Toward a Realistic Computerized Ocean

"A two-foot rise in sea level could have precisely the same effect on ocean waves that a 2 degree C. rise in temperature would have on heat waves. It would make the freak disaster commonplace."
-- J. Weiner, The Next One Hundred Years

The ocean is rising! The ocean is rising! It's a cry reminiscent of Chicken Little, who thought the sky was falling when an acorn hit him on the head. The important difference is the ocean is rising.

Scientists estimate sea level is roughly four inches higher than it was 100 years ago, and satellite measurements show it going up one to three millimeters a year. Do the rising levels reflect climate change associated with alarms about global warming, as many scientists believe? Or, as others argue, are they part of normal fluctuations in weather cycles that will even out over time? Either way, should we be worried? What are the potential effects for coastal population centers?

The need for answers runs smack up against the vast uncertainties inherent in a system as complex as Earth's climate. The best tool -- the only tool -- we have for assimilating the multitude of variables and trying to make rational predictions is computer modeling, and it's not yet good enough. "In order for climate models to be reliable," says University of Miami ocean scientist Rainer Bleck, "you want them to be able to accurately predict climate for stationary conditions, without the added variabilities of increasing carbon dioxide in the atmosphere. We're not at that stage yet."

Bleck and his collaborators, University of Minnesota computer engineers Matt O'Keefe and Aaron Sawdey, have taken a notable step in the right direction. Exploiting the parallel-processing ability of the CRAY T3D at Pittsburgh Supercomputing Center, they have advanced the state of the art is ocean modeling. In a simulation that ran for 10 days on 256 T3D processors, half the machine, their model of circulation in the Atlantic Ocean correctly predicted the course of the Gulf Stream. No other circulation model of the entire Atlantic Ocean has done this. "This degree of realism," says Bleck, "is very much a step up from previous simulations." These results prove the feasibility of a revised approach to ocean modeling.


Modeled Surface Temperature of the North Atlantic
Color shows surface temperature, red corresponding to high temperature. The model begins in winter, and as time proceeds through 30, 90 and 150 days, warm waters from the equator migrate northward. Unlike most prior modeling, the Gulf Stream breaks off from the North American continent at Cape Hatteras and establishes itself on a north-east course.

Initially, the Gulf Stream cuts through a warm-water mass north of the Cape, left over from a coarse-mesh simulation used to initialize this very high-resolution run. As the model adjusts to the Gulf Stream current, warm water left of the Gulf Stream rolls up into a clockwise spinning ring. By five months, the last image, this adjustment process is essentially complete.

(ANIMATION: 1.7 MB)
Each frame in this animation of the surface temperature of the Gulf Stream represents a seven day period.

Researchers: Matt O'Keefe, University of Minnesota; Aaron Sawdey, University of Minnesota; Rainer Bleck, University of Miami.
Hardware: CRAY T3D
Software: MICOM (Miami Isopycnic Coordinate Ocean Model)
Keywords: ocean modeling, computer modeling, Gulf Stream, Atlantic Ocean, grid size, MICOM, heat diffusion, oceanography, climate research, sea level, weather cycles, surface temperature, Northe Atlantic, isopycnic, global warming.

Related Material on the Web:
MICOM for T3D and SGI Home Page at the University of Minnesota, other information including graphics and animations.
MICOM Group Home Page at the University of Miami
Matt O'Keefe's Home Page
Projects in Scientific Computing, PSC's annual research report.

References, Acknowledgements & Credits