Source-backed Froggit

Advances in weather prediction or atmospheric modeling

## Key Findings - Recent advances in weather prediction and atmospheric modeling announced in October 2026 include: - 1. **AI models rivaling traditional dynamical models**: Deep learning autoregressive models trained on atmospheric reanalyses now achieve forecast quality comparable to physics-based dynamical models across nowcasting, medium-range, and subseasonal-to-seasonal lead times, while producing well-calibrated ensemble forecasts at reduced computational cost (Source [2]: https://arxiv.org/abs/2610.09770v1). - 2. **Data-driven subseasonal soil moisture forecasting for flash drought prediction**: Skillful data-driven approaches demonstrate subseasonal soil moisture forecasting capabilities, addressing t...

Source: https://arxiv.org/abs/2610.02511v1