Predicting Beyond Training Data via Extrapolation versus Translocation: AI Weather Models and Dubai's Unprecedented 2024 Rainfall
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914302445748224 |
|---|---|
| author | Sun, Y. Qiang Hassanzadeh, Pedram Shaw, Tiffany Pahlavan, Hamid A. |
| author_facet | Sun, Y. Qiang Hassanzadeh, Pedram Shaw, Tiffany Pahlavan, Hamid A. |
| contents | Artificial intelligence (AI) models have transformed weather forecasting, but their skill for gray swan extremes is unclear. Here, we analyze GraphCast, AIFS, and FuXi forecasts of the unprecedented 2024 Dubai storm, which had twice the training set's highest rainfall in that region. Remarkably, GraphCast and AIFS accurately forecast this event up to 8 days ahead. FuXi forecasts the event, but underestimates the rainfall. Fine-tuning and receptive field analyses suggest that these models' success stems from "translocation": learning from comparable/stronger dynamically similar events in other regions during training. Evidence of "extrapolation" (learning from weaker events) is not found. Even events within the global distribution's tail are poorly forecasted, which is not just due to data imbalance (generalization error) but also spectral bias (optimization error). These findings demonstrate the potential of AI models to forecast regional gray swans and the opportunity to improve them through understanding the mechanisms behind their successes/limitations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_10241 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Predicting Beyond Training Data via Extrapolation versus Translocation: AI Weather Models and Dubai's Unprecedented 2024 Rainfall Sun, Y. Qiang Hassanzadeh, Pedram Shaw, Tiffany Pahlavan, Hamid A. Atmospheric and Oceanic Physics Artificial intelligence (AI) models have transformed weather forecasting, but their skill for gray swan extremes is unclear. Here, we analyze GraphCast, AIFS, and FuXi forecasts of the unprecedented 2024 Dubai storm, which had twice the training set's highest rainfall in that region. Remarkably, GraphCast and AIFS accurately forecast this event up to 8 days ahead. FuXi forecasts the event, but underestimates the rainfall. Fine-tuning and receptive field analyses suggest that these models' success stems from "translocation": learning from comparable/stronger dynamically similar events in other regions during training. Evidence of "extrapolation" (learning from weaker events) is not found. Even events within the global distribution's tail are poorly forecasted, which is not just due to data imbalance (generalization error) but also spectral bias (optimization error). These findings demonstrate the potential of AI models to forecast regional gray swans and the opportunity to improve them through understanding the mechanisms behind their successes/limitations. |
| title | Predicting Beyond Training Data via Extrapolation versus Translocation: AI Weather Models and Dubai's Unprecedented 2024 Rainfall |
| topic | Atmospheric and Oceanic Physics |
| url | https://arxiv.org/abs/2505.10241 |