Tag
SimCast-S2S is a generative latent-diffusion framework for probabilistic subseasonal precipitation forecasting that leverages transfer learning from climate simulations to outperform deep learning baselines and compete with operational systems.