[palmer]

Here's one problem with your models: the MCS that came through No. VA this morning affected our climate temporarily affected our climate by bringing down colder air from aloft and leaving behind lingering clouds. None of that is in current models (resolution of about 200 miles). It certainly matters to accurately model the effects of water vapor in both condensed and uncondensed forms. Without that, you fail to model the consequences of forcing (man-made increases in CO2 causing slight warming causing increases in water vapor). There is no way to accurately depict the extra water vapor without clouds and precipitation that are often localized phenomena (a small local vortex in our case).

Fortunately, computation power is increasing and adequate models should arrive in the next decade or so.

[cogitator]

The short answer is, the current state of the models is sufficient to determine Earth's climate state and discern the primary effects of changing climate "drivers" There is still uncertainty in the size of some climate influences and feedbacks, clouds notably being one of the larger uncertainties. For this reason, every model output is always associated with some uncertainty, so that model output will not provide exact replication of events and trends. For example, while coupled-ocean atmosphere models generally reproduce ENSO frequency, they have yet to capture a good reproduction of individual event intensity or actual event timing. But they still characterize the ENSO contribution to global climate over decades well.