Forecasting Fails: Unveiling Evasion Attacks in Weather Prediction Models
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_ | 1866912755766788096 |
|---|---|
| author | Arif, Huzaifa Chen, Pin-Yu Gittens, Alex Diffenderfer, James Kailkhura, Bhavya |
| author_facet | Arif, Huzaifa Chen, Pin-Yu Gittens, Alex Diffenderfer, James Kailkhura, Bhavya |
| contents | With the increasing reliance on AI models for weather forecasting, it is imperative to evaluate their vulnerability to adversarial perturbations. This work introduces Weather Adaptive Adversarial Perturbation Optimization (WAAPO), a novel framework for generating targeted adversarial perturbations that are both effective in manipulating forecasts and stealthy to avoid detection. WAAPO achieves this by incorporating constraints for channel sparsity, spatial localization, and smoothness, ensuring that perturbations remain physically realistic and imperceptible. Using the ERA5 dataset and FourCastNet (Pathak et al. 2022), we demonstrate WAAPO's ability to generate adversarial trajectories that align closely with predefined targets, even under constrained conditions. Our experiments highlight critical vulnerabilities in AI-driven forecasting models, where small perturbations to initial conditions can result in significant deviations in predicted weather patterns. These findings underscore the need for robust safeguards to protect against adversarial exploitation in operational forecasting systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_08832 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Forecasting Fails: Unveiling Evasion Attacks in Weather Prediction Models Arif, Huzaifa Chen, Pin-Yu Gittens, Alex Diffenderfer, James Kailkhura, Bhavya Machine Learning With the increasing reliance on AI models for weather forecasting, it is imperative to evaluate their vulnerability to adversarial perturbations. This work introduces Weather Adaptive Adversarial Perturbation Optimization (WAAPO), a novel framework for generating targeted adversarial perturbations that are both effective in manipulating forecasts and stealthy to avoid detection. WAAPO achieves this by incorporating constraints for channel sparsity, spatial localization, and smoothness, ensuring that perturbations remain physically realistic and imperceptible. Using the ERA5 dataset and FourCastNet (Pathak et al. 2022), we demonstrate WAAPO's ability to generate adversarial trajectories that align closely with predefined targets, even under constrained conditions. Our experiments highlight critical vulnerabilities in AI-driven forecasting models, where small perturbations to initial conditions can result in significant deviations in predicted weather patterns. These findings underscore the need for robust safeguards to protect against adversarial exploitation in operational forecasting systems. |
| title | Forecasting Fails: Unveiling Evasion Attacks in Weather Prediction Models |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2512.08832 |