Early in the 20th century a Norwegian named Vihelm Bjerknes claimed that atmospheric physics was advanced enough to forecast weather by calculations. He came up with 7 equations that predicted "large-scale atmospheric motions." Unfortunately these equations did not provide fast calculating methods nor provide accurate forecasting techniques.
In 1922, Lewis Fry Richardson created the first numerical weather prediction system (NWP). His method used grid cells and provided a simpler method of calculating Bjerknes's equations plus adding an eight equation (a calculation of atmospheric dust). But again the method turned out to be too slow and ended miserably. It was not until the 40s and the invention of digital computers that his technique became practical.
In the 1940s with WWII beginning, Princeton's John von Neumann began to see parallels in computer forecasting and simulations in nuclear weapon explosions.
Both are non-linear problems of fluid dynamics. He believed that weather modeling might lead to weather control. Which could turn the war around with this new weapon of war. Under his wing, Jule Charney and Arnt Eliassen began The Meteorology Project, which ran its first model in 1950. Their model, like Richardson's earlier model, used grids and differential equations. Their worked started a new trend that sent researchers scrambling to find out more.
The Royal Swedish Air Force Weather Service in Stockholm was first in the world to begin the routine real-time numerical forecasting. Forecasting was done 3 times a week starting in December of 1954.
Around 1952, again von Neumann convinced members of the Weather Bureau and other agencies to establish a Joint Numerical Weather Prediction Unit (JNWP). Beginning in 1954 in Maryland and later starting real-time forecasting in May of 1955, these new models replaced the primitive models of Bjerknes and Richardson with barotropic and baroclinic models.
By the 50s the weather forecasting had become regional. There were a great many hoops to leap through and technological limitations hindered a lot of early work. General circulation models (GCM) were the pot of gold at the rainbow. But by mid 1950, Norman Phillips developed a 2 layer, hemisperic, quasi-geostrophic computer model. Today it is still regarded as the first GCM. The bullet had been fired and everyone was ready to play the game. Between the 50s and 60s, 3 new groups began to build and develop new 3 dimensional GCMs based on Bjerknes's and Richardson's work.
These new agencies were The General Circulation Research Section (later known as GFDL), UCLA Department of Meteorology, Lawrence Livermore Labs, and the National Center for Atmospheric Research. Each of these agencies over the last few decades have developed new and innovated technologies in the quest for the perfect circulation model.
By 1965, these groups had established the norm for modeling. Between 1965 and 1975, RAND Corporation, Goddard Institute for Space Studies, European Center for Medium Range Weather Forecast (ECMRW), and Australian Numerical Meteorological Research Centre began their work building on the previous research and including new innovations along the way.
Two of the newest innovations that were put to use were two types of models; coupled atmospheric- ocean models and spectral transform techniques. Coupled atmospheric- ocean models are a technique that uses grids. Basically you lay down a grid on the Earth, take your weather measurements and then calculate the differential equations needed in order to figure your data. The only disadvantage to this method is that calculating is slow and the models develop can only see so far into the future. The other method, spectral transform techniques, uses sine and cosine equations for infinite amount time. The advantage to this method is that it calculates models faster and you can develop models that see further into the future. Unfortunately, this method is very math oriented and the calculations are very complicated.
With the beginning of the technological age of faster and better, the GCMs became technologically advanced as well. Longer models runs were now capable and new data was developed. Political advisors began to take notice of what these models predicted and funding was available for even better techniques. With this funding new groups formed again, which included; Max Planck Institute, NASA Goddard Laboratory for Atmospheres, NASA Goddard Laboratory for Atmospheric Sciences, Colorado State University, Oregon State University, National Meteorological Center, and Lawrence Livermore National Labs. With these new groups, the future was becoming a reality with new and better techniques. Politics and science bound together. The public became more informed and now these models are increasingly becoming a norm in the science of predicting how we as a society are impacting our environment.
History taken from the website: http://www.stanford.edu/group/STS/sloan.project/intro.html
