> "Since the early cloud-seeding experiments reported by Kraus and Squires (1947), wherein it was hypothesized that the observed “explosive growth” of cumuli was the result of artificially induced glaciation of the cloud, there has been intensive interest and research in the relationships between evolution of the ice phase and the dynamics of cumuli."

As I understand it the idea is that a small amount of AGI or can cause nucleation which might trigger rainfall:

> "Thus, the latent heat liberated during the freezing of supercooled droplets and the vapor deposition growth of ice particles can augment the diminishing cloud buoyancy production during condensation growth of cloud droplets. In some cases, the additional buoyancy may be sufficient to allow a cloud to penetrate weak stable layers or to survive the entrainment of relatively dry environmental air."

The idea in cloud seeding is to hit a cloud 'on the verge of rain and push it over the boundary, but it doesn't seem straightforward at all to determine if a given cloud on that boundary will produce rain or not, as in a '50% chance of rain' forecast. It does indeed seem to be one of those chaotic boundary condition issues:

> "This demonstrates that there exists a very intimate, nonlinear coupling between buoyancy production by glaciation of a cloud and the evolution of the microstructure of the cloud, and the evolving cloud motion field."

https://www.sciencedirect.com/topics/earth-and-planetary-sci...