Earth-Science Reviews 201 (2020) 102974

Contents lists available at ScienceDirect

Earth-Science Reviews
journal homepage: www.elsevier.com/locate/earscirev

Megaflooding associated with glacial Lake Agassiz T

Timothy G. Fisher
Department of Environmental Sciences, MS 604 University of Toledo, 2801 West Bancroft St., Toledo, OH 43560-3390, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Significant baseline and episodic megaflooding from glacial Lake Agassiz was routed to the south, east and north
Outlets coasts of North America over the lake’s nearly 6000 year history. The five phases of lake-level change were
Boulders controlled by which outlet was active, which in turn was controlled by ice margin position and glacioisostatic
Flood routing adjustment. The southern outlet is the oldest and best understood outlet, while successively younger outlets are
Spillway
progressively less understood. Eastern drainage synchronous with the Younger Dryas chronozone had been
Younger dryas
Preboreal Oscillation
assumed by most studies, but a spillway to accommodate the ∼90 m drop in lake level is yet to be described.
With the south and east outlets unable to accommodate the necessary lake level drawdown, the northwest outlet
has become the default outlet with Arctic Ocean Core data supporting it. However, extant terrestrial data from
the continent only provides data for a large flood from the Fort McMurray area closer to the end of the Younger
Dryas, coinciding with the timing of the Preboreal Oscillation. The challenge remains to find agreement between
the marine and terrestrial records. Uncontroversial is the geomorphic and sedimentologic evidence for flooding
into and out of the lake consisting of large spillways, large boulders, and in places giant current ripples composed
of boulders. Flood discharge estimates into and out of the lake range from 0.04 to ∼1 Sv, with final subglacial
drainage into the Tyrell Sea estimated at ∼5 Sv, sufficient to raise global sea level by 0.18 m and initiate the 8.2
ka stadial.

1. Introduction corrections and identifying specific sources of the meltwater (i.e., water
from Lake Agassiz or some other lake or meltwater source). Terrestrial
Regional interest in glacial Lake Agassiz became international in- records are hampered by radiocarbon ages giving only maximum or
terest when Broecker et al. (1989) proposed that meltwater delivery to minimum ages of sediments and landforms—seldom the depositional or
the Gulf of Mexico switched to the North Atlantic and triggered a slow- formation age. Optically stimulated luminescence (OSL) dating can
down in thermohaline circulation initiating the Younger Dryas stadial. provide ages of deposition, but suffers from larger errors than radio-
Dubbed by many as the world’s largest glacial lake, Lake Agassiz carbon dating.
evolved over nearly 6000 years through a series of configurations across There is ample evidence of megaflooding into and out of the lake
the center of the North American continent (Fig. 1). Large volumes of consisting of large spillways and characteristic flood deposits. In par-
meltwater entered the lake along its western and ice-contact sides and ticular, two very large drops in lake level have been proposed. The first
were episodically routed to the southern, eastern, and northern sides of released ∼17,000 km3 near the start of the Younger Dryas
North America by outlet-sill incision, deglaciation of lower outlets and (Breckenridge, 2015), and 163,000 km3 when the lake finally drained
glacioisostatic adjustment (GIA). However, an understanding of routing near the start of the 8.2 ka event (Clarke et al., 2004). After a brief
history is incomplete. The routing problem has been approached from overview of lake evolution, lake phases are discussed emphasizing
two different directions—from marine basins looking inland for sources water routing, and followed by the megaflooding history associated
of meltwater, and from spillway channels delivering meltwater to the with various inlets and outlets. This review briefly summarizes perti-
oceans. Oxygen isotope and other proxy data from marine sediment nent Lake Agassiz meltwater routing history, and evidence of mega-
cores adjacent to the Mississippi, Saint Lawrence, and Mackenzie rivers flooding from Lake Agassiz. Uncalibrated radiocarbon ages have the
record fluctuations in meltwater delivery, the timing of which is ideally units 14C yr (or kyr) BP, calibrated radiocarbon ages use cal yr (or kyr)
paired with terrestrially-based reconstructions of ice-margin recession, BP, and OSL or cosmogenic ages use ka.
lake evolution, and drainage. Complicating the problem is reliable
chronology. Marine cores are hampered by carbonate reservoir

E-mail address: timothy.fi[email protected].

https://doi.org/10.1016/j.earscirev.2019.102974
Received 27 September 2018; Received in revised form 30 September 2019; Accepted 3 October 2019
Available online 11 December 2019
0012-8252/ © 2019 Elsevier B.V. All rights reserved.
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 1. Overview of the maximum extent of Lake Agassiz occupied throughout its nearly 6000 year history. Panel A modified from Leverington and Teller (2003) and
panel B modified from Dilworth and Fisher (2018).

2. Lake phases, levels and routing controlled by ice margin and projected to eastern outlets, it is possible that up to three outlets
position were active at once following the transgression at the end of the
Moorhead Phase. The Nipigon Phase eastern (Kelvin) outlets finally
Glacial Lake Agassiz formed once the Des Moines Lobe of the captured drainage as the LIS continued its recession northwards, un-
Laurentide Ice Sheet (LIS) receded north of the sub-continental drai- covering progressively lower outlets as recorded by the suite of beaches
nage divide separating the Gulf of Mexico and Hudson Bay watersheds at successively lower elevations than the Campbell Beach (Fig. 2).
(Upham, 1895). Meltwater was trapped between the receding lobe and During this phase water drained into Lake Kelvin and out through the
high topography of the Big Stone Moraine (Fig. 1A) by about 14 ka Nipigon channels into the Lake Superior basin, through the Great Lakes
(Lepper et al., 2007). Opening or closing major outlets routing water to basin and St. Lawrence River, and eventually into the North Atlantic
different oceans defines lake phases (Fig. 2). Volumes of water have Ocean (Fig. 1). With continued LIS recession glacial Lake Agassiz
been estimated for these phases (Leverington et al., 2000, 2002; eventually merged with glacial Lake Ojibway (Ojibway Phase), its
Table 1). The first major lake phase is the Lockhart Phase (Fig. 2) when counterpart further east in the Hudson Bay lowlands. Drainage was
drainage was through the southern outlet. The start of the Moorhead south through the Kinojévis outlet into the Ottawa River and then St.
Phase is thought to be synchronous with the start of the Younger Dryas Lawrence River. Final drainage of the lake occurred as the ice dam
chronozone (YD) when drainage shifted away from the southern outlet thinned and Lake Agassiz drained subglacially into the Tyrell Sea, the
to some combination of eastern or northwestern outlet, and/or eva- precursor of Hudson Bay. The final drainage is generally assumed to be
poration. The Moorhead Phase ended and the Emerson Phase began responsible for the 8.2 ka cooling event (Barber et al., 1999; Clarke
when flow returned to the southern outlet with lake levels at the et al., 2004). Detailed paleogeography maps of the different lake stages
Campbell Beach (cf. figures in Fisher et al., 2011). Because the Camp- and eastern routing of meltwater are available in Leverington et al.
bell Beach has been mapped to the southern and northwestern outlets, (2000, 2002), Leverington and Teller (2003), and Teller and

2
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 2. Lake Phase history of glacial Lake Agassiz with age control shown for the three oldest phases. Note change in elevation scale at 9.9 ka.

Table 1 strandlines on the west-facing slope (Fig. 4A). Black dots mark suc-
Select Volume and Area of Lake Stages (modified from Leverington et al., 2000, cessively lower water planes in both DEMs and the number of dots
2002) between the Tintah Beach and highest strandline (Herman?) is instead
Lake stage Lake Phase Area (km2) Volume (km2) similar on both maps. Strandlines below the labeled Norcross Beach
west of Fargo are very subtle; resembling the Norcross strandlines de-
Herman Lockhart 134,000 10,900 scribed by McMillan and Teller (2012) in southern Manitoba that they
Norcross Lockhart 166,000 13,300
suggest are storm beaches. The much larger and complex strandlines in
Tintah Lockhart 184,000 15,700
Early Moorhead 117,000 10,800
Fig. 4A offer greater opportunity for reconstructing paleo water planes
Late Moorhead 185,000 19,700 and understanding littoral processes that reflects control by sediment
Upper Campbell Emerson 263,000 22,700 supply, fetch, and aspect. The strandlines in Fig. 4A are currently under
McCauleyville Nipigon 219,000 16,400 study as they offer exceptional resolution of distinct water planes with
The Pas Nipigon 151,000 4600
large strandlines.
Kinojévis Ojibway 841,000 163,000
Fidler Ojibway 408,000 49,900 Strandline chronology of the Herman through Tintah strandlines has
been determined from primarily OSL dating because wood has not been
found within them, and which is not from a lack of effort—presumably
Leverington (2004). any wood was oxidized. While bones have been dated from the
Campbell Beaches (e.g., Nielsen et al., 1984) most radiocarbon dates
are from underlying basal organics (often peat) in wetlands dammed by
3. Lockhart phase
beach ridges (e.g., Bjorck and Keister, 1983; Teller et al., 2000).
Available OSL ages for the Lockhart Phase are shown on Figs. 2 and 4C.
Flights of strandlines from the Herman dropping to the Tintah Beach
Those by Lepper et al. (2011, 2013) cluster tightly and are from the
(Fig. 2) developed as the lake expanded northwards during the Lockhart
beach type area ∼100 km from the southern outlet. Those dated by
Phase through a combination of southern outlet incision and GIA. With
Teller et al. (2018) in southern Manitoba are weakly developed, contain
the southern outlet at an isobase with the lowest rate of uplift for the
poorly-sorted sediment, and are 400 km from the southern outlet where
lake basin, higher rates of rebound northwards resulted in strandlines
the higher number of beaches makes water plane assigned more diffi-
warped upwards and an increasing number of strandlines for any
cult. Most errors associated with OSL beach ages are ± ∼10% (Aitken,
named beach (Upham, 1895; Leverett, 1932; Johnston, 1946; Elson,
1998). However, Lepper et al. (2011, 2013) and Liu et al. (2014) rely on
1967; Teller and Thorleifson, 1983; Teller, 2001; Fisher, 2005; Yang
a methodological approach in Lepper et al. (2011) that also reports just
and Teller, 2012; Breckenridge, 2015). For example, there is a single
analytical error (∼3%; Fig. 4C). A test of the veracity of the OSL age is
Herman Beach graded to the southern outlet but more than a half dozen
shown in Fig. 4C, where the geomorphic record of Lockhart strandlines
in northern North Dakota up to 70 m higher in elevation. Episodic
records a regression and thus decreasing ages with decreasing elevation
outlet incision is recognized by the sequence of four strandlines graded
(Herman through Tintah strandlines). Independent support for this re-
to the southern outlet at successively lower elevations (Herman, Nor-
lationship is the decreasing gradient of lower elevation strandlines re-
cross, Upham, Tintah: Fisher, 2005) and terraces within the spillway
flecting lower rates of GIA through time, a observation made by all who
(Fig. 3).
have mapped these strandlines, and most recently by Breckenridge
Examples of strandlines are shown on Fig. 4 from the east side of the
(2015). Thus strandline age correlates with elevation, which is illu-
basin, southeast of Grand Forks (Fig. 1) and on the west side of the
strated on Fig. 4C where OSL ages decrease in age down slope. The OSL
basin, west of Fargo, ND. There is a striking difference in appearance of
record within the gray panel (Fig. 4C) is the time window available for
the strandlines between the two sites with seemingly many more

3
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 3. Upper segment of the southern outlet spillway by the town of Browns Valley. Terraces armored with boulders are common along the spillway with a few
examples mapped adjacent to the streamlined islands within the spillway. Core panel is modified from Fisher (2003).

4
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

strandline formation. Its older bracket is deglaciation of the Big Stone

Moraine and initiation of Lake Agassiz (Lepper et al., 2007) and the
younger bracket is the oldest in-situ dated terrestrial organics from the
Redwood Loop site (Fisher et al., 2008) that grew after lake level fell,
signaling the end of the Lockhart Phase. The OSL ages from southwest
of Fargo (Fig. 4C) are from sand deposited subaerially during the
Moorhead Phase (Liu et al., 2014). Note that nearly all of the OSL ages
fall within the gray panel, and the ages with their less conventional
reported error assignment agree with the geomorphic strandline record,
and radiocarbon dated stratigraphy.
During the Lockhart Phase large influxes of meltwater from cata-
strophic drainage of ephemeral glacial lakes to the west of Lake Agassiz
entered the lake through the Sheyenne, Pembina and Assiniboine rivers,
and is further discussed in Section 1.8. Terraces (a few mapped on
Fig. 3) within the southern outlet spillway (also known as River
Warren) may be associated with some of the strandlines and have
suggested to some workers that outlet incision was in response to
knickpoint migration (Matsch et al., 1983). Two sediment Cores
through lacustrine and gravel sediment into shale bedrock within the
southern outlet record a minimum two episodes of outlet activity
(Fisher, 2003). A thin lacustrine mud interbedded between gravel in
Core BVF-99 (Fig. 3) was interpreted to record the Moorhead Phase
when the southern outlet was abandoned.

4. Moorhead phase

The Moorhead Phase began when lake level dropped below the
southern outlet, and ended when a transgression (e.g., Teller, 2001)
returned flow to the southern outlet (Fig. 2). The chronology and
meltwater routing is uncertain where the lake drained at this time to
cause the ∼90 m drop in lake level (Breckenridge, 2015). Water levels
fell from the Tintah Beach to a short distance north of Grand Forks, ND
(Dilworth and Fisher, 2018). Drainage through eastern outlets into the
Great Lakes basin following abandonment of the southern outlet was
initially proposed by Upham (1895). However, few strandlines have
been mapped on the mostly bare Canadian Shield bedrock, which has
resulted in previous workers projecting strandlines from the main
Agassiz basin eastward to low passes along the subcontinental drainage
divide separating Hudson Bay and the Great Lakes basin. Historically,
possible outlets have been described south and north of the Kaiasch
Moraine (Elson, 1967; Zoltai, 1967; Teller and Thorleifson, 1983), here
simply referred to as the Kam and Kelvin outlets respectively (Fig. 5).
Detailed field investigations documenting any of the Kam outlets to the
start of the Moorhead Phase have never been reported. The most recent
projection of strandlines to any possible Kam outlets was by
Breckenridge (2015), but the results were much the same as earlier
work. Though Breckenridge emphasized that any early Moorhead flow
would have begun at the lowest Tintah level. The outlets north of the
Kaiasch Moraine have been associated with the younger Emerson Phase
drainage, which is discussed in section 1.5. The original reconstruction
of a > 40 m water level drop (Broecker et al., 1989) explained by
eastward routing of Agassiz to the North Atlantic is problematic be-
cause nowhere along the drainage divide is there a col with enough
relief to explain the drop in lake level. The low lake level reached
during the Moorhead Phase just north of Grand Forks, North Dakota
(Fig. 1; Dilworth and Fisher, 2018) projects well below any cols on the
Fig. 4. Strandline examples from glacial Lake Agassiz. See Fig. 1A for location. eastern drainage divide.
A) West facing slope in Minnesota illustrating a high number of individual The drop in water level at the start of the Moorhead Phase is re-
strandlines. Black dots represent distinct water planes, white dots possible corded by: 1) the transition to mud from gravel in lakes of the southern
strandlines within the Tintah gap area. Center of DEM is at 47.6671 °N, outlet, 2) the youngest age from a strandline abandoned when lake
96.3375 °W. B) Strandline examples along an east facing slope previously stu- level fell, 3) the oldest ages from within the basin exposed when lake
died by Lepper et al. (2011) as the Wheatland transect. Black dots represent level fell, 4) deglacial ages where the new outlets opened, and 5) the
distinct water planes. Center of DEM is at 46.9188 °N, 97.3859 °W. age of freshwater signals in marine sediment. The youngest wood
within basal gravel from the southern outlet spillway is
12,550–12,720 cal yr BP (10,675 ± 60 14C yr BP), which is a max-
imum age for when the southern was abandoned. The Tintan Beach is

5
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 5. Hillshaded DEM of the eastern outlet regions identifying major outlets. Flow from the Kelvin outlets was buffered by Lake Kelvin before passing through the
Nipigon outlets into the Lake Superior basin.

the lowest strandline graded to the southern outlet that is above the best minimum estimate for lake level fall, but plant colonization time
Campbell strandline and that records the end of the Moorhead Phase. must be taken into account. The oldest root from a drowned forest is
Only two OSL ages are available for the Tintah Beach and they average 12,540–12,740 cal yr BP (10,710 ± 75 14C yr BP, ETH-32674) and the
13.5 ± 1.1 ka (Lepper et al., 2011). The areal gap of strandlines be- youngest in situ organics (12,250–11,760 cal yr BP 10,000 ± 70 14C yr
tween the Tintah and Campbell beaches was used by Breckenridge BP, ETH-32679) are ∼1100 years younger. Based on the analysis of
(2015) to argue for rapid lake level fall at this time. The Tintah Beach available radiocarbon dates (Fig. 8 in Fisher et al., 2008) lake level fall
has not been mapped to any eastern outlet. High resolution LiDAR was ∼12.7 cal kyr BP plus time for colonization (∼100 years?), which
images (Fig. 4B) show possible weakly developed strandlines within gives an age close to the southern outlet chronology (Fig. 3; Fisher,
this gap area, but require additional work to confirm. After lake level 2003), and is overlapped by the OSL ages from the Tintah Beach, and
fell organic sediment could accumulate on the freshly exposed land- oldest peat (12,150–12,860 cal yr BP [10,700 ± 140 14C yr BP, WAT-
scape by fluvial deposition and plant growth at elevations below the 1910]) from the Rainy River Lowland (Bajc et al., 2000) in northwest
Tintah Beach. As discussed at length by Fisher and Lowell (2006) and Ontario (R on Fig. 1) that grew after lake level fell.
Fisher et al. (2008), the oldest in situ terrestrial organic material is the Streams flowing across the exposed lake basin during the Moorhead

6
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Phase deposited fluvial sand of the Poplar River Formation (PRF) on ages (cf. Fisher and Lowell, 2012) are Moorhead Phase aged. If this
lacustrine sediment of the Brenna Formation of Lockhart age. The PRF flood occurred at the start of the YD, then the older wood should pre-
was buried by lacustrine sediment of the Sherak Formation (Arndt, date the YD. While extant geomorphic and chronologic data from the
1977) during the Moorhead transgression leading to reopening the Agassiz basin thus far do not support northwest drainage at the start of
southern outlet. From two sites just south of Fargo, ND, the PRF is the YD (Fisher and Lowell, 2012), it has become the default outlet
expressed as a laterally continuous ribbon sand, with an average age of following initial drainage to the east (Breckenridge, 2015; Leydet et al.,
13.1 ± 1.6 ka (Fig. 2) that may record a regressive sand deposited 2018). Calculations demonstrating the feasibility of the Moorhead
when lake level fell (Liu et al., 2014). Radiocarbon dates from the Po- Phase to have resulted from enhanced evaporation as a closed basin
plar River Formation that are primarily channel sands elsewhere in the (Lowell et al., 2013) have not been well received (Carlson and Clark,
Fargo area are younger, beginning around 11,350 cal yr BP (Yansa and 2012; Teller, 2013; Breckenridge, 2015).
Ashworth, 2005). The available data from the Fort McMurray area of minimum
Before a new outlet can open to cause abandonment of the southern flooding ages of 11.3 ± 0.1 to 11.0 ± 0.2 ka BP overlaps with the
outlet, ice must deglaciate lower outlets. Minimum ages from lakes and Preboreal Oscillation (PBO) (Hald and Hagen, 1998). Fisher et al.
cosmogenic ages on boulders were used by Lowell et al. (2005, 2009) to (2002) suggested that drainage through the CLAS caused the PBO
suggest that ice had not retreated far enough eastward to open lower through increased output of freshwater and sea ice export to the North
spillways into the Great Lakes basin. More recently, Leydet et al. (2018) Atlantic Ocean through Fram Strait, and these ages overlap with re-
obtained older cosmogenic ages from boulders that suggest deglaciation interpreted (Carlson and Clarke, 2012) ages from the Mackenzie delta
was as early as 14.0 ± 0.4 ka to route meltwater eastwards before the (Murton et al., 2010).
start of the YD stadial. However, as most recently pointed out by Marine records have also been used to determine meltwater routing
Breckenridge (2015), none of the identified possible Kam spillway from Lake Agassiz. Data from the Gulf of Mexico record abandonment
routes (Fig. 5) have the relief necessary to drop the lake ∼90 m to cause of the south outlet (Broecker et al., 1989; Flower et al., 2004; Williams
the flood invoked by many earlier researchers. Deglaciation of eastern et al., 2010; Wickert et al., 2013) which requires diversion of meltwater
outlets may have initially dropped water levels at the start of the elsewhere. Eastern drainage through the Great Lakes basin is supported
Moorhead Phase if the magnitude of the drop exceeded the vertical gap by marine proxy data for freshwater beyond the St. Lawrence River
between the Tintah and Campbell beaches. A search for thick sedi- (Carlson et al., 2007; Levac et al., 2015) and evidence for freshwater
mentary sequences offshore of Thunder Bay, Ontario (Fig. 5) failed, excursions closer to land (Rayburn et al., 2011a, b; Cronin et al., 2012).
suggesting to Voytek et al. (2012) that no significant megaflooding Once the St. Lawrence River opened following drainage of Lake St.
occurred in this direction. If initial drainage was to the east before the Lawrence (Rodrigues and Vilks, 1994; Levac et al., 2015) drainage from
start of the YD and a connection to the North Atlantic Ocean through the Great Lakes is rerouted eastwards along with meltwater runoff from
the Great Lakes basin was not available, routing could have instead a large portion of the LIS. Data from the Arctic Ocean basin has also
been out the Lake Superior basin through the Brule spillway or Chicago been used to support drainage from Agassiz northwards at the start of
outlet to the Mississippi River (B and C, respectively Fig. 1) if ice had the YD (Hillaire-Marcel et al., 2013; Keigwin et al., 2018) to explain
retreated far enough before the Marquette readvance (Lowell et al., freshwater inputs. Modeling of meltwater runoff by Tarasov and Peltier
1999). Such routing would not interrupt the meltwater signal in the (2005, 2006) suggested that meltwater runoff from the Keewatin Dome
Gulf of Mexico. Because any eastward drainage cannot explain the of the LIS alone could explain freshening in the Arctic Ocean, but was
complete drop in lake level during the Moorhead Phase, drainage out of criticized by Carlson and Clark (2012) for their enhancing precipitation
the northwest outlet has been proposed (Teller et al., 2005; during the YD. Condron and Winsor (2012) supported northern versus
Breckenridge, 2015; Leydet et al., 2018). eastern drainage as freshwater delivery was more effect in their mod-
A northwest outlet was considered by Upham (1895), and Elson eling results at weakening circulation in the North Atlantic Ocean. Al-
(1967) identified a series of channels in the upper Churchill River valley ternative sources of meltwater beyond North America are from the
(Fig. 6), now termed the Wycherely channels mapped by Fisher and Greenland Ice Sheet that is much closer to sites of deep-water formation
Souch (1998). Further to the northwest is the large Clearwater lower- (Jennings et al., 2006) and drainage from the Baltic Ice lake
Athabasca spillway (CLAS) that was initially interpreted as the north- (Muschitello et al., 2016).
west outlet for Lake Agassiz (Smith and Fisher, 1993; Fisher and Smith, The routing of meltwater at the start of the Moorhead Phase is
1994), and is further described in section 1.11. Strandlines well above poorly understood. Available chronology from the Lake Agassiz basin
the head of the spillway were assumed to be a part of Lake Agassiz and supports an abandonment of the southern outlet close to the start of the
labeled as Norcross. Since then, Fisher (2005) retracted the Norcross YD, but field data revealing when routing transitioned is not available.
interpretation, primarily because the Norcross strandline does not ex- Evidence for meltwater excursions at the start of the YD from marine
tend north of southern Manitoba (Fig. 7) (cf. Fisher and Lowell, 2012). records off shore of the Saint Lawrence and Mackenzie Rivers compete
Instead the high strandlines above the head of the CLAS would now be for drainage from Lake Agassiz. The lack of field evidence for stran-
associated with glacial Lake Churchill in the upper Churchill valley, a dlines extending to the CLAS spillway at the start of the YD may reflect
product of merging formerly mapped glacial Lake McMurray in Alberta their erosion by younger lakes (Breckenridge, 2015) or a readvance.
and glacial Meadow Lake in Saskatchewan (Fisher et al., 2009), and Additional work is required to evaluate the megaflooding history from
drainage of glacial Lake Churchill is the source of water that eroded the the Fort McMurray area and sites further north where other large lakes
CLAS. Glacial Lake Churchill strandlines in this region remain undated. once existed (e.g., Smith, 1992; Lemmen et al., 1994), collect radio-
Minimum deglacial ages of 11.3 ± 0.1 to 11.0 ± 0.2 cal kyr BP from carbon ages from Lockhart-aged beaches, and to collect chronology
the youngest moraine (Fort Hills Moraine) in the Fort McMurray area data from spillways identified in northwest Ontario.
are a minimum limit for when the flood occurred (Fisher et al., 2009). If The strandline record during the Moorhead Phase is poorly under-
a hundred years are invoked for a lag time before plant colonization stood. Because the lake transgressed over the lake basin at the end of
occurred, then this chronology overlaps with OSL ages (11.9 ± 1.0, the phase, preservation potential of strandlines is low (Teller, 2001;
11.8 ± 1.0, 11.5 ± 0.7 ka) of sand conformably overlying the oldest Breckenridge 2015). Fisher et al. (2008) dated plane bedded sand they
flood gravel on the Mackenzie delta (Murton et al., 2010). Ages on considered a Moorhead Phase beach (Ojata) as it was buried by fine-
wood associated with flood gravel in the Fort McMurray area, and in grain lacustrine sediment. More recently Teller et al. (2018) suggested
the resultant delta built into Lake McConnell (Smith and Fisher, 1993) that the Gimili Beach in southern Manitoba (G on Fig. 1A) may also be a
are expected to be older than the flood because the Athabasca River can Moorhead Phase beach. The previously undated Gimili Beach had al-
transport wood from an older landscape westward, and the oldest wood ways been assumed to be a Nipigon Phase beach because of its lateral

7
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 6. Hillshaded DEM of the northwestern extent of Lake Agassiz within the upper Churchill River basin. See Fig. 1 for location. At the Campbell level flow was out
a series of the four Wycherley channels (B) into a broad lowland occupied by modern Lakes Wasekamio, Turnor and Frobisher (Lake Wagtufro) and then into the
already formed Clearwater lower-Athabasca spillway (CLAS) and eventually the Mackenzie River. (C) View (white arrow on B) of one channel looking east towards
Wycherley Lake. Panel B is modified from Fisher and Souch (1998).

extent (50 km) and low slope, but the much older than expected OSL (Table 2), some of which were discussed in Lepper et al. (2013) and
ages (Teller et al., 2018, Fig. 2) complicates this understanding. Older agree with the OSL ages. Three of the OSL ages (samples AB0710,
sand may be present in the beach that is not representative of the AB0711, E2; all not plotted on Fig. 2) are older than most Campbell
geomorphic landform, similar to older ages encountered elsewhere in ages and may result from Moorhead-aged sand around which the
other Lake Agassiz beaches (Lepper et al., 2013). Campbell Beach formed, thus the younger OSL ages are considered
more limiting and the E1 sample from Manitoba (Teller et al., 2018)
agrees with the Campbell ages near and at the southern outlet type area
5. Emerson phase (Lepper et al., 2013). Wood ages from within beaches are considered
maximum ages for those beach deposits. Where lagoons formed behind
Much of the earlier literature assumed that water drained eastwards Campbell beaches, the youngest basal dates from the lagoons are good
during the Moorhead Phase, and the Emerson Phase began when flow estimates for the Campbell Beach age. Many of these ages in Table 1
returned to the southern outlet and ended when drainage switched back overlap in time with the OSL ages, and the duration of the Emerson
to the east. Water level at this time is recorded by the Campbell Beach, Phase may be estimated from 10.7 to 10.0 ka.
which is divided into Upper and Lower beaches. The Campbell beaches Deglacial ages from the Kelvin outlets region that overlap in age
grade to the southern outlet and are projected to the eastern Kelvin with the Emerson Phase are from Lowell et al. (2005, 2009) and Teller
outlets (Figs. 5 and 8) and the Wycherely channels as the northwestern et al. (2005) from minimum limiting ages in lakes near moraines, and
outlet (Fig. 1). If an eastern or northwestern outlet was active during cosmogenic ages from boulders. In northern Saskatchewan, Schreiner
Moorhead time, then it was GIA driving the transgression that returned (1983), Fisher and Smith (1994), and Rayburn and Teller (2007)
flow to the southern outlet. mapped Campbell Beaches northwards from Nipawin, Saskatchewan
The duration of the Emerson Phase is perhaps best estimated by the along the eastern side of the Cub and Wapawekka Hills (Fig. 6). Well-
range of ages on the Campbell Beach. Flow returning to the southern developed strandlines and strandline fragments north and east of the
outlet is recorded by gravel overlying lacustrine mud and bedrock Wapawekka Hills were only associated with the Lower Campbell
within the spillway (Fisher, 2003). The youngest piece of wood from strandline by Rayburn and Teller (2007). Further to the northwest
the gravel is 10,520–11,080 cal yr BP (9460 ± 70 14C yr BP, Fisher and Souch (1998) mapped the Wycherely channels, and assumed
AA38302), and is the most constraining maximum age for reopening they drained Lake Agassiz at the Campbell level to the northwest into a
the southern outlet. The OSL chronology for the undifferentiated small regional lake dubbed Lake Wagtufro, which drained into the head
Campbell Beach is shown in Fig. 2, with each age listed in Table 2. of the Clearwater lower-Athabasca spillway (CLAS, Fig. 6). This
There is a substantial suite of radiocarbon ages on the Campbell Beach

8
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 7. Hillshaded DEM of the Manitoba Escarpment on the east side of the hills and mountains. Until ice lobes retreat northwards west of the escarpment, drainage
from the ice sheet and uplands was to the west and south into the Assiniboine River. On the eastern side of the escarpment, ice retreat eastwards and northwards
causing new spillway channels to drain into the reentrants along the east side of escarpment. North of Duck Mountain only the Campbell Beach has been mapped.

reconstruction requires that the CLAS spillway had already formed. 6. Nipigon phase
Radiocarbon dates and age assignments based on sedimentation rates
from lakes adjacent to the Wycherley channels and Lake Wagtufro gave Outflow through the eastern Kelvin outlets was into a highstand of
an age of 10.2–10.8 cal kyr BP for cessation of northwest drainage from Lake Nipigon known as Lake Kelvin. Water then passed through the
lake Agassiz, which overlaps with OSL ages (Lepper et al., 2013) from Nipigon channels into the Lake Superior basin (Fig. 5). Thick sequences
the Campbell Beaches elsewhere in the lake basin (10.0–10.7 ± 0.9 ka) of sediment within troughs of Lake Superior just offshore of Nipigon
and dates from boulders (Kelly et al., 2016) in the Kelvin outlets record high sediment loads carried through the Nipigon channels
(11.1 ± 0.4–10.8 ± 0.2 ka). Whether flow first occurred to the east, (Teller and Mahnic, 1988; Gary et al., 2012) as evidence for cata-
northwest or south is unknown, but flow was eventually captured by strophic flooding from Lake Agassiz.
the Kelvin outlets ending the Emerson Phase and starting the Nipigon Detailed paleogeographic maps of ice margin retreat and meltwater
Phase. Synchronous flow through three different outlets was possible routing through successively lower outlets west (Figs. 5 and 8), then
for a limited time, similar to the three-outlet history of mid-Holocene north of Lake Nipigon are associated with the suite of 14 strandlines
Lake Nipissing in the upper Great Lakes basin (Thompson et al., 2011). (Fig. 2) recording episodic regressive lake level fall in the southern
basin of Lake Agassiz (Leverington and Teller, 2003). Opening each
successively lower outlet has been postulated to result in a brief

9
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 8. Oblique aerial photographs of two examples of channels associated with the Kaiashk spillway system, location shown in Fig. 5. In A) the height of the cataract
above the Devil’s Lake plunge pool is about 75 m. In B) Ottertooth Lake is 3 km long in this image and narrows to 0.25 km wide.

catastrophic flood until lake level stabilizes at a new sill. The resulting drainage divide between Hudson Bay and the Atlantic Ocean and lasted
strandlines following the successive drop in lake level are all younger for about 2100 years as determined from varve counts (Breckenridge
than the Campbell beaches—but the only age control is based on their et al., 2012). Overflow was south to the Ottawa River through the Ki-
tilt, becoming younger with lower gradients. Some age estimates have nojévis outlet (KinO, Fig. 1). Proglacial lakes Agassiz and Ojibway
been proposed using this relationship (Yang and Teller, 2012). As ice merged at about 9040 cal BP (Breckenridge et al., 2012), with the Lake
recession continued northward, drainage was eventually diverted from Agassiz name preserved. Fluctuations in water level are recorded by
the Great Lakes basin as Lake Agassiz merged with glacial Lake Ojibway scarps (Roy et al., 2015), and Roy et al. (2011) described a 0.6 m thick
when lake level transitioned between The Pas and Gimili levels (Fig. 2; laminated unit with rounded clay balls and other clasts interbedded
Leverington and Teller, 2003). Using varve thickness records from the between lacustrine and marine sediments recording final drainage,
Lake Ojibway basin, Breckenridge et al. (2012) estimated this occurred building on earlier work by Skinner (1973). Additional analysis on this
at 9040 cal yr BP. unit by Breckenridge et al. (2012) and Stroup et al. (2013) alternatively
suggest a Lake Ojibway sedimentary deposit predating final drainage of
the lake.
7. Ojibway phase and final drainage The final drainage of Lake Agassiz is hypothesized as one or nu-
merous subglacial channels forming beneath the thinning ice sheet over
As Lake Agassiz was further expanding northwards against the re- modern Hudson Bay, which resulted in catastrophic drainage in one or
ceding ice sheet margin, further east proglacial Lake Ojibway was de- a few stages (Clarke et al., 2004). Geomorphic indicators for the
veloping over northeastern Ontario and northwestern Quebec. With an flooding include meltwater channels kilometers in width, fields of giant
initial stage known as glacial Lake Barlow (Vincent and Hardy, 1979) current ripples and arcuate iceberg scours recording unidirectional
glacial Lake Ojibway formed between the receding ice margin and the

Table 2
Campbell Beach Chronology.
Sample ID dated material Agea Comment Reference

Age of sand deposition—OSL

KG0505 fine sand 10.0 ± 0.9 ( ± 0.2) ka Undifferentiated Campbell Beach (CB) Lepper et al. (2011)
KG0505 very-fine sand 10.3 ± 0.9 ( ± 0.2) ka Undifferentiated CB Lepper et al. (2011)
AB0704 fine sand 10.7 ± 0.9 ( ± 0.3) ka Undifferentiated CB near Campbell, MN Lepper et al. (2013)
KL0902 fine sand 10.6 ± 0.9 ( ± 0.2) ka Undifferentiated CB near Campbell, MN Lepper et al. (2013)
KL0903 fine sand 10.7 ± 0.9 ( ± 0.2) ka Undifferentiated CB near Campbell, MN Lepper et al. (2013)
AB0710 fine sand 12.0 ± 1.0 ( ± 0.3) ka Undifferentiated CB near Campbell, MN Lepper et al. (2013)
AB0711 fine sand 13.6 ± 1.1 ( ± 0.3) ka Undifferentiated CB near Campbell, MN Lepper et al. (2013)
E1 fine sand 10.1 ± 0.7 Upper Campbell (UC) Beach Teller et al. (2018)
E2 fine sand 11.7 ± 1.4 UC Beach Teller et al. (2018)

Wood within beach sedimentb

TO-4855 wood frag. 10,250 – 10,770 (9340 ± 90) youngest age in beach sediment Teller et al. (2000)
GSC-5357 wood 11,320 – 12,030 (10,100 ± 90) transported wood in UC beach McNeely and Nielsen (2000)
GSC-5330 wood 11,210 – 11,710 (9940 ± 80) in sand of LC Beach McNeely and Nielsen (2000)
GSC-4732 wood 11,210 – 11,710 (9940 ± 80) in sand of LC Beach McNeely and Nielsen (2000)

Maximum ages from organics underlying beach or resultant lagoon

unknown peat 10,250 – 10,790 (9350 ± 100) peat immediately below beach sand Bjorck and Keister (1983)
TO-4856 charcoal 10,280 – 10,720 (9330 ± 80) basal lagoon sediment Teller et al. (2000)
TO-4873 organic frag 10,280 – 11,070 (9380 ± 90) from silty clay in base of lagoon Teller et al. (2000)
TO-4284 peat 10,500 – 11,100 (9460 ± 90) base of lagoon Teller et al. (2000)
GSC-5731 wood 11,110 – 12,120 (9990 ± 160) underlying LC Beach (max age) McNeely and Nielsen (2000)
BGS-1305 plant detritus 11,150–11,820 (9920 ± 110) soil buried by offshore bar or spit Bajc et al. (2000)
BGS-1304 wood 10,650 – 11,720 (9750 ± 170) beneath UC Beach Bajc et al. (2000)
WAT-1934 wood 10,490 – 11,220 (9530 ± 140) beneath UC Beach Bajc et al. (2000)
AA-50803 wood 10,570 – 11,090 (9490 ± 70) youngest age from valley fill beneath UC Beach Fisher et al. (2008)

a
For OSL ages, propagated error shown with analytical error in parentheses, cf. Lepper et al. (2011).
b
Radiocarbon ages in parentheses.

10
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 9. Hillshaded DEM of the interior plains west of Lake Agassiz illustrating large spillway systems and controlling ice margins. Most ice margins from Bluemle
(1988).

flows (Josenhans and Zevenhuizen, 1990; Lajeunesse and St-Onge, 8. Floods into Lake Agassiz
2008). Veillette (1994) worked in the Lake Ojibway basin and along
with others, estimated final drainage at 8.0 14C kyr BP (∼8.86 cal kyr Catastrophic drainage events into Lake Agassiz from ice-marginal
BP) based on marine shells associated with the Tyrell Sea transgression lakes hundreds of kilometers to the west are recorded by large spillways
following final drainage of Lake Agassiz. The estimate for flood timing ending at shorelines, and by large fans deposited into the lake
from marine records by Barber et al. (1999) from carbonate marine (Figs. 1,9). Many of these events and characteristic erosional and de-
shells in the sedimentary record of the flood in Hudson Strait and positional features are reviewed in Kehew et al. (2009) with peak flood
Hudson Bay is 8.45 cal kyr BP, and the 8.2 ka event isotopic record from discharge estimated at ∼0.8 Sv (Table 3). The general sequence of
the Greenland Ice Sheet may be associated with final drainage (see events involves meltwater flowing along and away from ice margins
discussion in Thomas et al., 2007). within the Missouri drainage basin when ice was along the Missouri

11
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Table 3
Reconstructed flows from spillways associated with Lake Agassiz.
Spillway Paleohydrology Reference

Souris Spillway v = 2.9 – 11.7 ms , Q = .0.58 – 0.8 Sv

-1
Lord and Kehew (1987)
Southern Outlet v = 8.4 ms-1, 6.4 – 30.4 ms-1; Q = 0.1 – 0.36 Sv Fisher (2004)
Kelvin Outlets Q = ∼0.1 – 0.2 Sv Teller and Thorleifson (1983)
Kelvin Outlets 1900 – 8100 km3 released from 7–58 m drops in water level Teller and Leverington (2004)
Kelvin Outlets 0.012 – 0.024 Sv; 0.066 – 0.16 Sv Kelly et al. (2016)
CLAS Q = ∼2.16 Sv Fisher et al. (2002)
Wycherley channels Q = 0.08 – 0.29 Sv this paper
Kinojévis Outlet Q = 0.03 – 0.06 Sv Veillette (1994)
Subglacial Q = 0.5 – 5 Sv Leverington et al. (2002); Clarke et al. (2004)

Coteau and receding northwards. The spillways west of the James (Nielsen, 1988). Short strandline segments associated with the local
spillway record some of the earliest flows (Fig. 9) around 13,800 cal yr lakes have been mapped above the upper Campbell beach in this area
BP (Brophy and Bluemle, 1983). On Fig. 9 selected ice margins are (Nielsen, 1988). The highest Lake Agassiz beach north of Duck Moun-
shown as broad dashed white lines to illustrate how their position tain mapped in these reentrants is the Upper Campbell beach, in-
controlled formation of ice-dammed lakes, and with subsequent reces- dicating that when ice retreated from the Manitoba Escarpment lake
sion of the ice margin, how new drainage routes evolved. For example, level was below the Campbell beach during the Moorhead Phase. Fur-
when ice retreated northwards from the Heimdal ice margin, flow was ther north, two large deltas developed where the Saskatchewan River
initially through the James spillway until lower land was uncovered; entered proglacial Lake Saskatchewan (Fig. 7). The Fort Á La Corne
flow was then captured and diverted down the Sheyenne River. Both delta formed when Lake Saskatchewan drained southward through the
rivers end at Lake Dakota (Fig. 9). Brophy and Bluemle (1983) describe Assiniboine spillway system, after Lake Assiniboine (Wolfe and Teller,
initial flow in the Sheyenne River as typical braided river flow before 1995) had drained (Fig. 7). Set below and eastward of the Fort Á La
the spillway developed, evidenced by high elevation terraces of out- Corne delta is the Nipawin delta. It formed at a level lower than glacial
wash. With further ice margin recession water was trapped at the Lake Saskatchewan; Elson (1967) and Christiansen et al. (1995) as-
margin of the Souris Lobe in glacial Lake Souris, and to the east in Lake sociate the delta with Lake Agassiz.
Minneoaukan. While the triggering mechanisms for what initiated lake
drainage and spillway formation is uncertain, headwater migration of
9. Southern outlet flows
knickpoints to the lakes, melting of stagnant ice near the sill, and large
inputs of meltwater from upstream have all been proposed (cf. Kehew
The southern outlet of Lake Agassiz (Figs. 1,3) is the oldest and most
and Lord, 1986; and Lord and Kehew, 1987). The James and Sheyenne
studied outlet of the lake. At the Herman and Norcross level, flow was
spillways that enter Lake Dakota record high-magnitude flows before
through the Cottonwood and central spillway (Fig. 3) and the Cotton-
the Des Moines Lobe had retreated north of the Big Stone Moraine.
wood channel is abandoned once lake level dropped to the Tintah level.
Once Lake Agassiz began forming, sometime after the ice margin had
The central spillway complex at the Tintah and Upham levels was very
receded north of the present day Sheyenne River, a large flood eroded
broad, up to 10 km width. Discharge estimates through the southern
the Sheyenne spillway and deposited the large Sheyenne fan into the
outlet range from 0.4 to 0.7 Sv based on a variety of techniques in-
southern basin of the lake. Drainage from ephemeral Lake Souris is
cluding boulder transport (Fig. 10A); see Fisher (2004) for more in-
often invoked to have caused this event (Kehew and Lord, 1986). The
formation. As the lake area increased through time, baseline discharge
Sheyenne fan area is ∼2000 km2 and consists of a wedge of sediment
would have increased through the southern outlet. As discussed in
fining outwards from gravel and sand, to silt and clay (Brophy, 1967).
section 1.8, ice retreat allows successively northern spillways to form,
With continued ice recession northwards, Lake Souris transitions
delivering catastrophic floods into the lake along its western side. While
into glacial Lake Hind (Fig. 9). Detailed mapping and reconstruction of
discharges have been estimated between 0.058 to 0.8 Sv, water levels
the lakes paleogeography can be found in Sun and Teller (1997). Once
remained at the Herman lake level when the Sheyenne and Pembina
the Assiniboine Lobe had retreated far enough northwards, drainage
deltas were deposited. The reasoning for this is that the Herman Beach
from Lake Hind was to the southeast through the Pembina spillway. The
extends north from the southern outlet to the Pembina delta. If the
spillway morphology records large flows, although the Pembina Fan is
southern outlet was incised to a lower lake level when a delta formed
much smaller than the Sheyenne Fan. The width of the spillway is larger
from a catastrophic flood, then a lower lake level would appear north of
than the Sheyenne’s, suggesting that much of the eroded sediment en-
the delta. Elson (1967) noted that the highest Herman strandline ends
tering the lake was transported away by littoral processes. Further re-
near the Pembina delta, and strandlines are difficult to map across the
cession of the Assiniboine Lobe opened a short spillway (box 4, Fig. 9)
Assiniboine delta. The Norcross is the highest strandline north of the
and sediment was deposited in the Assiniboine reentrant. Continuing
delta. Thus, it is likely that some combination of increasing baseflow
ice margin retreat increased the Assiniboine River discharge, with flow
and catastrophic floods through the Pembina and Assiniboine spillways
across the Lake Hind basin and into glacial Lake Agassiz contributing to
initiated incision of the southern outlet, causing lake level to fall to the
growth of the Assiniboine delta.
Norcross level. The River Warren spillway continued to evolve across
The Assiniboine spillway extends much further north, along the
Minnesota (cf. Johnson et al., 1998), and now is occupied by the Little
western side of the hills that make up the Manitoba Escarpment (Fig. 7).
Minnesota River that eventually joins the Mississippi River (Fig. 1).
Many spillways come off these headlands, and are tributaries to the
Assiniboine River and spillway, indicating that lower elevation land to
the east and north was ice covered when the Assiniboine delta was 10. Eastern outlet flows
forming. The westward flowing spillways were abandoned once lobes of
ice retreated from the reentrants along the Manitoba Escarpment, Flow out the Kelvin outlets was estimated by Teller and Thorleifson
presumably in a south to north direction. Deltas from local, small ice- (1983) using channel geometry and boulder dimensions to be
marginal lakes are present within the Swan River reentrant and along 0.1–0.2 Sv over a few years. Kelly et al. (2016) calculated lower dis-
the eastern side of the Porcupine Hills and Duck Mountain (Fig. 7) charges for the Roaring River site, using conservative estimates of
channel width (Table 1) and boulder dimensions (Fig. 10B). A

12
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 10. Example of boulders associated with

spillways. A) Boulders in the southern outlet
spillway. The boulder closest to camera is 2.6 m
high. B) Approximately 1 m diameter boulder
in a Kelvin outlet, west of Lake Nipigon. C)
Transverse giant current ripple composed of
boulders up to 1.8 m diameter adjacent to
Mundell Lake. Author at 1.75 m height is cir-
cled for scale. Flow direction was from right to
left.

maximum discharge from the Mundell Lake site within the Pillar outlet drainages (cf. Martini et al., 2002; Burr et al., 2009) that are commonly
complex is 0.12–0.016 Sv. At this site large giant current ripples associated with deglaciation. Highly erosive flows result in large
(Fig. 10C) were constructed from boulders up to 1.8 m b-axis. Pre- channels or sets of channels eroded into bedrock (Fig. 11A) or glaci-
servation of these bedforms adjacent to the lake likely indicates a short- genic sediment (Fig. 11B). Sediment transport is high and flood deposits
lived megaflood event followed by a much lower discharge steady flow. may be found within the spillway where velocities decrease from flow
Past workers have discussed or implied such a flooding history con- separation or expansion. Flood deposits are found in the CLAS where
trolled by the receding ice margin opening progressively lower outlets flow occurred in upper scoured zones and at its northern end where
further to the north as the sub-continental drainage divide is uncovered flow slowed and bifurcated around the Fort Hills (Fig. 6) (Smith and
(Elson, 1967; Zoltai, 1967; Teller and Thorleifson, 1983; Leverington Fisher, 1993; Fisher et al., 1995) with examples described next. Often
and Teller, 2003; Kelly et al., 2016). Leverington and Teller (2003) very-large boulders are transported or left behind as lags (Fig. 11D),
estimated water level drops of 7 to 58 m and corresponding water vo- while large piles of usually subangular to subrounded boulders record
lume releases of 1900–8100 km3 as new outlets became established. bedload transport, but they are too large to crush for most gravel op-
With better topographic data and ice margin positions, more accurate erations (Fig. 11E). In areas of flow separation large sets of cross-beds
estimate of paleodischarge can be determined. develop (Fig. 11C) consisting of graded beds with open- and closed-
The various Nipigon outlets (Fig. 5) and on-shore fan sediments into work characteristics (Fig. 11H). Some clast-supported gravel is open-
the Lake Superior basin were described by Teller and Mahnic (1988). work (Fig. 11F) requiring that all other material remained in suspen-
Evidence for flooding include sandy turbidities up to 0.65 m thick and sion, and elsewhere gravel clasts are supported by a matrix resulting in
large trough crossbeds. Gary et al. (2012) used marine seismic methods a bimodal particle size distribution (Fig. 11G, I). Intraformational clasts
offshore of Nipigon, and described deeply incised bedrock channels in consisting of loosely consolidate oil shale bedrock, till and glaciola-
the bay overlain by coarser-grained hummocky material interpreted as custrine sediment range in size from pebbles and boulders (Fig. 11J) to
sediments deposited during high energy events, which are in turn large rafts 15 m along an exposed axis.
overlain by finer-grained sediments reflective of non-catastrophic The Wycherely channels (Fig. 6) near the headwater of the Churchill
events. River were the northwest outlet active at the Campbell level (Fisher and
Souch, 1998; Fisher, 2007). Routing from the Wycherely channels was
11. Northwest and Ojibway outlet flows into Lake Wagtufro and then through the CLAS. One channel was
briefly visited on foot in 2003, with access by float plane, but was not
The main spillway and flood deposits associated with the Clearwater surveyed in the field. Using topographic maps and aerial photographs
lower-Athabasca Spillway (CLAS) were described in detail by Fisher (Fig. 6) hydrological variables are estimated at ∼400 m width and
and Smith (1993, 1994), Smith and Fisher (1993), and Fisher et al. ∼15 m depth. Individual boulders up to a few meters diameter were
(1995). A peak discharge estimate of ∼2.4 Sv used the slope-area observed as were deposits of boulders of ∼1 m diameter. Assuming
equation with mean velocities of 13.2 ms-1. Assumptions of duration velocities of between 3 to 12 ms-1 (methodology in Fisher, 2004), each
assumed a draw down of 46 m and an areal extent of 350,000 km2 channel may have experienced peak discharges of 0.018–0.072 Sv. If all
(Teller and Clayton, 1983), but the area would be considerably less for channels were active at once discharge could have been 0.08 to 0.29 Sv.
the much smaller glacial Lake Churchill. Such a value is higher than Lake Agassiz baseline flow estimated by
Sediments deposited from megaflooding or catastrophic flows have Teller et al. (2002) at 0.034 Sv for the Campbell level.
been described from glacial Lake Agassiz and from many other lake Veillette (1994) estimated paleoflows for the Ojibway Phase at

13
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

Fig. 11. Examples of flood deposits from the Clearwater lower-Athabasca spillway. A) 5 m of imbricate and cross-bedded gravel overlying bedrock. Person is 1.6 m
tall. B) Plane-bedded sand overlying glacial lacustrine rhythmites. Scale in cm and inches. C) Cross-bedded gravel, person is 1.95 m tall. D) 1 to 2.4 m diameter
boulders, person is 1.6 m tall. E) Sub-rounded boulders, person is 1.6 m tall. F) Open-work gravel in center of image. G) Very-coarse sand to fine-gravel matrix
supported pebbles and cobbles. H) Planar tabular cross-beds consisting of pebbly and open-work cobbles-to-boulder beds. Spade is 25 cm wide. I) Cobbles supported
within a matrix of granules normally graded to coarse sand. Note black laminae of tar sand at top of image. J) Boulder-sized clast of diamicton within a boulder-
gravel deposit.

different river levels downstream of the Kinojévis outlet in the Ottawa 163,000 km3, with an initial drainage of 113,100 km3 followed by
River valley. Water depths were estimated from fluvial erosional marks 49,900 km3 from the Fidler level. Resulting discharges were about 5 Sv
(e.g., potholes) cut into bedrock and the Manning equation was used to if drainage was in one stage, or 3.6 Sv and then 1.6 Sv if a two-stage
estimate discharge. Results ranged from 33,000–65,000 m3s-1, which drainage. Clarke et al. (2004) estimate a half-year flood duration, with a
are more conservative than an earlier estimate of 0.2 Sv from Lewis and discharge of about 5 Sv, and a maximum sea level rise of 0.19 m. The
Anderson (1989). final drainage either in one or two parts is explained as the forcing
mechanism for the 8.2 event recorded in the Greenland ice Cores and
elsewhere (Barber et al., 1999).
12. Final drainage

There is some uncertainty about the final drainage of Lake Agassiz, 13. Summary
whether it was in two stages or one (Leverington et al., 2002; Teller
et al., 2002; Clarke et al., 2004). Clarke et al. (2004) advocate for The history of drainage from Lake Agassiz extends over a nearly
subglacial drainage that could accommodate an initial drop in water 6000 year period as the lake evolved across the center of the North
level from the Kinovéjis to Fidler levels and then final drainage. American continent. With water trapped at its southernmost extent at
Leverington et al. (2002) calculated a total drainage volume of the sub-continental drainage divide, the large southern outlet was

14
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

episodically incised as recorded by numerous beaches graded to it, and Assoc. Can. Sp. Pap, pp. 173–186.
boulder-armored terraces within it. Most recent discharge estimates for Burr, D.M., Carling, P.A., Baker, V.R., 2009. Megaflooding on Earth and Mars. Cambridge
University Press, New York, pp. 319.
the southern outlet are up to 0.36 Sv. Large floods from deglacial lakes Carlson, A.E., Clark, P.U., 2012. Ice-sheet sources of sea-level rise and fresh- water dis-
further west deposited large fans into Lake Agassiz with estimated charge during the last deglaciation. Revs. Geophys 50, RG4007.
discharges of up to 0.8 Sv. The receding ice margins controlled where Carlson, A.E., Clark, P.U., Haley, B.A., Klinkhammer, G.P., Simmons, K., Brook, E.J.,
Meissner, K.J., 2007. Geochemical proxies of North American freshwater routing
and when lower outlets to the east and north could open. Uncertainty during the Younger Dryas cold event. PNAS 104, 6556–6561.
remains where the lake drained the first time the southern outlet was Christiansen, E.A., Sauer, E.K., Schreiner, B.T., 1995. Glacial Lake Saskatchewan and Lake
abandoned at the start of the Moorhead Phase, which is approximately Agassiz deltas in east-central Saskatchewan with special emphasis on the Nipawin
delta. Can. J Ear. Sci. 32, 334–348.
coeval with the start of the YD. While the ice margin may have re- Clarke, G.K.C., Leverington, D.W., Teller, J.T., Dyke, A.S., 2004. Paleohydraulics of the
treated far enough east to permit drainage into the Great Lakes basin, last outburst flood from glacial Lake Agassiz and the 8200 BP cold event. Quat. Sci.
spillways with vertical relief to explain the ∼90 m drop in elevation Revs. 23, 389–407.
Condron, A., Winsor, P., 2012. Meltwater routing and the Younger Dryas. PNAS 109,
remain to be documented. Drainage out a northwest outlet is now
19928–19933.
viewed as the only other candidate with discharges estimated at ∼2 Sv. Cronin, T.M., Rayburn, J.A., Guilbault, J.-P., Thunell, R., Franzi, D.A., 2012. Stable iso-
However, only the much younger Campbell beaches have been mapped tope evidence for glacial lake drainage through the St. Lawrence Estuary, eastern
to a northwest outlet unless an earlier phase of Lake Agassiz extended to Canada, ∼13.1–12.9 ka. Quat. Int. 260, 55–65.
Dilworth, J., Fisher, T.G., 2018. Determining the Lake Agassiz Moorhead Phase lowstand
the head of the CLAS spillway and evidence for it has been removed elevation from compaction ridges and newly identified strandlines in the Red River
from a readvance. When flow returned to the southern outlet during the Valley, USA. Geomorph. 319, 216–225.
Emerson Phase there is evidence that a northwest, eastern (Kelvin) and Elson, J.A., 1967. Geology of Glacial Lake Agassiz. In: Mayer-Oakes, W.J. (Ed.), Life, Land
and Water. University of Manitoba Press, Winnipeg, pp. 37–96.
southern outlet were active simultaneously, or in close succession with Fisher, T.G., 2003. Chronology of glacial Lake Agassiz meltwater routed to the Gulf of
the lake at the Campbell levels. Drawdown of water during the Nipigon Mexico. Quat. Res. 59, 271–276.
Phase through successively lower Kelvin outlets resulted in paleo- Fisher, T.G., 2004. River Warren boulders: paleoflow indicators in the southern spillway
of glacial Lake Agassiz. Boreas 33, 349–358.
discharge estimates of 0.1 to 0.2 Sv that entered the Lake Superior Fisher, T.G., 2005. Strandline analysis in the southern basin of glacial Lake Agassiz,
basin. Lakes Agassiz and Ojibway merged with continued retreat of the Minnesota and North and South Dakota, USA. Geol. Soc. Am. Bull. 117, 1481–1496.
ice margin, and outflow through the Kinojévis outlet into the Ottawa Fisher, T.G., 2007. Abandonment chronology of glacial Lake Agassiz’s northwestern
outlet. Paleogeorg Palaeocl. 246, 31–44.
River system was estimated at 0.033–0.065 Sv. Final subglacial drai- Fisher, T.G., Smith, D.G., 1994. Glacial Lake Agassiz: its northwest maximum extent and
nage of the lake occurred as ice thinned over Hudson Bay with drainage outlet in Saskatchewan (Emerson phase). Quat. Sci. Rev. 13, 845–858.
in one or two stages. With discharge estimated at ∼5 Sv, sea level was Fisher, T.G., Souch, C., 1998. Northwest outlet channels of Lake Agassiz, isostatic tilting
and a migrating continental drainage divide, Saskatchewan, Canada. Geomorph. 25,
raised 0.19 m and this megaflood freshened the surface of the North
57–73.
Atlantic and is often cited as the cause of the 8.2 ka event. Fisher, T.G., Lowell, T.V., 2006. Questioning the age of the Moorhead Phase in the glacial
Lake Agassiz basin. Quat. Sci. Rev. 25, 2688–2691.
14. Declaration of Competing Interest Fisher, T.G., Lowell, T.V., 2012. Testing northwest drainage from Lake Agassiz using
extant ice margin and strandline data. Quat. Int. 260, 106–114.
Fisher, T.G., Jol, H.M., Smith, D.G., 1995. Ground-penetrating radar used to assess ag-
There is no conflict of interest with this manuscript. gregate in catastrophic flood deposits, northeast Alberta, Canada. Can. Geotech. J.
32, 871–879.
Fisher, T.G., Smith, D.G., Andrews, J.T., 2002. Preboreal oscillation caused by a glacial
Acknowledgements Lake Agassiz flood. Quat. Sci. Rev. 21, 873–878.
Fisher, T.G., Waterson, N., Lowell, T.V., Hajdas, I., 2009. Deglaciation ages and meltwater
SRTM data for Figs. 5 and 9 provided by OpenTopography.com. routing in the Fort McMurray region, northeastern Alberta and northwestern
Saskatchewan. Canada. Quat. Sci. Revs. 28, 1608–1624.
Thomas Valachovics, Brian Samsen, Tom Lowell, and two journal re- Fisher, T.G., Lepper, K., Ashworth, A.C., Hobbs, H.C., 2011. Southern outlet and basin of
viewers are thanked for their constructive comments. glacial Lake Agassiz. In: Miller, J.D., Hudak, G.J., Wittkop, C., McLaughlin, P.I.
(Eds.), Archean to Anthropocene: Field Guides to the Geology of the Mid-Continent of
North America. Geological Society of America, Boulder, CO, pp. 379–400.
References
Fisher, T.G., Yansa, C.H., Lowell, T.V., Lepper, K., Hajdas, I., Ashworth, A.C., 2008. The
chronology, climate, and confusion of the Moorhead Phase of Glacial Lake Agassiz:
Aitken, M.J., 1998. An Introduction to Optical Dating: The dating of Quaternary sedi- new results from the Ojata Beach, North Dakota, U.S.A. Quat. Sci. Rev. 27,
ments by the use of photon-stimulated luminescence. Oxford University Press, New 1124–1135.
York. Flower, B.P., Hastings, D.W., Hill, H.W., Quinn, T.M., 2004. Phasing of deglacial warming
Arndt, B.M., 1977. Stratigraphy of offshore sediment of Lake Agassiz. North Dakota Geol. and Laurentide Ice Sheet meltwater in the Gulf of Mexico. Geology 32, 597–600.
Surv. Report of Invest. 60, 1–58. Gary, J.L., Colman, S.M., Wattrus, N., Lewis, C.F.M., 2012. Post-Marquette discharge from
Bajc, A.F., Schwert, D.P., Warner, B.G., Williams, N.E., 2000. A reconstruction of Glacial Lake Agassiz into the Superior basin. J. Paleolim. 47, 299–311.
Moorhead and Emerson phase environments along the eastern margin of glacial Lake Hald, M., Hagen, S., 1998. Early Preboreal cooling in the Nordic seas region triggered by
Agassiz, Rainy River basin, northwestern Ontario. Can. J. Ear. Sci. 37, 1335–1353. meltwater. Geology 26, 615–618.
Barber, D.C., Dyke, A., Hillaire-Marcel, C., Jennings, A.E., Andrews, J.T., Kerwin, M.W., Hillaire-Marcel, C., Maccali, J., Not, C., Poirier, A., 2013. Geochemical and isotopic tra-
Bilodeau, G., NcNeely, R., Southons, J., Morehead, M.D., Gagnon, J.M., 1999. Forcing cers of Arctic sea ice sources and export with special attention to the Younger Dryas
of the cold event of 8,200 years ago by catastrophic drainage of Laurentide lakes. interval. Quat. Sci. Rev. 79, 184–190.
Nature. 400, 344–348. Jennings, A.E., Hald, M., Smith, M., Andrews, J.T., 2006. Freshwater forcing from the
Bjorck, B., Keister, C.M., 1983. The Emerson Phase of Lake Agassiz, independently re- Greenland Ice Sheet during the Younger Dryas: evidence from southeastern
gistered in northwestern Minnesota and northwestern Ontario. Canadian Journal of Greenland shelf cores. Quat. Sci. Rev. 25, 287–298.
Earth Science 20, 1536–1542. Johnson, M.D., Davis, D.M., Pederson, J.L., 1998. Terraces of the Minnesota River Valley
Bluemle, J.P., 1988. Guide to the geology of north-central North Dakota. North Dakota and the character of Glacial River Warren downcutting. Contributions to Quaternary
Geological Survey, Education Series. 19. studies in Minnesota: Minnesota Geol. Surv. Rep. Investig 49, 121–130.
Breckenridge, A., 2015. The Tintah-Campbell gap and implications for glacial Lake Johnston, W.A., 1946. Glacial Lake Agassiz, with special reference to the mode of de-
Agassiz drainage during the Younger Dryas cold interval. Quat. Sci. Revs. 117, formation of the beaches. Geological Survey of Canada Bulletin 7, 20.
124–134. Josenhans, H.W., Zevenhuizen, J., 1990. Dynamics of the Laurentide Ice Sheet in Hudson
Breckenridge, A., Lowell, T.C., Evans, G., 2012. A review and analysis of varve thickness Bay. Canada. Mar. Geol. 92, 1–26.
records from glacial Lake Ojibway (Ontario and Quebec, Canada). Quat. Int. 260, Kehew, A.E., Lord, M.L., 1986. Origin and large-scale erosional features of glacial-lake
43–54. spillways in the northern Great Plains. Geol. Soc. Am. Bull. 97, 162–177.
Broecker, W.S., Kennett, J., Flower, B., Teller, J., Trumbore, S., Bonani, G., Wolfli, W., Kehew, A.E., Lord, M., Kozlowski, A.L., Fisher, T.G., 2009. Proglacial megaflooding along
1989. Routing of meltwater from the Laurentide Ice Sheet during the Younger Dryas the margins of the Laurentide Ice Sheet. Chapter 7. In: Burr, D., Carling, P.A., Baker,
cold episode. Nature 341, 318–321. V.R. (Eds.), Megaflooding on Earth and Mars. Cambridge University Press, New York,
Brophy, J.A., 1967. Some aspects of the geological deposits of the south end of the Lake pp. 104–127.
Agassiz Basin. In: Mayer-Oakes, W.J. (Ed.), Life, Land and Water. University of Keigwin, L.D., Klotski, S., Zhao, N., Reilly, B., Giosan, L., Driscoll, N.W., 2018. Deglacial
Manitoba, Winnipeg, pp. 97–105. floods in the Beaufort Sea preceded Younger Dryas cooling. Nat. Geo. 11, 599–604.
Brophy, J.A., Bluemle, J.P., 1983. The Sheyenne River; Its geological history and effects Kelly, M.A., Fisher, T.G., Lowell, T.V., Barnett, P.J., Schwartz, R., 2016. 10Be ages of
on Lake Agassiz. In: In: Teller, J.T., Clayton, L. (Eds.), Glacial Lake Agassiz. 26. Geol. flood deposits west of Lake Nipigon, Ontario: evidence for eastward meltwater

15
T.G. Fisher Earth-Science Reviews 201 (2020) 102974

drainage during the early Holocene Epoch. Can. J. Ear. Sci. 53, 321–330. Younger Dryas. Quat. Sci. Rev. 13, 923–944.
Lajeunesse, P., St-Onge, G., 2008. The subglacial origin of the Lake Agassiz–Ojibway final Roy, M., Veillette, J.J., Daubois, V., Menard, M., 2015. Late-stage phases of glacial Lake
outburst flood. Nature Geoscience 1, 184–188. Ojibway in the central Abitibi region, eastern Canada. Geomorphology 248, 14–23.
Lemmen, D.S., Duk-Rodkin, A., Bednarski, J.M., 1994. Late glacial drainage systems along Roy, M., Dell’Oste, F., Veillette, J.J., de Vernal, A., Heilie, J.-F., Parent, M., 2011. Insights
the northwestern margin of the Laurentide Ice Sheet. Quat. Sci. Rev. 13, 805–828. on the events surrounding the final drainage of Lake Ojibway based on James Bay
Lepper, K., Fisher, T.G., Hajdas, I., Lowell, T.V., 2007. Ages for the Big Stone moraine and stratigraphic sequences. Quat. Sci. Rev. 30, 682–692.
the oldest beaches of glacial Lake Agassiz: Implications for deglaciation chronology. Schreiner, B.T., 1983. Lake Agassiz in Saskatchewan. In: In: Teller, J.T., Clayton, L. (Eds.),
Geology 35, 667–670. Glacial Lake Agassiz 26. Geol. Assoc. Can. Sp. Pap, St. Johns Newfoundland, pp.
Lepper, K., Gorz, K.L., Fisher, T.G., Lowell, T.V., 2011. Age determinations for Lake 75–96.
Agassiz shorelines west of Fargo, North Dakota, U.S.A. Can. J. Ear. Sci. 48, Skinner, R.G., 1973. Quaternary geology of the Moose River basin, Ontario. Geol. Surv.
1199–1207. Can. Bull. 225, 77.
Lepper, K., Buell, A.W., Fisher, T.G., Lowell, T.V., 2013. A chronology for glacial Lake Smith, D.G., 1992. Glacial Lake Mackenzie, Mackenzie Valley, Northwest Territories,
Agassiz shorelines along Upham’s namesake transect. Quat. Res. 80, 88–98. Canada. Can. J Ear. Sci. 29, 1756–1766.
Levac, E., Lewis, M., Stretch, V., Duchesne, K., Neulieg, T., 2015. Evidence for meltwater Smith, D.G., Fisher, T.G., 1993. Glacial Lake Agassiz: The northwestern outlet and pa-
drainage via the St. Lawrence River Valley in marine cores from the Laurentian leoflood. Geology 21, 9–12.
Channel at the time of the Younger Dryas. Global Planet. Change 130, 47–65. Stroup, J.S., Lowell, T.C., Breckenridge, A., 2013. Stratigraphy from the demise of a large
Leverett, F., 1932. Quaternary geology of Minnesota and parts of adjacent states. U.S. glacial lake: the Lake Ojibway, Ontario, Quebec example. J. Paleolim 50, 105–121.
Geol. Surv. Prof. Pap. 161, 149. Sun, C.S., Teller, J.T., 1997. Reconstruction of glacial Lake Hind in southwestern
Leverington, D.W., Teller, J.T., 2003. Paleotopographic reconstructions of the eastern Manitoba, Canada. J. Paleolim. 17, 9–21.
outlets of glacial Lake Agassiz. Can. J. Ear. Sci. 40, 1259–1278. Tarasov, L., Peltier, W.R., 2005. Arctic freshwater forcing of the Younger Dryas cold re-
Leverington, D.W., Mann, J.D., Teller, J.T., 2000. Changes in the bathymetry and volume versal. Nature 435, 662–665.
of glacial Lake Agassiz between 11,000 and 9300 14C yr B.P. Quat. Res. 54, 174–181. Tarasov, L., Peltier, W.R., 2006. A calibrated deglacial drainage chronology for the North
Leverington, D., Mann, J.D., Teller, J.T., 2002. Changes in the Bathymetry and Volume of American continent: evidence of an Arctic trigger for the Younger Dryas. Quat. Sci.
Glacial Lake Agassiz Between 11,000 and 9300 14C yr B.P. Quat. Res. 54, 174–181. Rev. 25, 659–688.
Lewis, C.F.M., Anderson, T.W., 1989. Oscillations of levels and cool phases of the Teller, J.T., 2001. Formation of large beaches in an area of rapid differential isostatic
Laurentian Great Lakes caused by inflow from Glacial Lakes Agassiz and Barlow- rebound: the three-outlet control of Lake Agassiz. Quat. Sci. Rev. 20, 1649–1659.
Ojibway. J. Paleolim. 2, 99–146. Teller, J.T., 2013. Lake Agassiz during the Younger Dryas. Quat. Res. 80, 361–369.
Leydet, D.J., Carlson, A.E., Teller, J.T., Breckenridge, A., Barth, A., Ullman, D., Sinclair, Teller, J.T., Thorleifson, L.H., 1983. The Lake Agassiz-Lake Superior connection. In: In:
G., Milne, G., Cuzzone, J., Caffee, M., 2018. Opening of glacial Lake Agassiz’s eastern Teller, J.T., Clayton, L. (Eds.), Glacial Lake Agassiz 26. Geol. Assoc. Can. Sp. Pap, St.
outlets by the start of the Younger Dryas cold period. Geology 46, 155–158. Johns Newfoundland, pp. 261–290.
Liu, X., Fisher, T.G., Lepper, K., Lowell, T.V., 2014. Geochemical characteristics of glacial Teller, J.T., Mahnic, P., 1988. History of sedimentation in the northwestern Lake Superior
Lake Agassiz sediments and new ages for the Moorhead Phase at Fargo, North Dakota, basin and its relation to Lake Agassiz overflow. Can. J. Ear. Sci. 25, 1660–1673.
USA. Can. J. Ear. Sci. 51, 850–861. Teller, J.T., Leverington, D.W., 2004. Glacial Lake Agassiz: A 5000 yr history of change
Lord, M.L., Kehew, A.E., 1987. Sedimentology and paleohydrology of glacial-lake out- and its relationship to the ∂18O record of Greenland. Geol. Sco. Am. Bull 116,
burst deposits in southeastern Saskatchewan and northwestern North Dakota. Geol. 729–742.
Soc. Am. Bull. 99, 663–673. Teller, J.T., Leverington, D.W., Mann, J.D., 2002. Freshwater outbursts to the oceans from
Lowell, T.V., Larson, G.J., Hughes, J.D., Denton, G.H., 1999. Age verification of the Lake glacial Lake Agassiz and their role in climate change during the last deglaciation.
Gribben forest bed and the Younger Dryas advance of the Laurentide Ice Sheet. Can. Quat. Sci. Rev. 21, 879–888.
J. Earth Sci. 36, 383–393. Teller, J.T., Risberg, J., Matile, G., Zoltai, S., 2000. Postglacial history and paleoecology
Lowell, T.V., Fisher, T.G., Hajdas, I., Glover, K., Loope, H.M., Henry, T., 2009. of Wampum, Manitoba, a former lagoon in the Lake Agassiz basin. Geol. Soc. Am.
Radiocarbon deglaciation chronology of the Thunder Bay, Ontario area and im- Bull. 112, 943–958.
plications for ice sheet retreat patterns. Quat. Sci. Rev. 28, 1597–1607. Teller, J.T., Boyd, M., Yang, Z., Kor, P.S.G., Mokhtari Fard, A., 2005. Alternative routing
Lowell, T.V., Fisher, T.G., Comer, G.C., Hajdas, I., Waterson, N., Glover, K., Loope, H.M., of Lake Agassiz overflow during the Younger Dryas: new dates, paleotopography, and
Schaefer, J.M., Rinterknecht, V., Broecker, W.S., Denton, G.H., Teller, J.T., 2005. a reevaluation. Quat. Sci. Rev. 24, 1890–1905.
Testing the Lake Agassiz Meltwater Trigger for the Younger Dryas. EOS 86 (365) 372. Teller, J.T., McGinn, R.A., Rajapara, H.M., Shukla, A.D., Singhvi, A.K., 2018. Optically
Lowell, T.V., Applegate, P.J., Fisher, T.G., Lepper, K., 2013. What caused the low-water stimulated luminescence ages from the Lake Agassiz basin in Manitoba. Quat. Res.
phase of glacial Lake Agassiz? Quat. Res. 80, 370–382. 89, 478–493.
Martini, I.P., Baker, V.R., Garzon, G., 2002. Flood and megaflood processes and deposits: Thomas, E.R., Wolff, E.W., Mulvaneya, R., Steffensen, J.P., Johnsen, S.J., Arrowsmith, C.,
recent and ancient examples. International Association of Sedimentologists Special White, J.C.W., Vaughn, B., Popp, T., 2007. The 8.2 ka event from Greenland ice cores.
Publication, pp. 32. Quat. Sci. Rev. 26, 70–81.
Matsch, C.L., 1983. River Warren, the southern outlet to glacial Lake Agassiz. In: In: Thompson, T.A., Lepper, K., Enders, A., Johnson, J., Baedke, S.J., Argyilan, E.P., Booth,
Teller, J.T., Clayton, L. (Eds.), Glacial Lake Agassiz 26. Geol. Assoc. Can. Sp. Pap., St. R.K., Wilcox, D.A., 2011. Mid-Holocene lake level and shoreline behavior during the
Johns Newfoundland, pp. 231–244. Nipissing phase of the upper Great Lakes at Alpena, Michigan, USA. J. Great Lakes
McMillan, K., Teller, J.T., 2012. Origin of the Herman–Norcross–Tintah sequence of Lake Res. 37, 567–576.
Agassiz beaches in Manitoba, Canada. Geomorphology 151-152, 77–88. Upham, W., 1895. The Glacial Lake Agassiz. U.S. Geol. Surv. Monograph 25, 685.
McNeely, R., Nielsen, E., 2000. Manitoba radiocarbon dates: Geologial radiocarbon dates Veillette, J.J., 1994. Evolution and paleohydrology of Glacial Lakes Barlow and Ojibway.
(Section 1). Manitoba Industry, Trade and Mines Geological Survey Open File Report Quat. Sci. Rev. 13 945–944.
OF2000-1. Vincent, J.-S., Hardy, L., 1979. The evolution of glacial Lakes Barlow and Ojibway,
Murton, J.B., Bateman, M.D., Dallimore, S.R., Teller, J.T., Yang, Z., 2010. Identification of Quebec and Ontario. Geol. Surv. Canada Bull. 316, 18.
Younger Dryas outburst flood path from Lake Agassiz to the Arctic Ocean. Nature Voytek, E.B., Colman, S.M., Wattrus, N.J., Gary, J.L., Lewis, C.F.M., 2012. Thunder Bay,
464, 740–743. Ontario, was not a pathway for catastrophic floods from Glacial Lake Agassiz. Quat.
Muschitello, F., Lea, J.M., Greenwood, S.L., Nick, F.M., Brunnberg, L., Macleod, A., Int. 260, 98–105.
Wohlfarth, B., 2016. Timing of the first drainage of the Baltic Ice Lake synchronous Wickert, A.D., Mitrovica, J., Williams, C., Anderson, R.S., 2013. Gradual demise of a thin
with the onset of Greenland Stadial 1. Boreas 45, 322–334. southern Laurentide ice sheet recorded by Mississippi drainage. Nature 502,
Nielsen, E., 1988. Surficial Geology of the Swan River area. Geological Report GR80-7. 668–671.
Manitoba Energy and Mines. 51 pp. Williams, C., Flower, B.P., Hastings, D.W., Guilderson, T.P., Quinn, K.A., Goddard, E.A.,
Nielsen, E., Gryba, E.M., Wilson, M.C., 1984. Bison remains from a Lake Agassiz spit 2010. Deglacial abrupt climate change in the Atlantic Warm Pool: A Gulf of Mexico
complex in the Swan River valley, Manitoba: depositional environment and paleoe- perspective. Paleoceanography 25.
cological implications. Can. J. Earth Sci. 21, 829–842. Wolfe, B., Teller, James T., 1995. Sedimentation in ice-dammed glacial Lake Assiniboine,
Rayburn, J.A., Teller, J.T., 2007. Isostatic rebound in the northwestern part of the Lake Saskatchewan, and catastrophic drainage down the Assiniboine Valley. Geogr. Phys.
Agassiz basin: Isobase changes and overflow. Paleogeorg Palaeocl 246, 23–30. Quatern. 49, 251–263.
Rayburn, J.A., Cronin, T.M., Franzi, D.A., Knuepfer, P.L.K., Willard, D.A., 2011a. Timing Yang, Z., Teller, J.T., 2012. Using LiDAR Digital Elevation Model data to map Lake
and duration of North American glacial lake discharges and the Younger Dryas cli- Agassiz beaches, measure their isostatically-induced gradients, and estimate their
mate reversal. Quat. Res. 75, 541–551. ages. Quat. Int. J. Paleolim 32–42.
Rayburn, J.A., Cronin, T.M., Franzi, D.A., Knuepfer, L.K., Willard, D.A., 2011b. Yansa, C.H., Ashworth, A.C., 2005. Late Pleistocene palaeoenvironment of the southern
Corrigendum to "Timing and duration of North American glacial lake discharges and Lake Agassiz Basin, USA. J. Quat. Sci. 20, 255–267.
the Younger Dryas climate reversal". Quat. Res. 77, 214. Zoltai, S.C., 1967. Eastern outlets of Lake Agassiz. In: Mayer-Oakes, W. (Ed.), Life, Land
Rodrigues, C.G., Vilks, G., 1994. The impact of glacial lake runoff on the Goldthwaite and and Water. University of Manitoba Press, Winnipeg, pp. 107–120.
Champlain Seas: The relationship between glacial Lake Agassiz runoff and the

16