Global models have been used in numerical weather prediction and climate research for a long time. Numerical weather prediction models make a forecast of the atmospheric state for up to two weeks based on an optimally determined initial state. The spatial resolution of the model ought to be as high as allowed by available computer hardware.
Climate simulations must be extended over longer periods, in many cases over several decades. This is only possible at the expense of lower spatial resolution. The atmospheric initial state is of less importance than an adequate representation of climate system components beyond the purely atmospheric part, e.g. the ocean, the cryosphere or the biosphere. The inclusion of short lived trace substances (like ozone or aerosols) as an interactive component of climate models is becoming increasingly important. This poses a new scientfic challenge as these substances do not only force on the radiative balance of the climate system but are themselves modified by feedback changes of basic meteorological parameters.
In our institute most applications are performed with the ECHAM climate model.