasi-stochastic events such as jőkulhlaups. Overflows may be rare in arctic
lakes, but were found to be important distributors of sediment in Lake C2, the
only published high-resolution lacustrine process study in the High Arctic.
In light of the complexities described above, lake processes must be
thoroughly studied to understand how sediment is deposited. However, each
lake and watershed will respond differently to identical forcings.
The
distinctive character of a lake is expressed through a unique mix of external
controls such as watershed geology, vegetation, glacial cover, temperature,
precipitation and wind. Internal controls such as salinity, temperature, ice
cover and mixis are all extremely important. The lack of knowledge of the
complex interplay of these variables is particularly acute in the High Arctic,
where little process-oriented research has taken place.
Monitoring a
watershed over as long a period as possible is one of the best techniques
available for inferring environmental conditions of the past.