Traditional models still 'outperform AI' for extreme weather forecasts

The article discusses a study that evaluates the performance of traditional physics-based weather models against newer artificial intelligence (AI) models in forecasting extreme weather events. The key points are:

1. Traditional Models Outperform AI for Extreme Weather:

   • Traditional, physics-based weather models still outperform AI models when it comes to predicting unprecedented or record-breaking extreme weather events.

2. Evaluation of AI Models:

   • Three older generation deterministic AI models (deterministic means they simulate only one possible future outcome) were tested against traditional models.
   • These AI models struggled with accurately forecasting extremes, which is a critical shortcoming for operational forecasts.

3. Advancements in AI Weather Forecasting:

   • The field of AI weather forecasting is rapidly evolving.
   • Newer probabilistic models have emerged that claim to better capture extremes by simulating multiple possible future states rather than just one deterministic outcome.

4. Future Research Directions:

   • Evaluating the newer generation of probabilistic AI models for their ability to forecast extreme events.
   • Developing hybrid models that combine aspects of traditional physics-based forecasts with AI models, which could be a "best-of-both-worlds" approach.

5. **Importance of

Read the full article at Carbon Brief

---

Want to create content about this topic? Use Nemati AI tools to generate articles, social posts, and more.