Modeling very simple situations with accurately quantified data will yield models that diverge quickly enough.
Modeling complex systems with many loosely quantified parameters is an exercise in the shortcomings of computer modeling. Simply moving data within reasonable margins of error will give tremendously different results.
The oceans are the drivers of climate, to be sure, but the oceans are affected from below by geothermal phenomena that we may not even know about and they are affected from above by atmospheric phenomena that we barely understand.
A model based on simple trends is far more accurate than ANY of the climate models have ever been.
I'll produce a link to that soon.