Quantifying the Influence of Climate on Storm Activity Using Machine Learning
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_ | 1866915345454858240 |
|---|---|
| author | Hadas, Or Kaspi, Yohai |
| author_facet | Hadas, Or Kaspi, Yohai |
| contents | Extratropical storms shape midlatitude weather and vary due to the slowly evolving climate and the rapid changes in synoptic conditions. While the influence of each factor has been studied extensively, their relative importance remains unclear. Here, we quantify the climate's relative importance in mean storm activity and individual storm development using 84 years of ERA-5 data and convolutional neural networks. We find that the constructed model predicts over 90% of the variability in the mean storm activity. However, a similar model predicts about a third of the variability in individual storm properties, such as maximum intensity, showing their variability is dominated by synoptic conditions. Isolating the impact of present-day climate change on individual storms shows it contributes to about 0.1% for storm-intensity variability, whereas its contribution to storms' heat-anomaly variability is over three times greater, highlighting that focusing on variables directly tied to global warming offers a clearer attribution pathway. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_20521 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Quantifying the Influence of Climate on Storm Activity Using Machine Learning Hadas, Or Kaspi, Yohai Atmospheric and Oceanic Physics Extratropical storms shape midlatitude weather and vary due to the slowly evolving climate and the rapid changes in synoptic conditions. While the influence of each factor has been studied extensively, their relative importance remains unclear. Here, we quantify the climate's relative importance in mean storm activity and individual storm development using 84 years of ERA-5 data and convolutional neural networks. We find that the constructed model predicts over 90% of the variability in the mean storm activity. However, a similar model predicts about a third of the variability in individual storm properties, such as maximum intensity, showing their variability is dominated by synoptic conditions. Isolating the impact of present-day climate change on individual storms shows it contributes to about 0.1% for storm-intensity variability, whereas its contribution to storms' heat-anomaly variability is over three times greater, highlighting that focusing on variables directly tied to global warming offers a clearer attribution pathway. |
| title | Quantifying the Influence of Climate on Storm Activity Using Machine Learning |
| topic | Atmospheric and Oceanic Physics |
| url | https://arxiv.org/abs/2504.20521 |