WHOI  WHOI People  

Modeling Hudson Bay Fresh Water Discharge: Today and 8500 Years Ago

OCCI Project Funded: 2001

Proposed Research

Recent evidence suggests that the melting of the last major remnant of glacial maximum ice, an ice dome in Hudson Bay, occurred about 8500 years ago and released a huge volume of proglacial lake water and ice through Hudson Strait into the Labrador Sea. However, evidence for such a meltwater event has not yet been found in the Labrador Sea, although it is present farther south in Cabot Strait and off Nova Scotia. We hypothesize that this distribution occurred because the freshwater outflow stayed entirely over the continental shelf and, therefore, would not show up in Labrador Sea cores. We will test this hypothesis by studying the dynamics of large freshwater outflows onto a continental shelf using a primitive-equation numerical model. Our goal is to determine the most likely fate of the freshwater for a range of parameters appropriate for the outflow event of 8500 years ago. Our goal is to determine the most likely fate of the freshwater for a range of parameters appropriate for the outflow event of 8500 years ago.

Final Report

Some 8500 years ago, much of northern Canada was covered by a huge lake, called Lake Agassiz, which was formed from glacial melt water and had an estimated volume of 2x1014 m3 (nearly 10 times the total volume of the present day Great Lakes). Recent evidence suggests that a glacial ice dam failed (melted) about 8200 years ago, causing Lake Agassiz to drain through Hudson Strait within a year or less (Barber et al. 1999, Clarke et al. 2003). The resulting freshwater discharge rate into the Labrador Sea (5x106 m3/s) would have been enormous (250 times larger than the present discharge rate from the Mississippi River) and is conjectured to be the cause of an abrupt change in the global climate - the strongest cooling event in the last 10,000 years (Alley et al. 1997). The basic premise is that such a large volume of freshwater, if spread over the surface of the North Atlantic, could prevent (or at least severely reduce) the normal formation and sinking of dense water, and subsequently reduce the large-scale overturning circulation in the ocean which plays a critical role in the global climate through its redistribution of heat. Several numerical modeling studies support this scenario by suggesting that a reduction or halting of the overturning circulation may cause abrupt cooling of the global climate (Renssen et al 2002, Manabe and Stouffer 1997, Seidov and Maslin 1999). 

However, this entire argument hinges on the assumption noted in italics above; that the freshwater discharge spreads over the North Atlantic. Yet, the large-scale numerical models do not have enough resolution to realistically simulate the discharge of freshwater and its subsequent pathway. Rather, they simply add freshwater to some portion of the North Atlantic. Furthermore, these numerical model studies show that the climate response depends critically on the distribution of the freshwater. Therefore, knowledge of the pathway that a large freshwater discharge is likely to follow is crucial to our understanding of the impact of the drainage of Lake Agassiz on global climate.  

With support from the Ocean & Climate Change Institute, we have used both theory and a regional numerical model to investigate the likely pathway of a large freshwater discharge like that from Lake Agassiz. We know from theory and observations that freshwater discharges into the ocean from large rivers and estuaries do not simply spread uniformly away from their source. Instead, Earth's rotation causes the freshwater to turn (to the right in the Northern Hemisphere) and flow along the coast as a narrow, buoyant coastal current. For example, the present-day freshwater discharge through Hudson Strait into the Labrador Sea turns southward and flows along the Labrador Coast, resulting in lower salinities over the continental shelves all the way to Cape Hatteras, North Carolina (Loder et al. 1998). A recently developed theory (Lentz and Helfrich 2002) provides estimates of the width, thickness and speed of such a buoyant coastal current, all of which are consistent with numerous present-day observations of buoyant coastal currents. 

In the case of the drainage of Lake Agassiz, we know little about the details of the outflow; for example, how much ocean water was mixed into the lake water as it first entered the ocean and hence the salinity of the resulting buoyant current. Nevertheless, we have applied the theory for a reasonable range of plausible salinities and transports, and we predict that the resulting buoyant coastal current would have extended offshore only about 50 km and would have been about 100 m deep. Furthermore, the theory predicts that the buoyant current would have propagated along the coast at 2-3 m/s (175 - 260 km/day), suggesting that it would only have taken about 21 days for the freshwater to travel from Hudson Strait to Cape Hatteras (3600 km). Thus, the theory suggests that the freshwater may have been transported a considerable distance to the south before entering the open ocean.  It would not have simply spread over the North Atlantic. 

Given the tremendous volume of freshwater drained from Lake Agassiz relative to the weaker present-day discharge, it is not obvious that a buoyant coastal current would form in the same way. Perhaps such a huge freshwater discharge would immediately separate from the coast as an unstable jet and break up into eddies that would spread the buoyant water farther offshore. To address this issue, we used a regional numerical model with relatively high resolution to simulate a large freshwater discharge from Hudson Strait. The model included both realistic coastline and bathymetry. In general agreement with the theory, the freshwater discharge in the numerical model turned to the right, filled the continental shelf with less salty water, and flowed as a buoyant coastal current along the Labrador Coast with roughly the same width and thickness predicted by the theory. The primary difference from the theory is that the buoyant coastal current in the numerical model is not smooth, but rather exhibits a "turbulent" character, with velocities and salinities that vary on short spatial scales. Nonetheless, the overall southward flow persists as a buoyant coastal current over the entire extent of the numerical model (about 600 km).   

The results of our study are clearly preliminary and not definitive at this point. We have not answered the most important question of whether or not the freshwater discharge from Lake Agassiz could have eventually reduced or halted the large-scale overturning circulation, thereby causing the abrupt cooling of the global climate, as hypothesized. However, we have raised some important questions that we feel must be addressed to understand the linkage between any freshwater discharge and global climate. For example, what are the true pathways of freshwater discharges throughout the world oceans? How long does it take the freshwater to reach climate-sensitive locations? And, more importantly, what is the salinity of the water when it gets there (i.e. how much mixing with ocean water has occurred along the pathway)? Are there other climate implications associated with unusually large buoyant coastal currents? Our plan is to pursue these issues in future studies, using this project as our motivation and starting point. 


Alley, R.B. et al., 1997. Holocene climatic instability:  a prominent, widespread event 8200 yr ago. Geology 25,  483-486.

Barber, D.C. et al., 1999. Forcing of the cold event of 8,200 years  ago by catastrophic drainage of Laurentide lakes. Nature  400, 344-348.

Clarke, G., D. Leverington, J. Teller and A. Dyke, 2003. Superlakes,  megafloods, and abrupt climate change, Science 301, 922-923. 

Lentz, S.J. and K.R. Helfrich, 2002. Buoyant gravity currents along  a sloping bottom in a rotating fluid. J. Fluid Mech. 464,  251-278.

Loder, J.W., B. Petrie and G. Gawarkiewicz, 1998. The coastal ocean  off northern North America: a large-scale view, in The Sea:  The Global Coastal Ocean. Regional Studies and Syntheses, K.H.  Brink and A.R. Robinson eds., John Wiley and Sons, 11, 105-133. 

Manabe, S. and R.J. Stouffer, 1997. Coupled ocean-atmosphere model  response to freshwater input: Comparison to Younger Dryas event.  Paleoceanography 12, 321-336.

Renssen, H., H. Goosse and T. Fichefet, 2002. Modeling the effect  of freshwater pulses on the early Holocene climate: The influence  of high-frequency climate variability. Paleoceanography  17, Art. No. 1020.

Seidov, D. and M. Maslin, 1999. North Atlantic deep water circulation  collapse during Heinrich events. Geology 27, 23-26.

Originally published: January 1, 2001