File last modified 3 December 1998
23.1 A History, in Brief
General circulation modeling has its own subculture within the scientific community. Whether harmful or beneficial all subcultures have their own jargon and history. Sometimes it becomes very difficult for one of the uninitiated to converse with a member of this subculture unless they are "up" on the lingo (kind of like an FBI agent trying to talk to a Berkeley student in the late 1960's). So to help our students we begin this section of the course with a brief history of GCM modeling in the United States and introduce you to some of the legendary names of the period. It's not important that you know the capabilities of the ol' "CDC stretch", but it helps to know that in the late 1960's it was considered the hottest box around. Based on your own experience with computers and this time index, you will be able to draw your own conclusions about its capabilities.
In 1969 Kirk Bryan published his seminal paper in J. of Computational Physics (Bryan, 1969, JCP, 4, 347-376) that laid out the primitive equations that 3-D general circulation modeling would be using for the next 30 (or more?) years. Then very shortly after, or actually in conjunction with, Mike Cox produced the first FORTRAN coding of Bryan's primitive equations on the old CDC "stretch" 7600. As time went on and computers became better suited for doing this type of computation Mike Cox continued to work at NOAA's GFDL in Princeton, NJ eventually moving on up to the UNIVAC 1108. Then in the early 70's Bert Semtner rewrote this code for the new CRAY "super"computers (Semtner, 1974). Mike Cox continued to work on and improve his code and in the late 1970's early 1980's produced a vectorized version that ran on the first Texas Instruments ASC. Then in 1984 (Cox, 1984) Mike Cox adopted Bert Semtner's code and optimized it for the CDC Cyber 205. Then in the early 1990's Ron Pacanowski (Pacanowski et al., 1991) rewrote Mike Cox's code in modular programming style and gives birth to MOM-1 with Dixon and Rosati. Later Pacanowski, Goldberg, Rosati, and Dixon generalized this code further to produce MOM-2 (see http://www.gfdl.gov and follow the links to the latest version MOM-2 v2.2).
The goal all along was to produce a flexible research tool sufficiently generalized to make it useful for ocean and coupled air-sea modeling over a wide range of time and space scales. The pursuit of this goal continues today. At the National Center for Atmospheric Research (NCAR) they are developing the NCAR Climate System Model (CSM). The Journal of Climate recently had a special issue dedicated to the release of version one (volume 11, June 1998) and a great deal of this climate system modeling code is available for not only the ocean, but also for the atmosphere and terrestrial systems at http://www.ucar.edu/ just follow the links.
GoTo Next Section