In weather forecasting, traditional deterministic local atmosphere-driven models (NWP) rely on physical equations to simulate atmospheric behavior.
Reanalysis models combine historical observations with modeling techniques to create consistent, long-term datasets. They often achieve SOTA performance in evaluations but are less practical for real-time applications. In contrast, observation-based models directly utilize current data, offering greater practicality despite lower accuracy.
Additionally, global models excel at capturing large-scale atmospheric patterns, while local models focus on finer, region-specific predictions, catering to localized needs.
Weather Forecast Models
Weather Forecast Notion