> supercomputer to predict the chaotic effects.

This might be the hardest part of your proposal to achieve. Chaotic systems are fundamentally difficult to simulate, not just because of their differential equations but our lack of perfect knowledge of their initial conditions. Blanket solar reduction seems more feasible?