Climate models initially served purely heuristic purposes. This means that models were developed and used to investigate and better understand the processes governing climate and its variations. Historically, an ingenious inventor of heuristic modeling was Jule Charney. Charney was hired by the famous mathematician John von Neumann in 1948 as a young and bright meteorologist for the so-called Meteorology Project. His task was to help develop numerical weather prediction at the Institute of Advanced Study in Princeton.
Charney promoted the idea of experimenting with models and deliberately accepting even so radical simplifications that a model could hardly be considered a representation of the atmosphere. He suggested a hierarchy of models in which specialized simple models could serve for testing the understanding of individual processes, which represented only a part of the complex atmospheric interactions. The model space served as a laboratory for experimentation; and the models resembled an idealized experimental apparatus in the laboratory. Running experiments with this apparatus allowed valuable insights, informed theoretical reasoning and improved scientific understanding (Dahan 2001).
It was such type of experiment, which meteorologist Norman Phillips, one of Charney’s associates in the Meteorology Project, undertook in 1955. Phillips wanted to test whether a weather forecasting model could be used to simulate atmospheric conditions over a longer period of time than a few days. Phillips results proved surprisingly successful and produced patterns of atmospheric circulation, which resembled observed patterns. These results provoked enormous excitement. When Phillips was invited to deliver a seminar on his experiment to the Royal Meteorological Society in 1956, British Meteorologist Eric Eady grasped the importance of this work in and wrote:
"Numerical integrations of the kind Dr. Phillips has carried out give us a unique opportunity to study large-scale meteorology as an experimental science" (Eady quoted in Lewis 2000: 117).
Complex systems like weather and climate had become accessible to quantitative understanding based on the laws of physics. Phillips approach became the foundation for the development of so-called General Circulation Models (GCMs), atmospheric models which represented large scale processes in the atmosphere and served for investigating and understanding the emergence of different climates. Climate modelers of the first generation such as Joseph Smagorinski and Syokuro Manabe at Princeton and Yale Mintz and Akio Arakawa at the University of California, who started climate modeling efforts in the late 1950s, had purely heuristic interests (Heymann, Hundebøl 2017). They strove to tackle the problem of climate piece by piece and contribute to what climate scientists William W. Kellogg and Stephen H. Schneider later called a “theory of climate” (Kellogg and Schneider 1974).
References:
Dahan, D.Amy. 2001. History and Epistemology of Models: Meteorology as a Case Study (1946–1963). Archive for the History of the Exact Sciences 55, pp. 395–422.
Heymann, Matthias, Nils Randlev Hundebøl. 2017. From heuristic to predictive: Making climate models political instruments, in: Matthias Heymann, Gabriele Gramelsberger and Martin Mahoney (eds.): Knowledge and Authority: Epistemic and Cultural Shifts in Computer-based Environmental Science (New York: Routledge), pp. 100-119.
Kellogg, William W., Stephen H. Schneider 1974. Climate Stabilization: For Better or For Worse? Science 186, pp. 1163–72.
Lewis, John M. 2000. Clarifying the dynamics of general circulation: Phillips’s 1956 Experiment, in: David A. Randall, General circulation model development, past, present, future (San Diego: Academic Press), pp. 91-125.
