Odekon M. Encyclopedia of paleoclimatology and ancient environments

Подождите немного. Документ загружается.

Barbados, the Sunda Shelf, and the Bonaparte Gulf are represen-

tative of sites far from former ice margins, where sea levels

experienced a sustained rise until about 7,000 years ago. The

onset of this rise appears to occur shortly after 20,000 years

ago and its rate appears to have been variable. The record from

the Bonaparte Gulf indicates that the low sea levels persisted for

an interval of about 10,000 years, starting at ~30,000 years ago.

Detailed observations for the past 7,000 years from such local-

ities distant from the former ice sheets, e.g., Orpheus Island,

Australia, also indicate that sea levels at continental margins

have actually been falling slowly as a result of the oceanic man-

tle response to the changing water load.

Figure G39 Observations of relative sea-level change from different locations and for different time intervals. All observation sites are

believed to have been tectonically stable except for Barbados where a small correction for such motion has been applied. Note the different

scales used for the various time-height plots. All observed depths and elevations have been reduced to mean sea level and all timescales

correspond to calendar years (adapted from Lambeck and Chappell, 2002).

GLACIAL ISOSTASY 377

This spatial variability can be understood in terms of the

several components contributing to the signal and, in particular,

from the interplay between local rebound and global ocean

volume change. For sites far from the ice sheets, the isostatic

contributions are relatively small, about 10–15% of the eustatic

change, and insensitive to the details of the melting history.

Observations from “far-field” sites such as Barbados or the

Bonaparte Gulf therefore constrain the change in total ice

volume once the small isostatic corrections are applied. For

the sites near the centers of former glaciation, the observed sig-

nal is predominantly from the crustal rebound that dominates

the ocean volume signal. If the eustatic change is known from

the analysis of far-field data, the rebound term can be deter-

mined and this is primarily a function of rheology parameters

and ice thickness. When melting ceases abruptly, a small cusp

may occur in the sea-level data and this becomes more pro-

nounced when the rebound contribution is reduced relative to

the eustatic change. This occurs, for example, at Andøya and

establishes the time of the end of major glaciation. At these

sites the signal is initially dominated by the rebound but at a

certain time this is reduced and equal in magnitude but oppo-

site in sign to the global eustatic term so that sea level is con-

stant for a period. Later the eustatic term dominates and sea

level again rises until melting ceases and it is the residual

relaxation of the unloading that dominates the signal. This lat-

ter part of the signal provides a useful constraint on mantle

rheology while the earlier part of the signal provides constraints

on the ice distribution at its margins. The Andøya signal, for

example, indicates that the outer Vesterålen of Norway were

ice free at about 19,000 years ago, but that the ice must have

extended offshore before that time in order to produce the ele-

vated sea levels once the area became ice free.

Beyond the ice margin, the rebound signal during deglacia-

tion becomes one of subsidence. During the loading phase,

mantle material is displaced beneath the area of loading and

because mass is conserved, a broad but shallow bulge forms

around the periphery of the ice sheet (c.f. Figure G38). Then,

as the ice melts, this bulge subsides and results, in the absence

of other contributions, in an apparent rise in sea level. Thus,

here the sea-level change is initially rapid as the ice decays

and, when melting has ceased, the rise is slow and reflects

the isostatic change. The early part of the record represents

mainly the eustatic contribution and the latter part reflects

mainly the rheology. Much further from the ice sheets, the

deformation resulting from the changing ice load is reduced

and the major isostatic term comes from the change in water

load (Figure G38). This effect is most pronounced at continen-

tal margins where the extra water during deglaciation depresses

the ocean floor, pulling sea level down with it, and resulting in

a fall in sea level at the coast. This signal becomes most visible

when melting has ceased and results in small-amplitude high-

stands at about 6,000–7,000 years ago, as observed in many

places along the Australian coast (the Orpheus Island result).

What these examples illustrate is that the isostatic signals,

by taking advantage of their spatial and temporal variability,

provide a means of estimating both the mantle rheology and

aspects of the ice history. This is important because generally

the distribution of the ice during the recent glaciation is not

well known. Most of the observational evidence is from the last

20,000 years, older information being scarce because it has

been largely destroyed by an earlier ice advance or phase of

sea-level rise and also because the commonly used radiocarbon

dating method becomes unreliable or imprecise.

The limits of the ice sheets over North America and Europe

for the LGM and for the subsequent ice retreat have been

mostly well mapped but major differences of opinion remain

for some locations, most notably for Arctic Russia. Sea-level

observations may help here. For example, if a large ice sheet

existed over the Kara Sea and western Siberia during the

LGM, then a raised shoreline pattern similar to that observed

for adjacent Fennoscandinavia resulting from glacial rebound

would be expected, but the absence of such a pattern indicates

that any LGM ice over the region must have been unimportant.

For regions such as this, the glacial isostasy analysis provides

more insight into the ice history than into the Earth rheology.

Debate also occurs over the extent to which the large ice sheets

extended onto the continental shelves. Did the Scandinavian

Ice Sheet extend across the North Sea and overrun the northern

British Isles or did the two ice sheets form separately? Again,

the glacial isostatic analyses for the two areas provide useful

constraints. If thick ice did extend across the North Sea during

the LGM then the rebound in southwestern Norway would

have been greater than that actually observed and the center

of rebound over the British Isles would lie in eastern Scotland

rather than where it is observed in western Scotland. Where

were the LGM North American ice margins in northeastern

Canada? Was the ice restricted to the upper fiords of Baffin

Island or did it extend far onto the continental shelf? The pre-

sence of well-elevated shorelines at outer coastal sites, with

early Holocene ages, points to substantial rebound having

occurred here and to early ice over the region and onto the

shelf. Another area of uncertainty concerns the ice thickness

at the LGM and during the subsequent deglaciation phase.

Only in special cases where mountains protruded through the

ice are there observational constraints on the thickness and this

is not the case for the large former ice sheets other than near

their margins. Instead, the ice thickness is based on glaciologi-

cal and climate assumptions, for example about the conditions

at the base of the ice sheet, the coupling of the ice to the under-

lying rock and sediments, and the supply of moisture. Alterna-

tive, but yet plausible, assumptions can lead to distinctly

different results for the ice thickness, and the glacial isostatic

inversions must include a strategy for estimating ice thickness.

Because of the good distribution in time and space of the obser-

vational evidence across northern Europe, the Fennoscandian

rebound problem is particularly well posed and inversion

schemes can be developed that solve the problem for both

rheology and ice sheet parameters.

Figure G40 illustrates the isostatic rebound of Scandinavia

based on model calculations constrained by a variety of geolo-

gical, geodetic, and geomorphological data. The results are

shown for three epochs: (a) an early stage of the LGM when

ice margins were at their maximum extent, (b) about 12,000

years ago when the southern ice margins had retreated across

the Baltic Sea and when the retreat was momentarily halted

during the climatic cold interval of the Younger Dryas, and

had vanished. At the time of the second epoch, the Baltic was

closed to the Atlantic both by the ice barrier in southern

Sweden and by the land barrier through the Danish Bælts. This

formed the Baltic Ice Lake at an elevation of about 25–30 m

higher than coeval sea level. It left well-defined shorelines

across the region, which today are strongly inclined relative

to present sea level, and this information has been used to con-

strain the model parameters. Tide-gauge observations of sea-

level change for the past century as well as Global Positioning

378 GLACIAL ISOSTASY

Figure G40 Paleogeographic reconstructions for Fennoscandia at three epochs: ~20,000 years ago corresponding to the end of the Last

Glacial Maximum (LGM), ~12,000 years ago, corresponding to the Younger Dryas (YD), and for the present. The ice sheets are shown by the

translucent regions with the ice thickness contours, in white, corresponding to 400 m intervals. Areas with underlying blue correspond to

regions below sea level for the epoch given. The paleoshorelines are defined by the green-blue boundary for each epoch. During the Younger

Dryas, the Baltic Ice Lake was about 25–30 m higher than the Atlantic Ocean and formed a frozen freshwater lake. For the LGM and YD the

yellow and red contours correspond to the change in local sea level that occurred since the present. The yellow contours are positive and in

this area the paleoshorelines, if preserved, will occur above present sea level. The zero value is identified as the first yellow contour. The red

contours are negative. The contour values are 100 m for the LGM, and 50 m for the YD. The third panel illustrates the present-day rate of

change in sea level rise (contour interval is 1 cm yr

1

) with the maximum rate in the northern Gulf of Bothnia approaching 8 cm yr

1

GLACIAL ISOSTASY 379

System (GPS) observations of crustal rebound for the past dec-

ade provide additional constraints on the rebound models.

The essential rheological result from this and inversions of

data sets from other localities is that the viscosity can be ade-

quately described by a linear relation between stress and rate of

strain over the range of stress magnitudes and duration of stress

cycles encountered. The resulting “effective” viscosity is depth-

dependent, with the lower mantle below about 670 km depth being

of distinctly higher viscosity than the upper mantle. Some viscos-

ity stratification occurs within the upper mantle, with the

depth dependence broadly following the depth of seismic para-

meters of shear velocity and attenuation. Lateral variation in the

upper mantle also occurs, again broadly following the seismic

parameters. Beneath the continental margins of Australia, for

example, the upper mantle viscosity is less than beneath Scandina-

via, where in the former case the estimates of viscosity come

from analysis of the water loading effects. Such variation is not

unexpected since both mantle viscosity on time scales of thou-

sands of years and the propagation of seismic waves at periods

of seconds and minutes are likely to be temperature controlled.

Mantle viscosity is a fundamental parameter for describing

the Earth’s dynamical behavior because it controls the rates of

convection and thermal evolution of the mantle. The challenges

for future work in glacial isostasy are to refine this viscosity

structure, not only its depth and lateral variability but also its

behavior over time. Is the evidence for linear behavior supported

when some of the longer cycles in glacial loading are examined?

This will require new information for the earlier periods of the

glacial cycles of both the ice sheets and the Earth’s response.

Improved observational evidence for the present rates of post-

glacial deformation will also contribute to an improved under-

standing of the viscosity structure. These measurements will

include high-precision measurements of the displacements of

points on the Earth’s surface and changes in gravity. The chal-

lenge here will be to separate these changes from tectonic signals.

Kurt Lambeck

Bibliography

Lambeck, K., 2002. Sea-level change from mid-Holocene to recent time:

An Australian example with global implications. In Mitrovica, J.X.,

and Vermeersen, B. (eds.), lce Sheets, Sea Level and the Dynamic

Earth. Washington, DC: American Geophysical Union, pp. 33–50.

Lambeck, K., 2004. Sea-level change through the last glacial cycle. Compt.

Rendus Geosci., 336, 677–689.

Lambeck, K., and Chappell, J., 2001. Sea-level change through the last

glacial cycle. Science, 292, 679–686.

Watts, A.B., 2001. Isostasy and Flexure of the Lithosphere. Cambridge:

Cambridge University Press.

Cross-references

Glacial Eustasy

Sea-level Change, Post-Glacial

Sea-Level Change, Quaternary

Sea-Level Indicators

GLACIAL MEGALAKES

Continental ice sheets that form during epochs of glaciation

exert immense influences on water drainage across the land.

Their huge loads depress the underlying land surface, and lakes

form in the moats that surround the ice sheets. They can block

the lower courses of major rivers, impounding flow, and even

diverting it into adjacent drainage basins. Meltwater from gla-

cial margins may introduce huge discharges into land-surface

depressions that hold much smaller lakes during nonglacial per-

iods. The lakes may climatically alter water balances, promot-

ing further glaciation by a kind of positive feedback.

Glacial megalakes are exceptional for their extent and/or

their volume. They may form in direct contact with glacial ice,

or they may form in nearby depressions under the pronounced

influence of glaciers in the regional drainage basin. There are

modern examples for all these situations, but the relatively small

size and different thermal regimes of modern versus ice-age gla-

ciers pose problems of extrapolation to the past conditions

of major continental glaciation that favored the development of

glacial megalakes.

Types of glacial megalakes

Ice-marginal lowland megalakes

The classic example of an ice-marginal lowland megalake is

the Baltic Ice Lake, which formed about 14,500 calendar years

ago, in the ice-marginal moat to south and east of the retreating

Fennoscandian Ice Sheet. Prominent, well-dated glacial mor-

aines mark the recession of this ice sheet from northern

Germany and Poland to positions in Denmark and southern

Norway and Sweden, which blocked the outlet of glacial melt-

water from the Baltic to the North Sea until about 11,200 years

ago. The lake then drained through a spillover across the

Mt. Billingen Area, southern Sweden (Bjorck, 1995). There

followed a period of alternation between marine and saline

conditions, as land rebounded from the load of the disappearing

ice sheet and global sea level rose from the melting of ice on

all the continents. At its maximum extent, the Baltic Ice Lake

covered nearly 400,000 km

in a relatively shallow inundation

with a volume of about 9,000 km

. Its modern successor, the

Baltic Sea, is the world’s largest brackish water body.

Structural/tectonic basin megalakes

Basins formed by tectonic processes may hold lakes that are

strongly influenced by glaciers in their drainage basins. Classic

examples occur in the Basin and Range tectonic province of the

western United States, where Lake Bonneville, the mega-lake

predecessor of the Great Salt Lake, formed during the last ice

age. This paleolake achieved a maximum area of 52,000 km

and a volume of 7,500 km

between about 22,000 and 17,500

calendar years ago. The lake filled to the level of a spill point

in south-central Idaho, and then dropped 100 m as it eroded into

the outlet, releasing the catastrophic Bonneville Flood down the

Snake River (O’Connor, 1993).

Another megalake formed at the elevation of 3,800 m in

the Altiplano of southern Bolivia and northern Peru. The modern

Altiplano depression contains Lake Titicaca (8,800 km

in area),

Lake Poopo, and the great salt pans of Ciopasa and Uyuni. Paleo-

lake Tauca covered the whole depression (about 50,000 km

)toa

depth of around 100 m. It achieved its maximum stage after the

Last Glacial Maximum in the nearby mountains (about 22,000–

20,000 calendar years ago).

Intermontane, ice-dammed megalakes

Very deep lakes can form when a glacier advances down

a mountain valley to block a river . Perhaps the most famous

example is Glacial Lake Missoula, which formed between about

380 GLACIAL MEGALAKES

17,500 and 14,500 calendar years ago (Atwater, 1986). At max-

imum extent this lake covered 7,500 km

, and it held a volume

of about 2,500 km

. Cataclysmic failure of the ice dam impound-

ing this lake resulted in the release of water over 600 m deep,

thereby generating some the largest known glacial megafloods

(Baker and Bunker, 1985).

Subglacial megalakes

Ice-penetrating radar mapping of the sub-ice floors of portions

of the Antarctic Ice Sheet has identified over 70 subglacial

lakes. Largest of these is Lake Vostock, which occupies an

interior subglacial trench of tectonic origin. The lake measures

240 by 50 km and holds between 2,000 and 5,000 km

water (subject to interpretive uncertainty as to the lake depth).

Because the water in this lake is isolated from the surface

environment by 4 km of overlying ice, it likely contains very

interesting biota, surviving in an extreme environment similar

to those inferred to be present on Mars and on Jupiter’s icy

moon, Europa. A project is currently underway to sample the

water in this remarkable lake.

Glacial Lake Agassiz and the Laurentide Ice Sheet

Perhaps the largest well-documented glacial megalake formed

in north-central North America in association with the largest

glacier of the last ice age, the Laurentide Ice Sheet. This ice

sheet is known to have been highly unstable throughout much

of its history. Not only did freshwater discharges from the gla-

cier result in ice-marginal lakes; outbursts of meltwater into

the Atlantic Ocean may have generated climate changes by

influencing the thermohaline circulation of the North Atlantic

Ocean (Teller et al., 2002).

During the Last Glacial Maximum, the Laurentide Ice Sheet

achieved its maximum extent about 20,000 years ago. The

initial glacial retreat generated relatively small lakes because

of the regional river drainage to the south, away from the ice

front. The ice sheet surged and readvanced at times, leading

to a very complex history in the basins that now contain the

Great Lakes. However, when the ice sheet advanced north of

this area into the Canadian Shield, Lake Agassiz formed in

the trough between the ice and drainage divides to the south

and west, in parts of present-day North Dakota, Manitoba, and

western Ontario. The megalake seems to have been initiated

close to the time of the Younger Dryas cooling event, which

began about 12,700 years ago. One explanation for this cooling

is diversion to the St. Lawrence River of the Laurentide melt-

water that previously flowed to the Gulf of Mexico via the

Mississippi. The outflow of great quantities of freshwater into

the North Atlantic disrupted the salinity gradient that drives

the thermohaline circulation of the oceans by the formation of

deepwater in the North Atlantic (Broecker and Denton, 1989).

As the southern margin of the Laurentide Ice Sheet retreated

northward into the present Hudson Bay region, Lake Agassiz

adjusted to a succession of overflow or spill points, approxi-

mately as follows: (a) It drained southward via the Minnesota

River to the Mississippi and Gulf of Mexico (about 12–13,000

years ago). (b) It then entered Lake Superior via a route near

Thunder Bay. (c) It shifted to the northwest via the Clearwater

spillway to the Mackenzie River and the Arctic Ocean (about

11,000 years ago). (d) It then flowed eastward again to Lake

Superior through a succession of spillways (about 10,000 years

ago). (e) It shifted to the Ottawa River and the St. Lawrence.

(f) Finally, it flowed to the Hudson Bay area (about 8,400 years

ago). This 5,000-year history resulted in inundation at one

time or other of 1.5 million km

,butitsmaximumone-time

extent was not achieved until a union occurred with glacial Lake

Ojibway in northern Ontario about 8,400 years ago. At this max-

imum, terminal stage, the resulting megalake covered about

840,000 km

, and held about 160,000 km

of water (Leverington

et al., 2002). This is double the volume of the largest modern

lake, the Caspian Sea, but the paleolake was highly unstable.

It was dammed by remnants of the Laurentide Ice Sheet, which

was severely weakened by of the influx of marine water

into Hudson Bay. The stage was set for a massive sub-glacial

outburst of the Lake Agassiz water through Hudson Strait

into the Labrador Sea, a probable trigger for the 8,200 yr B.P.

cold event – the most abrupt and widespread global climatic

cooling event to have occurred within the last 10,000 years

(Clarke et al., 2003).

Central Asian paleolakes

In the mountain areas of central northern Asia, there were prob-

able ice-dammed expansions of modern Lakes Baikal (31,500

area; 23,000 km

volume) and Issyk-Kul (6,000 km

area;

4,000 km

volume). In the Altai Mountains, great megafloods

emanated from the late Pleistocene ice-dammed Chuya-Kuray

paleolake, covering 12,000 km

and holding 1,000 km

water, perhaps 900 m deep at the ice dam (Baker et al., 1993).

The lowland glacial lakes of Siberia involved the expansion

of huge ice sheets that covered the shallow seas north of

Eurasia and blocked north-flowing rivers, notably the Ob,

Irtysh and the Yenesei. Grosswald (1980) interpreted this

blockage to be Late Weichselian in age (about 15–20,000 years

ago), but more recent work considers the events to be Early

Weichselian (about 90,000 years ago) (Arkhipov et al., 1995).

The largest lake, formed on the west Siberian plain, was esti-

mated by Mangerud et al. (2001) to cover 600,000 km

at a sur-

face elevation of 60 m. However, both Arkhipov et al. (1995)

and Grosswald (1980) postulate a much larger paleolake, about

1,200,000 km

in area, with a volume of about 75,000 km

a surface elevation of 128 m. This west Siberian megalake

was the Asian equivalent of Agassiz. It drained southward,

through the Turgai divide of north-central Kazakhstan, to the

basin of the modern Aral Sea. The latter rose from its 1960 ele-

vation of 53 m to 70 or 80 m, enlarging from 60,000 km

(1960

area) to about 100,000 km

in the Pleistocene. This paleolake

then drained southwestward through the Uzboi Channel into

the basin of the modern Caspian Sea. Also fed by glacial melt-

water from northern Europe via the Volga, the Caspian

expanded during its late glacial Khvalyn phase to an area of

950,000 km

, holding a volume of 135,000 km

at an elevation

of 50 m (the modern Caspian level is –28 m). The Khvalyn

paleolake drained via the Manych paleochannel and the Don

River to the basin of the modern Black Sea. The Black Sea

was then disconnected from the Agean and filled by freshwater

(the New Euxine phase); the glacial meltwater filled the basin

to an elevation of –60 m. This basin functioned in two modes

during the Quaternary. Its cold-climate mode was a freshwater

lake (that may have filled and drained through the northwestern

Turkey to the Agean). Its warm-climate phase involved rising

global sea level, inducing salt-water invasions through the

Turkish straits to form the saline Black Sea, the last of which

occurred about 8,000 years ago, an inundation that may have

inspired the story of Noah (Ryan et al., 2003).

Victor R. Baker

GLACIAL MEGALAKES 381

Bibliography

Arkhipov, S.A., Ehlers, J., Johnson, R.G. and Wright, H.E. Jr., 1995. Gla-

cial drainage towards the Mediterranean during the Middle and Late

Pleistocene. Boreas, 24, 196–206.

Atwater, B.F., 1986. Pleistocene Glacial-lake Deposits of the Sanpoil River

Valley Northeastern Washington. U.S. Geological Survey Bulletin

1661, 39pp.

Baker, V.R., and Bunker, R.C., 1985. Cataclysmic late Pleistocene flooding

from glacial Lake Missoula: A review. Quaternary Sci. Rev., 4,1–41.

Baker, V.R., Benito, G., and Rudoy, A.N., 1993. Paleohydrology of late

Pleistocene superflooding, Altai Mountains, Siberia. Science, 259,

348–350.

Bjorck, S., 1995. A review of the history of the Baltic Sea, 13.0–8.0 ka BP.

Quaternary Int., 27,19–40.

Broecker, W.S. and Denton, G.H., 1989. The role of ocean-atmosphere

reorganizations in glacial cycles. Geochim. Cosmochim. Acta, 53,

2465–2501.

Clarke, G., Leverington, D., Teller, J., and Dyke, A., 2003. Superlakes,

megafloods, and abrupt climate change. Science, 301, 922–923.

Grosswald, M.G., 1980. Late Weichselian Ice Sheet of northern Eurasia.

Quaternary Res., 13,1–32.

Leverington, D., Mann, J.D., and Teller, J.T., 2002. Changes in the bathy-

metry and volume of glacial Lake Agassiz between 9,200 and

7,700

C yr B.P. Quaternary Res., 57, 244–252.

Mangerud, J., Astakhov, V., Jakobsson, M., and Svendsen, J.I., 2001.

J. Quaternary Res., 16, 773–777.

O’Connor, J.E., 1993. Hydrology, hydraulics and sediment transport of

Pleistocene lake Bonneville flooding on the Snake River, Idaho. Geolo-

gical Society of America Special Paper 274, 83pp.

Ryan, W.B.F., Major, C.O., Lericolais, G., and Goldstein, S.L., 2003. Cat-

astrophic flooding of the Black Sea. Annu. Rev. Earth Planet. Sci., 31,

525–554.

Teller, J.T., Leverington, D.W., and Mann, J.D., 2002. Freshwater outbursts

to the oceans from glacial Lake Agassiz and their role in climate change

during the last deglaciation. Quaternary Sci. Rev., 21, 879–887.

Cross-references

Cordilleran Ice Sheet

Last Glacial Termination

Late Quaternary Megafloods

Laurentide Ice Sheet

Proglacial Lacustrine Sediments

Quaternary Climate Transitions and Cycles

Scandinavian Ice Sheet

The 8,200-year BP Event

Thermohaline Circulation

Younger Dryas

GLACIAL SEDIMENTS

Glacial sediments are formed in association with glacier ice in

subglacial, ice marginal, lacustrine and marine environments.

In such diverse environments, sediment can be reworked and

deposited by a very wide range of processes, including subglacial

lodgement, deformation and melt-out, subaerial and subaqueous

mass-movements, fluvial processes, and settling through a water

column. Numerous attempts have been made to classify the

resulting range of sediments in schemes of greater or lesser com-

plexity (Dreimanis, 1989), incorporating many varieties of “till”

(e.g., flow till, lowered till, waterlain till, iceberg till). Many of

these proposed sediment types are difficult to distinguish in the

field, however, and a simple classification scheme is now favored

by most workers. A fundamental distinction can be made between

primary glaciogenic deposits, formed by exclusively glacial pro-

cesses (such as subglacial lodgement, melt-out and deformation),

and secondary deposits, formed by the reworking of glaciogenic

debris by processes not unique to glacial environments (Lawson,

1989). In terms of their internal characteristics, secondary glacio-

genic deposits can be indistinguishable from sediments formed

by similar processes in non-glacial environments and so, seen in

isolation, they do not provide unequivocal evidence for the former

presence of glacier ice. A glacial origin, however, can be deter-

mined from the context of such sediment facies, particularly the

range of associated facies and structures, and the form and distri-

bution of overlying landforms. It is therefore helpful to take a hier-

archical approach to the analysis and interpretation of glacial

sediments, in which individual facies are interpreted in terms of

their process of deposition, and sediment-landform associations

(comprising suites of genetically-related facies) are used to recon-

struct former depositional environments (cf. Walker, 1992; Benn

and Evans, 1998).

The characteristics of glacial sediments reflect the processes

of entrainment, transport, and deposition experienced by debris

as it travels through a glaciated basin. One of the most distinc-

tive characteristics of glacial sediments is the presence of erra-

tics, or exotic, far-traveled material. For Quaternary deposits,

the provenance of erratics can often be pinpointed, allowing

patterns of glacier flow to be reconstructed with some accuracy.

For older successions, however, this may not be possible, espe-

cially if sediment exposure or preservation is poor. Erratics do

not occur in all glacial deposits, however, and some are dominated

by local material,eroded and redeposited close to its source. Trans-

port processes can also impart a distinctive glacial signature

on sediments. Debris transported at glacier beds is commonly

subjected to high inter-particle contact forces, and the resulting

abrasion and crushing of particles produces distinctive clast

morphologies and particle size distributions. Clasts are commonly

striated (particularlyfiner lithologies)with smooth, polished facets

on their upper and lower surfaces, and may have asymmetrical

stoss-lee or bullet-shaped forms. The presence of such clasts in a

deposit does not necessarily mean, however, that it was deposited

in a subglacial environment, because glacial debris can be carried

far beyond glacier margins by various processes prior to final

deposition. Some of these processes (such as fluvial transport) will

result in rapid clast rounding and obliteration of striae and other

glacial signatures, whereas others (such as iceberg rafting) may

carry debris for hundreds of kilometers with little or no modifica-

tion. Particle size distributions of subglacially-transported debris

commonly exhibit a very broad range of grain sizes, from clay or

siltup to pebbles or larger, and are typically bimodal or polymodal.

The finer particles (<63 mm) are formed during subglacial abra-

sion, and are sometimes known as rock flour. Other particle size

modes may reflect fracturing or crushing of larger clasts along

joints, grain boundaries and other weaknesses. Whereas subglacial

transport produces distinct debris characteristics, debris trans-

ported in englacial and supraglacial positions generally undergoes

little or no modification and commonly retains the character of the

parent debris. For this reason, supraglacial and englacial debris

transport is known as passive transport, in distinction to active

subglacial transport (Boulton, 1978; Benn and Evans, 1998),

although some limited clast modification can occur during passive

transport. Passive transport is particularly important in mountain

environments, where large quantities of debris are delivered to gla-

cier surfaces by rockfall and other mass movement processes. Flu-

vial processes can also transport significant quantities of sediment

through some glaciers, resulting in significant clast rounding and

size sorting. Thus, although striated, faceted clasts and rock flour

are typical of subglacially transported debris, they are not essential

382 GLACIAL SEDIMENTS

characteristics of glacial sediments, nor do they provide unequivo-

cal evidence of ice-contact deposition.

Debris can be released from glacial transport at the glacier

bed, or at subaerial or subaqueous glacier margins. In the for-

mer case, tills can be formed by the frictional lodgement of par-

ticles against the bed, the deformation of weak subglacial

materials, or the melt-out of debris-rich ice. Subglacial defor-

mation is now thought to be a very important till-forming pro-

cess, especially where ice-sheets override soft sedimentary

rocks or unlithified sediments. High cumulative strains may

result in the complete mixing of the parent materials, creating

diamicts which may be hard to distinguish from other till types.

Paradoxically, subglacially deformed sediments are often easier

to identify where cumulative strains are lower, because structures

and textures inherited from the parent material are more likely to

survive, and patterns of deformation easier to recognize. Such

glaciotectonites are typically heterogeneous, incorporating lami-

nae, boudins or rafts of pre-existing materials, often complexly

folded and faulted. Glaciotectonites and deformation tills often

occur in close association. Lodgement is an important process

beneath wet-based ice; large boulders are especially likely to

lodge against the bed, and are then molded by overpassing ice

into asymmetrical stoss-lee forms resembling roches mouton-

nées. Although the term lodgement till is widely used (especially

in the older literature) it is unlikely that lodgement, acting alone,

can result in thick till units, and it is likely that most subglacial

tills have undergone some degree of deformation. Melt-out tills

are susceptible to reworking by slope processes or during dewa-

tering, and thick, undeformed melt-out till sequences generally

have low preservation potential. Sediments deposited below

active ice are commonly streamlined. Long, narrow ridges of till

(fluted moraines, or flutes) may extend downglacier from

boulders or other obstructions on the bed, reflecting till deforma-

tion into cavities at the glacier sole. Drumlins are larger stream-

lined bedforms, commonly with complex internal stratigraphies.

They may have rock or sediment cores (which may be deformed

or undeformed), with an outer carapace of till or glacitectonite.

The most widely accepted origin for drumlins involves selective

deformation, erosion and deposition of a soft glacier bed,

although other ideas have been proposed.

Debris released by melting at a glacier surface or along ter-

restrial ice margins is commonly reworked by a wide variety of

processes, including debris flow, other mass movements such

as falls, slides and slumps, and the action of flowing water

and wind. All of these processes will modify sediment charac-

teristics to some degree. In particular, grain-size distributions

are strongly influenced by fluvial action, fall sorting, winnow-

ing by wind, and other processes, and many secondary glacial

deposits exhibit a narrower range of grain sizes than subglacial

tills. The simplest ice-marginal deposits consist of aprons of

diamicts (debris flow and other mass movement deposits),

and gravels and sands (glaciofluvial deposits) built up around

the ice margin. Such dump moraines have a fan-like internal

geometry, with beds dipping out from the ice margin at angles

up to 35



. Dump moraines are particularly common around

debris-covered glaciers in high mountain environments, and may

form steep-sided ramps >100 m high. In many ice-marginal envir-

onments, however, sediments may be deposited and reworked

several times as the local topography is altered by ice melting

and oscillations of the glacier margin. Consequently, many ice-

marginal sediment sequences have complex stratigraphies consist-

ing of complexly interbedded diamicts, gravels, sands and silts,

affected by faults and folds. Advances of the ice margin can

bulldoze ice-marginal and proglacial sediments into push mor-

aines, or larger-scale glaciotectonic thrust-block moraines, with

complex internal deformation structures (Benn and Evans, 1998).

Many glaciers terminate in lakes or the sea, and a large part

of the glacial geologic record consists of glaciolacustrine and

glaciomarine sediments. The character of subaqueous glacial

sediments depends on the activity of the ice margin, sediment

supply, meltwater discharges, and whether the ice is grounded

or floating. At temperate glacier margins, where meltwater dis-

charges are high, sedimentation may be dominated by subaqu-

eous outwash, forming fans of crudely-sorted gravels and sands

interbedded with fine-grained drapes deposited from meltwater

plumes. If glacier margins are sufficiently stable, subaqueous

fans can accumulate to the waterline, whereupon they evolve

into ice-contact deltas. These have a three-fold internal strati-

graphy, consisting of (a) low-angle topsets of gravel or sand,

formed at the delta top; (b) dipping foresets formed on the delta

foreslope; and (c) low gradient bottomsets, formed by the set-

tling of fine sediments from turbidity currents beyond the delta

front. Progradation of a delta can result in these sediments forming

a vertical succession. Some subaqueous ice margins are dominated

by mass movements, including debris flows, slumps and slides.

Such mass movements tend to become more dilute and finer-

grained down-flow, forming turbidity currents. These are particu-

larly common in glaciolacustrine environments, where they form

distinctly laminated successions consisting of basal graded sands,

rippled sands and silts, laminated silts, and an upper drape of fine-

grained sediment. Glaciolacustrine turbidites should not be

confused with varves, which are annually deposited couplets

consisting of a clay layer deposited during the winter (when the

lake surface is frozen and sediment input is small) and a coarser

layer deposited in summer. In contrast, turbidites are deposited

rapidly (over a few minutes to hours in glacial settings), and lack

a winter clay layer. Where glacier ice is in direct contact with lake-

or seawater, calved icebergs can transport debris far beyond ice

margins, then release it as they progressively melt or overturn.

Dropstones can occur as isolated particles within fine-grained

sediments, or in iceberg dump mounds. Sediments with high

concentrations of dropstones are termed dropstone muds or drop-

stone diamicts.

During the last three decades, glacial geologists have amassed a

large amount of data on modern glacial sedimentary environments

in different climatic and topographic settings. This information has

been used to define a range of glacial land system models, which

can guide the interpretation of the glacial geologic record of the

Pleistocene and older glacial epochs (Evans, 2003). Caution must

be employed when appealing to modern analogs in glacial geol-

ogy, because some former events or environments may not have

modern equivalents, either in kind or in magnitude. With due care,

however, modern analogs provide a powerful means of extracting

maximum information about former depositional processes, envir-

onments,ice-margin activity, and climatic settingfrom Earth’srich

glacial geologic record.

Douglas I. Benn

Bibliography

Benn, D.I., and Evans, D.J.A., 1998. Glaciers and Glaciation. London:

Arnold.

Boulton, G.S., 1978. Boulder shapes and grain-size distributions of debris

as indicators of transport paths through a glacier and till genesis. Sedi-

mentology, 25, 773–799.

GLACIAL SEDIMENTS 383

Dreimanis, A., 1989. Tills: Their genetic terminology and classification. In

Goldthwait, R.P., and Matsch, C.L. (ed.), Genetic Classification of

Glacigenic Deposits. Rotterdam: Balkema, pp. 17–84.

Evans, D.J.A. (ed.), 2003. Glacial Landsystems. London: Arnold.

Lawson, D.E., 1989. Glacigenic resedimentation: Classification concepts and

application to mass-movement processes and deposits. In Goldthwait, R.P.,

and Matsch, C.L. (eds.), Genetic Classification of Glacigenic Deposits.

Rotterdam: Balkema, pp. 147–169.

Walker, R.G., 1992. Facies, facies models and modern stratigraphic con-

cepts. In Walker, R.G., and James, N.P. (eds.), Facies Models: Response

to Sea-level Change, 1–14. Toronto: Geological Association of Canada.

Cross-references

Drumlins

Glacial Erratics

Glaciofluvial Sediments

Glaciomarine Sediments

Ice-Rafted Debris (IRD)

Moraines

Roche Moutonnées

Tills and Tillites

GLACIATIONS, PRE-QUATERNARY

Introduction

Earth has experienced recurring periods of glaciation (icehouse

climatic conditions) alternating with periods of warm climate

(greenhouse climatic conditions) (Figure G41; Hambrey and

Harland, 1981; Eyles, 1993; Crowell, 1999). Glaciations have

occurred on one continent or another, except for the Mesopro-

terozoic (ca. 2–1 Ga) in Precambrian times and the Mesozoic

(250–65 Ma) in Phanerozoic times (Figure G42).

Changes in insolation, agglomeration, splitting and drifting

of continents and associated orogenies are the first-order con-

trols for changing climatic conditions on Earth. Some glacial

periods correlate reasonably well with major orogenic phases

(Figure G41 ); that is, with the agglomeration of continents

and the change in distribution and volume of oceanic basins.

The relative position and elevations of continents affected

oceanic and atmospheric circulation, along with redistribution

of heat, as well as degree of weathering of exposed rock and

associated sequestration of atmospheric CO

. These factors all

contributed to the development of greenhouse and icehouse

conditions and glaciations. Other factors such as orbital para-

meters further modulated the behavior of glaciers during ice-

house periods.

Recognition of ancient glaciations is based on type and distri-

bution of sedimentary deposits, fossil assemblages and their iso-

topic composition, and, in some cases, on erosional features.

However, the older the glaciation, the less reliable and abundant

the information is. Ancient glacial deposits may have been dee-

ply buried, modified by metamorphism or selectively preserved

in basins not subsequently affected by erosion. In addition, it is

more difficult both to interpret the paleoenvironment of older

deposits as different processes may have formed similar features,

and also to date them because fossils or other suitable dateable

material may be absent, and chemical (isotopic) composition of

waters and atmosphere may have been different or altered.

The following are the most common, effective lines of evi-

dence used to establish the occurrence of an ancient glaciation.

No single indicator, however, provides unequivocal proof of

glaciation; it is the assemblage of several criteria, abundantly

present in an area, which can be used to infer cold climate

and presence of glaciers.

Depositional features

The presence of widespread diamictite (thick when marine),

associated with cold-climate indicators is the most frequently

used criterion to establish glacial conditions. Diamictite is a

rock derived from diamicton – a poorly sorted mixture of

mud, sand and gravel (Figure G43). Diamicton may be depos-

ited directly by glaciers (till; tillite when cemented), or formed

by other sedimentary processes, such as rainout of fine-grained

suspended sediment and ice rafted debris, and debris flow in

terrestrial or subaqueous conditions. The presence of extensive

diamictite that contains faceted clasts is a strong initial clue of

glaciation, when tectonic influences (such as non-glacial debris

flow caused by tectonic activity) can be confidently discarded.

However, it is difficult to determine whether diamictite is of

glacial or non-glacial origin. It is therefore critical to look at

the characteristics and distribution of associated deposits in

order to establish the paleoclimatic significance of diamictite.

When abundant, shape and surface texture of clasts, from

bullet-shaped, faceted, and striated pebbles and boulders, to

imprints and chattermarks on sand grains, are powerful indica-

tors of glacial action. However, these may not always demon-

strate direct deposition by ice, as clasts may be reworked

under other depositional conditions. Although boulder pave-

ments are commonly used as glacial indicators in the Quatern-

ary, they are rarely identified in more ancient examples.

Dropstones (lonestones) and deposits with textural inversions

(such as randomly distributed lenses or pockets of sand

and/or coarser-grained material in fine-grained deposits) are

interpreted to indicate ice rafting and dumping from iceberg

and seasonal ice floes, although they need to be distinguished

from material rafted by floating organic debris (algae, trees)

in the Phanerozoic. Presence of sand, silt, and clay rhythmites

in lacustrine and marine deposits may contribute to a glacial

interpretation, although they may also be formed by turbidity

currents not necessarily derived from glaciers. True varves,

the typical clay-silt or silt-sand couplets formed annually in

ice-covered lakes, are rarely identified in pre-Quaternary suc-

cessions because it is usually impossible to confirm that each

couplet represents a year of deposition. Of interest is the fact

that loess, so common in Pleistocene periglacial settings, is

rarely reported from pre-Quaternary glacial deposits. This

may be a matter of poor preservation of exposed terrestrial

material; its ancient equivalent would likely be subaqueous

silty deposits. Depositional landforms such as drumlin and

eskers have rarely been used to establish glaciation, one of

the debatable exceptions being for the Carboniferous in Brazil.

Biological and geochemical features

Cold climate flora (mainly terrestrial) and fauna (terrestrial and

marine) and their related deposits, such as coal and ooze, are

strong indicators of cold settings during Phanerozoic times.

Isotope (such as O, C) analyses of skeletal marine microfossils

can also provide indication on the presence and amount of gla-

cial ice.

Erosional/deformation features

Abraded pavements on bedrock are often found with striations,

and some chatter marks, crescent gouges, and lunate fractures.

384 GLACIATIONS, PRE-QUATERNARY

Striated pavements are common throughout the ages (Laajoki,

2002). They can be formed by glaciers, but also by other unrelated

tectonic (slickensides) and depositional (debris flows) processes.

Large-scale erosional features such as roche moutonnées, crag

and tails, and rock channels have rarely been used to establish

glaciation, except in exhumed Ordovician surfaces in northwest

Africa. When extensive, furrows in marine and lacustrine deposits

are occasionally taken to indicate scouring by icebergs or seasonal

ice floes (such as for the Ordovician in Brazil). Associated with

other cold climate evidence, sandstone wedges have also been

used to infer frozen ground conditions.

Precambrian glaciations

Archean

Archean-age glacial deposits (2,990–2,870 Ma) have been

reported from South Africa. Diamictites are observed with

occasional striated and faceted clasts at three different strati-

graphic levels within the Witwatersrand Basin, both in outcrops

and subsurface mining walls. Diamictite up to 80 m thick con-

taining glacially striated and faceted clasts has also been found

in the Pongola Basin. Other diamictites have been observed

in northern Europe and North America but their glacial origin

is debated.

Paleoproterozoic

Glacigenic deposits of Paleoproterozoic age have been documen-

ted in North America (Huronian), South Africa (Makganyene),

Finland-Russia (Karelian) and western Australia (Hammersley).

Good exposures of Paleoproterozoic glacial deposits occur within

the Huronian Supergroup in Ontario, Canada. The Gowganda

Formation consists of diamictites with some faceted clasts over-

lying striated pavements, and thinly laminated argillites with

abundant dropstones. These diamictites have accumulated in a

predominantly glaciomarine setting. The laminated argillites were

thought to be varves formed in a glacially influenced, subaqueous

setting, but this cannot be confirmed in the absence of high resolu-

tion (annual) dating of these deposits.

The age of the Gowganda Formation has been bracketed

between 2,670 and 2,220 Ma by underlying volcanic rock

and a cross-cutting diabase. The ages of other Paleoproterozoic

glacigenic successions are equally poorly constrained and

range between 2,050 and 2,700 Ma. whereas the deposits in

South Africa, Finland-Russia and western Australia are thought

to be broadly correlative with the North American Huronian

glacial deposits, their synchronous deposition is speculative.

Neoproterozoic

The striated pavement and overlying diamictite exposed on the

shores of Varangerfjorden, Norway (Figure G43) were first

described in 1891 and used to infer a Precambrian-age glacia-

tion. This was one of the first Neoproterozoic glacial deposits

to be described; others have since been documented from every

continent (Crowell, 1999). Evidence for glacial conditions

includes striated pavements, striated and faceted clasts, marine

diamictite and laminated mudstone with dropstones.

Figure G41 Major tectonic, environmental and climatic change on Earth (after Doyle et al., 1995; with data from Berner, 1990; Young, 1999).

GLACIATIONS, PRE-QUATERNARY 385

Two major glacial periods have been established in the Neo-

proterozoic, with growing evidence for a third. Glacial deposits

are commonly ascribed to either of the two glacial periods

based on lithological characteristics, biostratigraphy (acritarchs

and Ediacaran fauna), chemostratigraphy (d

C) of associated

carbonates and radiometric dates (Hambrey and Harland,

1985; Knoll, 2000). The oldest glaciation (ca. 750–670 Ma)

is commonly referred to as Sturtian, whereas the youngest

(ca. 635 Ma) has been referred to as Marinoan (Australia,

Canada) or Varangian (North Atlantic). However, there is still

much debate about the age and stratigraphic correlation of

these glacial deposits and the number and extent of Neoproter-

ozoic glaciations (Knoll, 2000; Kendall et al., 2006).

Various aspects of the Neoproterozoic geologic record sug-

gest that glaciations of this time may have been more severe

than those of the Phanerozoic. The widespread distribution of

diamictites, including some at low paleolatitude, has been used

to infer that ice possibly covered the entire world (Harland,

1964). Inferred glacigenic deposits are commonly associated

with carbonates that exhibit strong fluctuations in d

C. The

close juxtaposition of glacigenic deposits and presumably

warm water carbonates has been used to suggest relatively

rapid climate change between glacial and warm conditions,

whereas the d

C fluctuations imply significant perturbations

in seawater geochemistry (Harland, 1964; Hoffman et al.,

1998). The low latitude (equatorial) deposition of some glacial

deposits has been suggested by paleomagnetic analyses,

although it has only been confidently demonstrated for the

Elatina Formation in Australia (Evans, 2000). Nevertheless,

Figure G42 Maps of locations showing the recorded glacial features

on Earth: (a) Precambrian; (b) Early Paleozoic; (c) Late Paleozoic;

(d) Cenozoic (after Hambrey and Harland, 1981; Martini et al., 2001;

and some data from Evans, 2000).

Figure G43 Neoproterozoic diamictite of the Smalfjord Formation

overlying a striated pavement, Bigganjargga, northern Norway.

386 GLACIATIONS, PRE-QUATERNARY