Focal Interest
Focal Interest

Numerical weather prediction

  1. Self-supervised graph model improved multi-horizon weather forecasts
  2. ICL Characterization of Climate Foundation Models: When Can Transformers Learn Weather and Climate?
    Theoretical analysis explains why climate foundation models succeed at field prediction but fail at extreme event detection through in-context learning complexity.
  3. Kilometre-scale simulations improved extreme rainfall forecasts in eastern Qinghai
    Kilometre-scale convection-permitting simulations significantly improve extreme precipitation forecasting accuracy in Qinghai's eastern valleys by better representing valley circulation patterns.
  4. Weather regimes change short-term solar forecast error
    Study reveals how North Atlantic weather regimes significantly influence satellite-based solar forecast accuracy, with seasonal variations up to 20% in error magnitude affecting renewable energy.