I'm interested in the process that goes from, for example, huge data sets of temperature, pressure, precipitation etc readings to a model of the atmosphere. I'd like to know about how much relies on a priori knowledge of Navier-Stokes, and how much curve fitting is done.
Some sort of general overview of the process (ideally a fairly accessible one) would be perfect.
I'm looking for a reference, specifically. I know more or less the science from talking to people who do this stuff, but I would like to have some summary article or book chapter overview I can refer to in written work.