 |
||||
|
||||
|
|
 |
||
|
||||
      |
Great Lakes Sensitivity to Climatic Forcing
Background
Introduction | Paleogeography | New Findings and Hypothesis | References Cited | NSF Proposal
Introduction
The watershed of the Great Lakes lies between the Hudson Bay and Gulf of Mexico watersheds of North America, and drains to the North Atlantic Ocean via the St Lawrence River. Approximately one-third of its 770,000 km2 area is the water surface of the 5 Great Lakes (Fig.1).
The climate and net water supply is influenced by the frequency of occurrence of the Arctic, Pacific, and Maritime Tropical air masses over the watershed (Fig. 2). The moist Maritime Tropical air mass is the main source of precipitation. All three air masses overlap the watershed, and slight variations in long-term atmospheric dynamics can induce variations in the degree of overlap with resulting changes in climate and moisture supply (1,2).
For the last century, fluctuations in moisture supply have been modest, resulting in an overall range of 2 m in annual water levels (3). Global warming scenarios project future temperature increases and basin runoff decreases that could result in reductions of outflow river discharge and drops in lake level ranging from 0.23 m to 2.48 m (4). The more extreme projections are beyond the limit of natural variability (5). This fact combined with the high level of dependence of aquatic ecosystems and economic and other societal interests on future water levels makes knowledge of the sensitivity of the lakes to climate change an important issue.
Figure 1: Modern Great Lakes (blue areas labeled with lake name). The Great Lakes drainage basin is outlined in white on this 30-arc second shaded-relief digital elevation model coloured in 100-m intervals of topography and bathymetry. Color bar below figure indicates land elevation relative to present-day sea level (reds and greens) and lake depth (blues). (Files are available to this project; a contoured printed version was published 1996 as Canadian Hydrographic Service Map 880A). Also illustrated is location of Volo Bog (V) near southern Lake Michigan, and North Bay (NB). Owing to glacial isostatic depression, the northern Great Lakes drained northeast through the North Bay outlet to Ottawa River (10-5 ka). Small lakes proposed for coring and investigation include: Lake of the Clouds (C), Pretty Lake (P), Lake Le Boeuf (B), Seneca Lake (S) (core already at hand), Derby Lake (D), Green Lake (G), Found Lake (F), and Jack Lake (J).
Figure 2: Present-day distribution of airstreams (5-month dominance) over North America (adapted from Bryson, 1966; Bryson and Hare, 1974).
Paleogeography of the Great Lakes
The Great Lakes basins are generally thought to be incrementally eroded depressions in relatively softer zones in the underlying Paleozoic sedimentary rock, and in weaker structural zones of the adjoining Precambrian metamorphic Canadian Shield rocks (6) during glaciations of the last two million years. The latest Laurentide Ice sheet reached a maximum position about 18 to 21 ka BP in the northern United States. After 14 ka proglacial lakes in front of the northward retreating ice margin overflowed the continental drainage divide to the south. Punctuated by intervals of readvance, the northward retreat of Laurentide ice and proglacial lakes finally uncovered the entire watershed before 8 ka. Major ice sheet drainage subsequently bypassed the Great Lakes via the Ottawa and St. Lawrence rivers to the Atlantic Ocean (Fig. 3). During deglaciation a complex series of lakes formed as a consequence of the interaction of oscillatory ice retreat and advance with topography. Water surfaces fluctuated as lower outlets through previously ice covered lowlands were opened and closed. Lake surfaces also rose as ice sheet retreat allowed hydrologic connections with upstream Lake Agassiz discharge, and/or with enhanced ice sheet meltwater discharge.
Throughout this period, lake elevations also changed as outflow channels were eroded and downcut. Relative water levels were also progressively altered as the basins were tilted up to the north-northeast due to glacio-isostatic rebound. This differential rebound affected lakes by reducing water depths in the north while increasing depths in the south. For the northern Great Lakes differential rebound caused complete diversion of discharge from a northern outlet to southern outlets about 5 ka (7-12).
Only in Lake Michigan was an interpretation made, based on study of ostracode fossils, that climate had affected the limnology and possibly forced this lake into hydrologic closure about 7000 BP (8,13). Apart from the above interpretation for Lake Michigan, the traditional understanding of the evolution of the Great Lakes generally implied 3 assumptions:
(1) the water balance was always positive so lake levels were mainly determined by the elevation of their outlets;
(2) evidence of water levels lower than present could be explained by relative lake level change forced by differential crustal rebound; and
(3) climatic change over the basin was modest and resulted only in level modulation, similar to or only slightly greater than present conditions, and major episodes of dry climate and hydrologic closure had not occurred.
Figure 3: Location of the Laurentide Ice sheet about 8900 cal BP (7.9 ka - orange shading) and 8400 cal BP (7.5 ka - yellow shading)(Outline from (39) and A. F. Dyke, pers.comm., 2002). Note that drainage from proglacial lakes Agassiz and Ojibway (blue shading) after 8 ka bypassed the Great Lakes basin, discharging via Ottawa and St. Lawrence rivers to Atlantic Ocean (blue arrows show direction of flow).
New Findings and Hypothesis
The above view of the Great Lakes being relatively unaffected by high-amplitude climate change began to be questioned by us when we discovered that the south basin of Lake Winnipeg (Fig. 4) had dried up between 7.5 ka and 4 ka (14). Our questions were supported by existing and emerging evidence in the Huron basin.
Recent seismic reflection data and sediment cores in northern Lake Huron had revealed a sediment architecture that implied lowstands in the lake history, the last of these being a strong event dated 7.9-7.5 ka (15-17) (Fig. 5). Subsequent studies on the sill at the entrance to Georgian Bay discovered 6 submerged tree stumps with 14C ages between 7230 and 8560 BP in water depths ranging down to 43 m (18). A beach deposit, recognized at 53 m depth, together with pollen evidence of shallow-water in a deepwater core suggested that lake level at 7.5 ka was ~ 35 m below the basin outlet at North Bay (19).
Figure 4: Shows Great Lakes (inside blue line) and their drainage basin (blue line), outflowing St. Lawrence River (SLR), present Prairie Peninsula vegetation zone (PP), locations of Deep Lake (D) (28) and Lake Winnipeg (LW) (size of dots not to scale), and Hudson Bay.
Figure 5. Seismic boomer profile (2-6 kHz) showing sediment sequence boundaries (labeled with color names) in relatively deep water in northern Lake Huron. The boundaries are conformities which can be traced to unconformities in shallower water, indicating erosion by wave activity during lowstands in the early-middle Holocene history of Lake Huron. Our target study interval (9400-7700 cal BP; 8.4-6.8 ka) includes the period of the last (Light Blue) lowstand dated 8900-8400 cal BP (7.9-7.5 ka).
Our re-assessment of Great Lake levels included development of a new synthesis of glacio-isostatic rebound using an empirical model to describe vertical earth movement with time (20). The parameters of the model were controlled by the upwarped glacial lake shorelines throughout the Great Lakes. With this model we were able to restore indicators of former water levels of any age to their original elevation in the Huron basin. The indicator data (21,22) consisted mostly of isolation basins (11,23), but also the evidence of dated submerged tree stumps (18). As a result of the restoration, indicators of similar age would cluster at similar positions in the field of a plot of original elevation versus age(Fig. 6). A lake level curve was estimated through the restored indicator data. Episodes of low lake level agreed remarkably well with the ages of the seismically determined lowstands (15). The Huron basin lowstand in the target interval is tied to the submerged beach of Blasco (19) and to a lithological unconformity named after Stanley by Hough (24). A similar lowstand is seen in the Michigan basin (8,25-27).
Model estimates of the elevation history of potential Huron overflow outlets were added (Fig. 6). The trajectory of the North Bay outlet passes 40-50 m above the inferred water surface at 7.9-7.5 ka (8900-8400 cal BP). A 40-50m depression of the lake below its outlet can only be attributed to a severe dry climate. This result contrasts with an open overflowing lake preceding and following the episode of closed lake status, and implies the impact of a high-amplitude climate change.
Figure 6: Paleo-lake levels (thick blue line) in the Huron basin. Within the target time interval of this study (9490-7700 cal BP), the water level was below the lowest possible basin overflow route (dashed red line marked North Bay outlet) from ca. 8900- 7800 cal BP, indicating the lake was hydrologically closed, drawn down by a climatic shift to a drier regime with more evaporation and/ or less precipitation and runoff. In this plot of elevation vs age, radiocarbon-dated lake level indicators (olive-colored symbols and lines showing one-sigma uncertainty) and overflow outlets (colored dashed lines) have been re-constructed by removing the effects of glacial rebound, and are shown at their original altitudes. The symbols indicate the type of indicator: rightpointing triangles represent isolation basins from which the large lake has regressed; left-pointing triangles represent basins which the lake has transgressed and inundated; circle represents water level at a delta or beach; downward-pointing triangles represent in situ tree stumps requiring a lake level below their root elevations. Seismic lowstands are low lake levels identified by seismo-stratigraphic sequence boundaries in seismic profiles. Stanley unconformity is a sedimentary erosional discontinuity in northwestern Lake Huron indicating low lake levels (9), believed to be correlative with the Flowerpot beach now submerged 53 m beneath Georgian Bay.
A hypothesis for the early severe dry climate impact from 7.9 ka to 7.5 ka (8900-8400 cal BP) is suggested by a correlative episode of more negative δ18O values of the sediments of Deep Lake, MN (28) (Figs. 4,7).
It was noted that the Laurentide ice in Hudson Bay was downwasting rapidly during this period, and Yu and Wright (29) suggested that this reduction allowed for more frequent outbreaks of Arctic air. The frequent cold air cover would cause southern air masses to rise with the effect that moisture would have been condensed at lower temperatures and the resulting precipitation would carry a more negative isotopic signal. Because Arctic air lacks moisture (30,31), we suggest that the outbreaks were extensive enough to cause a net reduction in water supply and an increase in evaporation leading to drawdown and closed lake status. Thus, the early part of the closure, until 7.5 ka (8400 cal BP) is thought to reflect a cold dry climate. We attribute the latter part of the lowstand to an increase in warm, dry westerly air flow. An increase in warm, dry westerly airflows at this time is also consistent with reorganization of atmospheric circulation observed in the Elk Lake, MN record.
This change is attributed to rapid change in the relative surface areas of ice, land and ocean (2), and with the 8.2 cal BP cool event recorded in Greenland ice cores (38) attributed to final drainage of glacial lakes Agassiz and Ojibway (39,40). This increase was suggested by Forester et al. (13) based on ostracode evidence of inferred low water (Fig. 8) in Lake Michigan that peaked around 8200 cal BP (7.4 ka). An enhanced warm, dry climate is also shown by reconstructed mean annual temperature and precipitation from pollen data (32) which suggest increased aridity in the southern Lake Michigan area peaking in the 8400-7800 cal BP interval (Fig. 9). This suggestion is consistent with an eastward expansion of prairie toward the southern Great Lakes basin (33-37) (Fig. 4, prairie peninsula). The occurrence of very negative δ18O values in Huron and Michigan (Fig. 10) during the target interval presents a challenge for our hypotheses of climate and closed lake status. High rates of evaporation implied by the postulated dry climate would be expected to shift isotope ratios to more positive values. The δ18O depleted values are more characteristic of meltwater (41-43a).
Alternate explanations include glacial groundwater reflux, backflow from proglacial lake discharge down the Ottawa River, the influence of cold Arctic air on precipitation, or melt from residual ice masses. Sources of the water with very negative δ18Ovalues will be sought by establishing spatial gradients of isotopic composition within the Great Lakes and between small atmosphere-charged lakes around the basin for the target time interval.
Figures 7: Note correlation of lowest Huron water levels and the most negative δ18O values in Deep Lake (D in Fig. 4) (28). Green bar indicates similarity in timing of end lowest Huron levels, final drainage of proglacial lakes Agassiz and Ojibway to Hudwon Bay, and 8.2 cal KA Greenland cool event (39).
Figure 8: Similarity in changes of lake levels and an ostracode shore index (with lag) (adapted from (13)) is indicated in the Michigan basin for the target study interval (yellow).
Figure 9: Mean July temperature and mean annual precipitation derived by transfer functions from pollen content at Volo Bog (Fig. 1) suggest increased evaporation and reduced water supply for southern Lake Michigan about 7800 cal B.P. Present values are 22.4°C and 933 mm.
Figure 10: δ18O ratios in Huron ostracodes and Michigan mollusks An explanation will be sought in this project for the much larger shift (about 4-fold greater) to more negative δ18O values in Huron basin compared with Michigan basin at 8900 cal BP.
Overall, we recognize the need to quantify the high-amplitude paleoclimate changes and their impacts on the hydrology and lake levels of the Great Lakes basin, with particular reference to the Huron-Michigan portion of the drainage system, for the time interval of 9400 to 7700 cal BP.
References Cited
NSF Proposal
Last updated: 2006-08-08 ks