N.Wells
Posts: 1836
Joined: Oct. 2005
|
In climate models, you model all the physical interactions as best you can from fundamental chemical and physical processes, balancing mass and energy, while paying attention to fluxes, volumes of reservoirs and sinks, and so on and so forth. You feed in all the basic data and best estimate for a known region for a given time interval, say the American SW (topography, available moisture sources, incoming radiation, etc.) and you match results to data from all available meteorological stations, data like lake levels, stream flows, snow pack depths, etc., etc. You tweak all this until your model is capable of correctly reproducing the actual data.
Generally, you make this as fundamental as possible, rather than "as simple as possible" because reality is not simple, but fundamentals avoid fudges and false simplications and proxies.
Once you've got that accomplished, you are ready to test the model. First, you will likely put in parameters for a different time interval with different weather (say, drought years, or very wet years, El Nino conditions, or some such), and you will test whether your program predicts results that match meteorological and climatic data for that new period. Then you try some other conditions. Then you expand your program to cover adjacent areas to see if it still works. If all of that holds up, then you can start playing with hypothetical inputs to see what happens, for example during an ice age.
|
