Eos, Vol. 86, No. 40, 4 October 2005

EOS, TRANSACTIONS, AMERICAN GEOPHYSICAL UNION

VOLUME 86 NUMBER 40

4 OCTOBER 2005

PAGES 365-373

Testing the Lake Agassiz Meltwater Trigger for the Younger Dryas PAGES 3 6 5 , 3 7 2

Meltwater drainage from glacial Lake Agas­siz has been implicated for nearly 15 years as a trigger for thermohaline circulation changes producing the abrupt cold period known as the Younger Dryas. On the basis of initial field reconnaissance to the lake's proposed outlets, regional geomorphic mapping, and prelimi­nary chronological data, an alternative hypoth­esis may be warranted.

Should ongoing data collection continue to support preliminary results, it could be concluded that Lake Agassiz did not flood catastrophically into the Lake Superior basin preceding the Younger Dryas (Figure 1). All preliminary findings imply a retreating ice sheet margin approximately 1000 years young­er than previously thought, which would have blocked key meltwater corridors at the start of the Younger Dryas.

If Lake Agassiz meltwater passing into the North Atlantic is not the trigger for the Younger Dryas, then perhaps there were different sources of water or triggers. At this point, it seems prudent to carefully examine the role of glacial Lake Agassiz in any abrupt climate change scenario.

The current paradigm for driving abrupt climate change is the modification of thermo­haline circulation by the addition of external freshwater to the North Atlantic Ocean. Nu­merous modeling experiments have demon­strated the extreme sensitivity of this system, and attributing the source of that freshwater to glacial Lake Agassiz has evolved with numer­ous investigations.

In the mid-1970s, Kennett and Shackleton [1975] noted that the isotopic composition of seawater in the Gulf of Mexico fluctuated sub­stantially during deglacial time, and they at­tributed the fluctuation to changing sources of meltwater from the Laurentide Ice Sheet. Ap­proximately coeval with the isotope changes,

B Y T. V LOWELL, T. G. FISHER, G. C. COMER,

I. HAJDAS, N. WATERSON, K. GLOVER, H. M . LOOPE,

J . M . SCHAEFER,V RlNTERKNECHT, W BROECKER,

G . DENTON, AND J.T.TELLER

the Laurentide Ice Sheet retreated northward into an isostatically depressed basin behind the subcontinental drainage divide. Researchers in the Great Lakes reconstructed lake-level history, and they recognized variations in melt­water routing either through the Mississippi River to the Gulf of Mexico or through the St. Lawrence River to the North Atlantic (Figure 1).

By the late 1980s ,Broecker et at. [1989] had honed the concept of a freshwater trigger upsetting thermohaline circulation, and mod­eling experiments defined the necessary melt­water fluxes.The climate connection to glacial Lake Agassiz arose because organic sediments (10 ,960-9900 radiocarbon years [ 1 4 C ] B.P) de­posited between two sequences of deepwater clays would require a major drop in lake level,

i.e. a meltwater releasing event coeval with the Younger Dryas.

Subsequently Teller and colleagues [e.g., Teller and Leuerington, 2004] employed rebound models (delayed glacio-isostatic up­lift of the Earth's crust from ice-sheet loading following deglaciation),lake-level histories, and ice-retreat patterns to calculate meltwater volumes reaching the North Atlantic via an eastern route.These calculations were compat­ible with modeling estimates needed to affect ocean circulation.Thus, a terrestrial meltwater drainage reconstruction for triggering the Younger Dryas existed that was compatible with ocean records.

Two Preliminary Trips

An initial aerial survey on 4 - 7 May 2003 of areas north and west of Thunder Bay Ontario, referred to as the eastern outlets (Figure 2 ) , was followed by a second survey on 20-22 September 2003 north and east of Fort McMur-ray Alberta, along the Clearwater River.This

Fig. 1. Digital elevation model of central North America showing the southern, eastern, and north­western outlets of Lake Agassiz. Original color image appears at the back of this volume.

Eos, Vol. 86, No. 40, 4 October 2005

Fig. 2. Core sites and preliminary radiocarbon and beryllium ages from the Thunder Bay region. Radiocarbon ages indicate a sequential northeast retreat of the ice sheet, while beryllium ages show greater variability along one ice margin. The blue arrow west of Thunder Bay is the hypoth­esized eastern outlet meltwater route. Original color image appears at the back of this volume.

area is referred to as the northwestern outlet (Figure 3 ) . Boulders from moraine crests and flood channels were collected in each area for cosmic ray exposure dating.The results of these surveys, and dates obtained, have directed further ongoing research activities.

In the Thunder Bay region, channels and dry waterfalls in bedrock west and south (Ouimet Canyon) of Lake Nipigon (Figure 2) were examined, and they were considered by all researchers as unconnected to discharge coeval with the onset of the Younger Dryas.

Possible older outlets had been proposed west, rather than north, of Thunder Bay, but neither maps nor aerial survey revealed any large continuous channels, dry waterfalls, or spillways cut into bedrock, as is the case west of Lake Nipigon or in the southern outlet of Lake Agassiz.This was surprising, given the proposed catastrophic nature of the flood nec­essary to trigger a climate change. Alternatively the resistant bedrock of the Canadian Shield prevented formation of a well-developed spill­way resulting in non-catastrophic flow instead.

In contrast to the uncertain flood routing west of Thunder Bay, the Clearwater spillway (Figure 3) served as a route for meltwater flows.The geomorphic evidence is stunning: a wide, linear channel with numerous feeder channels at its eastern end, and a large delta at its downstream end. Catastrophic flood deposits contain wood giving a maximum age of 9860±230 1 4 C years B. P [Fisher et ai, 2002] . Strandlines (water-plane indicators such as beaches, spits, or escarpments) near the head of this system are discontinuous and are cov­ered by boreal forest, with the only known Agassiz strandline projecting to the base of the spillway Evidence that Lake Agassiz existed at the head of the spillway is based on the distribu­tion of scattered high-elevation strandlines, la­custrine sediment, and radiocarbon age-dated flood gravels.

Given the importance of unraveling Lake Agassiz's drainage history as a trigger for abrupt climate change, and given that the nature of the field evidence for Younger Dryas-aged drainage at either the eastern or north­western outlets is ambiguous, it was decided to construct and apply a chronological test. Did the ice sheet margin, either at Thunder Bay or Fort McMurray, withdraw enough to allow passage of meltwater at the start of the Younger Dryas? An affirmative response al­lows, but does not prove, an Agassiz meltwater trigger; a negative response rules out that trigger, forcing other explanations for the cause of the Younger Dryas.

Research Design

Conducting this test involves reconstructing the pattern of ice sheet retreat and dating significant retreat positions.The first task em­ploys topographic digital elevation models from the NASA Shuttle Radar Topography Mission (SRTM) to identify ice margins and regional deglaciation patterns.

In the Thunder Bay area (Figure 2 ) , this ap­proach shows several subparallel ice margins from just north of the Mesabi Range of north-

em Minnesota northward to Lake Nipigon, reaf­firming mapping by previous investigators. No single meltwater choke point was found here; instead, the topography allows several scenari­os for meltwater drainage depending upon the rebound history. For example, the lowest pres­ent topography is through Shebandowan Lake and the Kaministikwia Rivers, and in this area the ice margin slowly retreated.

In the Fort McMurray area (Figure 3) , the ice sheet had a lobate geometry between bedrock uplands and filled the Athabasca River valley. The key plug holding in water north of Fort McMurray was the Firebag Moraine (Figure 3).Several large abandoned channels cut this moraine, and the age of both the channels and moraine constrains breaching of the moraine dam.

The second task is to assign ages to these various ice positions. We take 11,000 1 4 C years B.P (R. Alley, personal communication, 2005) from the Greenland Ice Sheet Project 2 (GISP2) ice core record as the start of the Younger Dryas.

Radiocarbon plateaus and resolution of the accelerator mass spectrometer (AMS) radiocarbon dating technique prohibit match­ing the resolution of the ice cores. Rather, samples from the first organic material that accumulated on the deglaciated landscape are obtained, to try to bracket the age of lakes on either side of moraines. Because the oldest organic material from each site may not have

been recovered, or does not coincide with ice recession from that site, we employ a brute force approach by sampling as many lakes as possible.To date, we have recovered samples from 84 sites. Keeping uncertainties in dating as low as possible has been resource-intensive but critical to this test.

Preliminary Findings

In the Fort McMurray area, three well-devel­oped ice margins (Stony Mountain, Firebag, and Cree Lake Moraines) that spread over some 100 km are currently assigned radiocar­bon ages of 10,030,9595, and 9665 1 4 C years B.P, respectively (Figure 3).Consequently, it appears that the plug was pulled and melt­water could have flowed through the Clearwa­ter channel between 10,030 and 9700 1 4 C years B.P, consistent with the maximum age of 9900 1 4 C years B.P for flood gravel north of Fort McMurray (Figure 3 ) .

Even with the earliest bracket imposed by the radiocarbon ages, deglaciation and melt­water routing may have been 1000 1 4 C years lat­er than the Younger Dryas.This routing is more closely associated with the Preboreal Oscilla­t i o n ^ brief return to cooling after the Younger Dryas [Fisheretal, 2002],but precise temporal relationships have yet to be determined.

In the Thunder Bay area, the current data indicate sequential deglaciation with the She­bandowan lowland submerged by a glacial

Eos, Vol. 86, No. 40, 4 October 2005

Depth (m)

3 1

0 40 Organic %

Sample Sites A Core Sites Mar 05

# Replicate Sites Mar 05

• Cores Sites Jan 04

If Exposure Site Sep 03

A Fisher Prior Work

0403 Crescent Lake l i thostratigraphic units

— peaty gytt ja marl

•.".*•« sand

AMS 1 4 C ages 9,335±70 ETH-28527 [9,090±70 ETH-29208]

^[9,180±70 ETH-29210]

9,295±70 ETH-28528 [9,290±70 ETH-29209] [9,660±40 Beta-194056]

\ 9,595±75 40 ETH-30174

Carbonate %

1 4 C ages in brackets are same sample dated tw ice

• Dyke et al. (2003)

^ Moraine

Fig. 3. Fort McMurray area with an example of core stratigraphy from Crescent Lake on the proximal side of the Firebag Moraine. The lowest radio­carbon age of9595±75 is taken as a minimum age estimate for this moraine. The geometry and position of the ice sheet at the Firebag Moraine would have prevented any meltwater draining northward at that time. Original color image appears at the back of this volume.

lake until 10,200 1 4 C years B.P, based on a combination of radiocarbon ages and varve counts on the inorganic, laminated lake clays below the radiocarbon sample level (Figure 2) .

One interpretation is that sequential recession of the ice margin did not open an eastern out­let until well after the beginning of the Young­er Dryas. An alternative interpretation has been that recession from a glacial re-advance associated with the Younger Dryas cooling is being dated, not the original deglaciation.

At present, the data do not differentiate be­tween a re-advance or not. However, the sample sites, some with glaciolacustrine sediments, plot above known elevations of Lake Agassiz and the subcontinental drainage divide.This implies a different rebound history or deglaciation pattern than has been proposed.

A central point for reconstructing the drain­age of Lake Agassiz is the significance of the terrestrial macrofossils in fluvial sediment that indicate subaerial exposure between two formations of lake clay deep in the Agas­siz basin.The oldest age on these deposits is 10,960±300 1 4C years B.P, indicating a lower water level at that time. However, reexamina­tion of this date within the context of other numerous ages for this low lake level may indicate the wood was reworked into younger sediment, which would negate the temporal coincidence of the lowering lake level and the start of the Younger Dryas.

Thus, some key questions are: When and from what water plane did the drop occur? How fast was the drop? Where was the ice margin relative to the outlets? What was the basin volume at that time? All of these factors constrain any estimate of freshwater flux from Lake Agassiz into the oceans. Finally, if both outlets were blocked at this time, how extensive was the lake, and where was water draining at that time? Obviously, an under­standing of the chronology of Lake Agassiz is incomplete at this time.

Implications

Preliminary results indicate that ice reces­sion at both outlet areas is later than sup­posed, and that large volumes of meltwater were not catastrophically released from Lake Agassiz at the beginning of the Younger Dryas. Thus, if the Lake Agassiz floods did not upset the circulation pattern, the question becomes: What did? Could other pathways of the hydro-logical cycle alter the thermohaline circula­tion pattern at the beginning of the Younger Dryas, or alter other climate fluctuations that preceded Lake Agassiz?

These investigations indicate that the geo­logical understanding of past abrupt climate changes is only preliminary This does not bode well for predicting future, abrupt climate changes.

References Broecker,W S., et al. (1989) , Routing of meltwater

from the Laurentide Ice Sheet during the Younger Dryas cold episode,Nature, 347,318-321.

Dyke, A. S., et al. (2003), Deglaciation of North America, Geol. Sum. Can. Open File Rep., 1574, CD-ROM.

Fisher,T.G., D.G.Smith,and J.T.Andrews ( 2 0 0 2 ) , Pre-boreal Oscillation caused by a glacial Lake Agassiz f l o o d , Q u a t . S c i . R e a , 2 7 , 8 7 3 - 8 7 8 .

Kennet t ,J .P,and N.Shackleton (1975) , Laurentide ice sheet meltwater recorded in Gulf of Mexico deep-sea cores, Science, 188, 147-150.

Teller, J.T., and D.WLeverington (2004) , Glacial Lake Agassiz: A 5000 yr history of change and its rela­tionship to the 3 1 8 0 record of Greenland, Geol. Soc. Am. Bull, 116, 729 -742 .

Author Information

Thomas Lowell and Nicholas Waterson, University of Cincinnati,Ohio;Timothy Fisher and Henry Loope, University of Toledo, Ohio; Katherine Glover, University of North Carolina at Charlotte; Gary Comer, GCI, Waukesha,Wisconsin; Irka Hajdas, ETH-H6nggerberg, Zurich, Switzerland; George Denton, Institute for Climate Change, University of Maine at Orono; Joerg Schaefer, Vincent Rinterknecht, and Wallace Broecker, Lamont-Doherty Earth Observatory of Columbia University Palisades, N.Y; and James Teller, University of Manitoba,Winnipeg, Canada

Eos, Vol. 86, No. 40,4 October 2005

Northwestern outlet to Arctic Ocean via Mackenzie River

Page 365

Fig. 1. Digital elevation model of central North America showing the southern, eastern, and north­western outlets of Lake Agassiz.

95°W 90°W 85°W 80° W 75°W 0 5 10 15 0 10 20 30 0 30 60 90 i i i i i i i i i i

30°N

25°N

20°N

30°N

25°N

20°N

30°N

25°N

20°N

V

i 1 1 1 1 i 1 1 1 1 i 95°W 90°W 85°W 80°W 75°W

NOAA GOES-12 Infrared

TOPEX

J a s o o - 1

. . . . i . . . . i . . . . 1 1 1 1 1 « . i . 1 1 1 1 1

ERS-2

_

t • t . 1 1 . 1 1 1 1 . i . i V Envisat -

i W i i i i i i i

0 5 10 15 0 10 20 30 0 30 60 90

wave height (m) wind speed (m/s) sea level (cm)

Fig. I. The left column shows a comparison of GOES 12 infrared images and altimeter data collected by (top) Jason 1 and TOPEX, (middle) Envisat and ERS-2, and (bottom) GFO during near-coincident overflights of Hurricane Katrina on 26,27, and 28 August 2005. The images were taken within 20 minutes of the altimeter passes. The three columns on the right show the altim­eter measurements of wave height, wind speed, and sea level anomaly, respectively, as a function of latitude along the altimeter tracks shown on the infrared images.

Page 366

Eos, Vol. 86, No. 40,4 October 2005

51 °N

Lake Nipigon

8400 10Be y r s / y 7

V

V

10,190±40 BETA195959

10,400 10Be yrs 13,100 10Be yrs

9100 10Be yrs

• A Y <7s>_

°o,

\ Ouimet Canyon

Thunder Bay

is ^ 7

14,050±100^7 ETH-30180

20±40 BETA195959

10,550±75 ETH-28939

12,000±85 ETH-28945

46°N

V \ Brule Spillway 100 km

Sample Locations with 14C select ages

V Core Sites Jan 051 • Core Sites Jul 0 4 ^ V Core Sites Mar 0 4 ]

V Teller Feb 041 A Exposure Sites Sep 03 \

• Dyke et al. (2003) ]

Fig. 2. Core sites and preliminary radiocarbon and beryllium ages from the Thunder Bay region. Radiocarbon ages indicate a sequential northeast retreat of the ice sheet, while beryllium ages show greater variability along one ice margin. The blue arrow west of Thunder Bay is the hypoth­esized eastern outlet meltwater route.

Page 372

Petrology lanetary Science

Chemistry /Physics Phase Equi l ibr iaTV of Interiors, Impact M a g m a Formation Processes

Elastic and Anelast ic Properties

Volati le Degassing Retention

Mineral Physics

Interior Chemistry,

Part i t ioning, Diffusion

• i

Electromagnetic and Iron Alloy

Properties

Superhard and Novel Materials

Geomagnetism

Thermal and

Rheological Propert ies

Page 373

Fig. 1. Links of mineral physics with other Earth science fields.

Eos, Vol. 86, No. 40,4 October 2005

EWG 1970s: 16.7

Fig. 2. Summer fresh water content (meters) averaged over decades. The top panel is from an AOMIP model; the bottom panel is based on temper­ature and salinity fields from the Environmental Working Group Atlas of the Arctic Ocean for the 1950s to the 1980s, and on hydrographic stations (black dots) in the 1990s and 2000s.

Page 371

Depth (m)

0 40 Organic %

0403 Crescent Lake l i thostrat igraphic units

WBSBBk peaty gytt ja . a _ _ marl mmm gytt ja ?7YS* peat • •V • i sand

A M S 1 4 C ages 9,335±70 ETH-28527 [9,090±70 ETH-29208] [9,180±70 ETH-29210]

N 9,295±70 ETH-28528 [9,290±70 ETH-29209] [9,660±40 Beta-194056]

\ 9,595±75 40 ETH-30174

Sample Sites A Core Sites Mar 05 # Repl icate Sites Mar 05 • Cores Sites Jan 04 V Exposure Site Sep 03

• Fisher Prior Work

\ 0 Carbonate %

1 4 C ages in brackets are same sample dated tw ice

Dyke et al.

Moraine

(2003)

Fig. 3. Fort McMurray area with an example of core stratigraphy from Crescent Lake on the proximal side of the Firebag Moraine. The lowest radio­carbon age of 9595±75 is taken as a minimum age estimate for this moraine. The geometry and position of the ice sheet at the Firebag Moraine would have prevented any meltwater draining northward at that time.

Page 372