[gorkish]

> I'd assume it's also a long way to go from doing it for a small section of the world to doing it everywhere

This probably isn't a good assumption. It's likely more about it being much faster to iterate/validate the methodology on the smaller dataset of just the Mediterranean (2.5 million km^2) before spending the effort to run it on the entire ocean (361 million km^2, 144x larger data).