Joseph Smagorinsky - Leadership of Geophysical Fluid Dynamics Laboratory

Leadership of Geophysical Fluid Dynamics Laboratory

Following his apprenticeship and work with von Neumann and Charney, in 1953, at age 29, Smagorinsky accepted a position at the U.S. Weather Bureau and was among the pioneers of the Joint Numerical Weather Prediction Unit. In 1955, at von Neumann's instigation, the U.S. Weather Bureau created a General Circulation Research Section under Smagorinsky's direction. Smagorinsky felt that his charge was to continue with the final step of the von Neumann/Charney computer modeling program: a three-dimensional, global, primitive-equation general circulation model of the atmosphere. The General Circulation Research Section was initially located in Suitland, Maryland, near the Weather Bureau's JNWP unit. The section moved to Washington, D.C. and was renamed the General Circulation Research Laboratory in 1959 and then renamed again as the Geophysical Fluid Dynamics Laboratory in 1963. The lab moved to its current home at Princeton University in 1968. Smagorinsky continued to direct the lab until his retirement in January, 1983.

Smagorinsky's key insight was that the increasing power of computers would allow one to move beyond simulating the evolution of the atmosphere for a few days, as in weather prediction, and move toward the simulation of the Earth's climate. The intention of such simulations is not to predict the detailed evolution of the weather, but by integrating the equations of motion, thermodynamics, and radiative transfer for long enough time periods to simulate the statistics of the weather—the climate—enabling one to study how these statistics were controlled by the atmospheric composition, the character of the Earth's surface, and the circulation of the oceans.

Among Dr. Smagorinsky’s many talents was attracting creative scientists to the staff of the Geophysical Fluid Dynamics Laboratory. Two of them were climate modeler Syukuro Manabe in 1959 and ocean modeler Kirk Bryan in 1961, who spearheaded the development of the first climate model in 1969, a general circulation model that was the first approach to take into account the interactions of oceans and atmosphere. Smagorinsky assigned Manabe to the General Circulation Model coding and development effort. By 1963, Smagorinsky, Manabe, and their collaborators had completed a nine-level, hemispheric primitive-equation General Circulation Model. Manabe was given a large programming staff and was thus able to focus on mathematical structure of the models, without becoming overly involved in coding. In 1955-56, Smagorinsky collaborated with John von Neumann, Jule Charney, and Norman Phillips to develop a 2-level, zonal hemispheric model using a subset of the primitive equations. Beginning in 1959, he proceeded to develop a nine-level primitive-equation General Circulation Model (still hemispheric). By the end of the next decade, general circulation models emerged globally as a central tool in climate research. Other researchers who worked with Smagorinsky in Washington and Princeton included Isidoro Orlanski, Jerry Mahlman, Syukuro Manabe, Yoshio Kurihara, Kikuro Miyakoda, Rod Graham, Leith Holloway, Isaac Held, Garreth Williams, George Philander, and Douglas Lilly.

Development of this first climate model was based on Smagorinsky’s belief that individual inquiry would be inadequate for addressing such a complex problem. He realized that it would take large-scale numerical modeling with teams of scientists using commonly shared high-speed computers to achieve such a breakthrough. As stated in a 1992 Bulletin of the American Meteorological Society, “Dr. Smagorinsky’s almost relentless pursuit of excellence at Geophysical Fluid Dynamics Laboratory set a standard for other laboratories and centers that have contributed immensely to the growth of meteorology as a science” throughout the world. Michael MacCracken, President of International Association of Meteorology and Atmospheric Sciences, wrote following Smagorinsky's death that “From its earliest days, GFDL has been world renowned, with an outstanding set of scientists doing outstanding work that attracted scientists from around the world to come to learn and collaborate – and then return to their home countries or other institutions as outstanding scientists. Not only a whole new scientific field of investigation, but a community of scientists capable of doing it well has been created.”

Smagorinsky invited many scientists from outside the normal circle to provide the broadest perspective on weather forecasts. Very early in his career, he brought pioneering oceanographer Kirk Bryan to GFDL to account for oceanic influences on the weather; and shortly following World War II, with the nation still leery of Japan, he invited Suki Manabe, Yoshio Kurihara, and Kikuro Miyakoda to GFDL, valuing their scientific expertise and potential and ignoring the xenophobia that might have discouraged such international collaboration. He continued this practice of inviting scientists to GFDL who could take on the project of producing a comprehensive theory of atmospheric processes, valuing talent and creativity over what he regarded as irrelevant factors such as field or nationality. Jerry Mahlman, who succeeded Smagorinsky as director of GFDL at Princeton, writes that Smagorinsky "had no real interest in the 'university scientific culture' that still has a tendency to count scientific publications, rather than scientific achievements, as its measure of faculty success. Joe would have none of that. He wanted junior scientists such as us to focus on solving difficult scientific challenges of major relevance to NOAA, the United States, and the world. . . . Without Joe’s support and encouragement, would Manabe have written the first paper on the science of global warming in 1967? Would Bryan have produced the world’s first ocean model in 1970? Would Manabe and Bryan have produced the world’s first coupled atmosphere–ocean model in 1972? Would I have produced the first comprehensive stratospheric dynamical/chemical model? Would Miyakoda have pioneered extended-range weather forecasting? For my research, the answer is: almost certainly not. Without the level of scientific and computational support provided by Joe, these achievements would have required at least another decade of development to achieve success."

Smagorinsky was among the earliest researchers who sought to exploit new methods of numerical weather prediction to extend forecasting past one or two days. Smagorinsky published a seminal paper in 1963 on his research using primitive equations of atmospheric dynamics to simulate the atmosphere’s circulation. This paper fundamentally changed the approach to modeling climate. He extended early weather models to include variables such as wind, cloud cover, precipitation, atmospheric pressure and radiation emanating from the earth and sun. In order to make these simulations possible, a method was needed to account for atmospheric turbulence that occurred on scales smaller than the model's grid size but still played a crucial role in the atmospheric energy cycle. With colleagues Douglas Lilly and James Deardorff, both at the National Center for Atmospheric Research, he developed one of the first successful approaches to large eddy simulation (e.g., the Smagorinsky-Lilly model), providing a solution to this problem that is still in use, not only in meteorology, but in all fields involving fluid dynamics.

Smagorinsky earned fame for his ability to secure the world's fastest computers for his laboratory time and time again. At the memorial gathering at Princeton University following Smagorinsky's death, Suki Manabe playfully suggested that Joe always attended meetings with government officials with a resignation letter in hand, ready to present it if his needs weren't met. However he achieved his goals, he did so with remarkable consistency, much to the amazement of those who wondered how a single government scientist had such leverage in the highly competitive battle for limited resources. Jerry Mahlman wrote that “Without the level of scientific and computational support provided by Joe, these achievements would have required at least another decade of development to achieve.” This remark that Smagorinsky had advanced his field by at least a decade was echoed by several speakers at his memorial.