I don't understand, maybe I'm misinterpreting your position.
Based solely on the scale of the problems yourself presented, you must surely agree that (mega)engineering is the most factible solution in terms of time. Greening the desert might be the best idea but the scale certainly sounds right.