Skip to content

Extreme Weather Event Forecasting

Home / Open Questions / Extreme Weather Event Forecasting

Background: Deep learning-based weather forecasting models often exhibit decreased predictive accuracy and robustness during extreme weather events.

Question / Future Work: Research is needed to improve the sensitivity and reliability of deep learning architectures when forecasting rare or high-impact weather events, which are typically underrepresented in standard training datasets.

Why It Matters: Improving performance on extreme weather events is vital for the societal utility of forecasting systems, specifically for disaster preparedness and climate risk management.

Evidence: The current graph structure is static, and the performance under rare extreme weather conditions remains to be further improved.

Metadata & Links