Buy the book


Email: pedwards at stanford dot edu

Center for International Security and Cooperation
Freeman Spogli Institute
C226 Encina Hall
616 Serra St
Stanford, CA 94306-6165

Office: 650.725.2707
Fax: 206.337.1523

The UCLA Dept. of Meteorology

Jacob Bjerknes, who founded the UCLA Meteorology Dept. in 1940, had a strong interest in the problem of the atmospheric general circulation. Yale Mintz, a graduate student of Bjerknes, received his Ph.D. in 1949. He continued to work at UCLA, becoming associate project director with Bjerknes. In the late 1950s, he began to design numerical general circulation experiments.[1]

Mintz and Arakawa

Like Smagorinsky, Mintz recruited a Tokyo University meteorologist, Akio Arakawa, to help him build general circulation models. Arakawa, known for his mathematical wizardry, was particularly interested in building robust schemes for the parameterization of cumulus convection. Mintz and Arakawa constructed a series of increasingly sophisticated GCMs beginning in 1961. "Ironically, Arakawa's first role after joining the project was to persuade him to slow the development, giving first priority to designing model dynamics suitable for long-term integrations."[2] The first-generation UCLA GCM was completed in 1963. Arakawa then went back to Japan, but returned to UCLA permanently in 1965.

In the latter half of the 1960s, IBM's Large Scale Scientific Computation Department in San Jose, California, provided important computational assistance and wrote a manual describing the model.[3]

Widespread Influence

Of all the general circulation modeling groups in the world, the UCLA laboratory probably had the greatest influence on other modeling groups, especially in the 1960s and 1970s. This was due not only to continuing innovation (particularly in cumulus parameterization), but also to the openness of the UCLA group to collaboration and sharing. Whereas GFDL, and to a lesser extent NCAR, were pure-research institutions, UCLA operated in the mode of an academic graduate program. The Dept. of Meteorology's graduates carried the UCLA model with them to other institutions, while visitors from around the world spent time at the group's laboratories.[4]

The UCLA Models

The key characteristics of the UCLA model series and its spinoffs are neatly pictured in a chart made by Arakawa. Until the 1980s, UCLA typically focused on model development, leaving "production" of the models (i.e. use in experimental studies) to other institutions. Version number given here are my own.

UCLA Prototype

The first Mintz/Arakawa model was a 2-level global, primitive-equation GCM at a 7° latitude x 9° longitude horizontal resolution. It included realistic land-sea distributions and surface topography. Mintz never learned to program computers; Arakawa carried out all the model coding.

This prototype model was abandoned about 1965.


When Arakawa returned to UCLA from Japan in 1965, he and Mintz began work on the first-generation "production" UCLA GCM. It increased model resolution to 4° latitude x 5° longitude, although it still had only two vertical levels, and introduced a new horizontal grid structure (the Arakawa/Lamb "B" grid).[5] This was an extremely influential GCM. About 1970 Lawrence Gates, a UCLA graduate, carried the model with him to the RAND Corporation, where he used it in a series of studies sponsored by the Advanced Research Projects Agency of the U.S. Defense Dept. The RAND version of the model was eventually carried to Oregon State University.[6]


The second-generation UCLA model essentially extended the vertical resolution of the second-generation model. Three-level and nine-level versions were built. This model was carried to three NASA laboratories. In 1972, it was adopted by the Goddard Institute for Space Studies in New York (GISS), whose current model is a direct descendant. Later in the 1970s it traveled to the Goddard Laboratory for Atmospheric Sciences and the Goddard Laboratory for Atmospheres.


This 12-level model used the Arakawa/Lamb "C" finite-difference horizontal grid. All subsequent UCLA models have also employed this scheme. Two versions of this model, with slightly different sets of prognostic variables, were built in the mid-1970s. One version "was exported to the U.S. Naval Environment Prediction Research Facility and the Fleet Numerical Oceanographic Center, both in Monterey, California. This model evolved into the operational NOGAPS forecasting system."[7] It was also given to the Meteorological Research Institute in Tsukuba, Japan, where it continues to be used in a wide variety of forecasting and climate studies.


Work on the fourth-generation UCLA model began in the late 1970s. The chief innovation of this model generation was a new vertical coordinate system which used the top of the planetary boundary layer as a coordinate surface. A version of this model remains in use at UCLA into the present.

UCLA IV was also adopted by the Navy research centers mentioned above. In addition, it was taken to the Goddard Laboratory for Atmospheres, in the early 1980s. Code for this model was extensively rewritten.[8] In 1988, the model was brought to Colorado State University by David Randall, another former student of Arakawa.

Versions of this model made their way to Lawrence Livermore National Laboratory, and also to the Central Weather Bureau of the Republic of China.

{link to chart of Arakawa model}

Current UCLA Models


UCLA Dept. of Atmospheric Sciences

Back to the AGCM Family Tree


[1] Y. Mintz, "Design of Some Numerical General Circulation Experiments," Bulletin of the Research Council of Israel 76 (1958): 67-114.

[2] D.R. Johnson and A. Arakawa, "On the Scientific Contributions and Insight of Professor Yale Mintz," Journal of Climate 9, no. 12 (1996): 3211-3224.

[3] W.E. Langlois and H.C.W. Kwok, "Description of the Mintz-Arakawa Numerical General Circulation Model," (Dept. of Meteorology, University of California at Los Angeles, 1969).

[4] A. Arakawa, interviewed by Paul N. Edwards, July 17-18, 1997, at University of California, Los Angeles.

A. Arakawa, n.d. Chart.

[5] A. Arakawa and V.R. Lamb, "Computational Design of the Basic Dynamical Processes of the UCLA General Circulation Model," in General Circulation Models of the Atmosphere, ed. J. Chang, Methods in Computational Physics: Advances in Research and Applications (San Francisco: Academic Press, 1977), 173-265.

[6] D. Randall, Colorado State University General Circulation Model: Introduction (n.d. [cited ); available from

[7] Ibid.

[8] Ibid.