You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This new [article](https://doi.org/10.1175/JCLI-D-24-0258.1) explores the predictability of Antarctic sea ice, which varies by region and is influenced by ocean and atmospheric conditions. Using climate models, Marika Holland and co-authors, confirm that **sea ice changes can be predicted months in advance, with cycles of predictability linked to ice growth and retreat**: it remains predictable during growth, loses predictability when melting, and regains it as it reforms. Ocean temperature patterns near the ice edge play a key role, with variations across different regions. These ice changes also influence marine ecosystems by affecting light availability. Understanding these patterns can **improve climate predictions and support the management of the Southern Ocean’s biodiversity**.
11
+
This new [article](https://doi.org/10.1175/JCLI-D-24-0258.1) explores the predictability of Antarctic sea ice, which varies by region and is influenced by ocean and atmospheric conditions. Using climate models, Marika Holland and co-authors, confirm that **sea ice changes can be predicted months in advance, with cycles of predictability linked to ice growth and retreat**: it remains predictable during growth, loses predictability when melting, and regains it as it reforms. Ocean temperature patterns near the ice edge play a key role, with variations across different regions. These ice changes also influence marine ecosystems by affecting light availability. Understanding these patterns can **improve climate predictions and support the management of the Southern Ocean’s biodiversity**.
Autoregressive surrogate models (emulators) make fast predictions for dynamical systems but become unstable over time due to error buildup. This [study](https://doi.org/10.48550/arXiv.2503.18731), led by Chris Pedersen, introduces thermalization, a technique leveraging diffusion models to adaptively correct errors during inference. **By stabilizing predictions, it extends emulated rollouts of chaotic and turbulent systems by several orders of magnitude**. This breakthrough application of diffusion models enhances the utility of emulators, and can be applied to autoregressive models across science and engineering.
11
+
Autoregressive surrogate models (emulators) make fast predictions for dynamical systems but become unstable over time due to error buildup. This [study](https://doi.org/10.48550/arXiv.2503.18731), led by Chris Pedersen, introduces thermalization, a technique leveraging diffusion models to adaptively correct errors during inference. **By stabilizing predictions, it extends emulated rollouts of chaotic and turbulent systems by several orders of magnitude**. This breakthrough application of diffusion models enhances the utility of emulators, and can be applied to autoregressive models across science and engineering.
New tools like kilometer-scale modeling, parameter tuning, and AI/machine learning are reshaping the climate modeling field, sparking debate on the best path forward. Five internationally renowned female climate scientists, including Laure Zanna, sign this [perspective piece](https://doi.org/10.1038/s41612-025-00955-8) in Nature climate and atmospheric science. Their analysis draws lessons from past research to guide the future of climate modeling. The conclusion: **the future of climate modeling depends on embracing diverse tools and methodologies to drive meaningful progress!**
12
+
New tools like kilometer-scale modeling, parameter tuning, and AI/machine learning are reshaping the climate modeling field, sparking debate on the best path forward. Five internationally renowned female climate scientists, including Laure Zanna, sign this [perspective piece](https://doi.org/10.1038/s41612-025-00955-8) in Nature climate and atmospheric science. Their analysis draws lessons from past research to guide the future of climate modeling. The conclusion: **the future of climate modeling depends on embracing diverse tools and methodologies to drive meaningful progress!**
A new framework from NCAR, called CREDIT (Community Research Earth Digital Intelligence Twin), is making it easier for researchers to **develop and test AI-based weather prediction models**. Introduced in this [paper](https://doi.org/10.1038/s41612-025-01125-6), CREDIT supports flexible model design and training, helping **address key challenges in AI-based forecasting**. Using this platform, researchers introduced WXFormer, a novel model that outperforms traditional forecasting systems like ECMWF’s IFS on 10-day forecasts, while being much more computationally efficient. CREDIT aims to accelerate innovation and collaboration in AI-driven weather prediction. **William Chapman** and **Judith Berner** contributed to this research.
12
+
A new framework from NCAR, called CREDIT (Community Research Earth Digital Intelligence Twin), is making it easier for researchers to **develop and test AI-based weather prediction models**. Introduced in this [paper](https://doi.org/10.1038/s41612-025-01125-6), CREDIT supports flexible model design and training, helping **address key challenges in AI-based forecasting**. Using this platform, researchers introduced WXFormer, a novel model that outperforms traditional forecasting systems like ECMWF’s IFS on 10-day forecasts, while being much more computationally efficient. CREDIT aims to accelerate innovation and collaboration in AI-driven weather prediction. **William Chapman** and **Judith Berner** contributed to this research.
**Carlos Fernandez-Granda** new book is out! Published by [Cambridge University Press](https://www.cambridge.org/core/books/probability-and-statistics-for-data-science/CC7DC7E53ED92074008803C96A67620B), the book is a self-contained **guide to the two pillars of data science, probability theory, and statistics**. The materials, which include 200 exercises with solutions, 102 Python notebooks using 23 real-world datasets, 115 YouTube videos with slides, and a free preprint, can be found at this **[website](https://www.ps4ds.net/)**.
**Gregory** et al. present in this [preprint](https://doi.org/10.48550/arXiv.2505.18328), a hybrid modeling approach that **integrates machine learning (ML) into the GFDL SPEAR climate model to correct sea ice biases in real time**. Two versions are tested: one that includes coupled feedbacks (HybridCPL) and one that does not (HybridIO). HybridCPL **significantly improves Arctic and Antarctic sea ice forecasts on seasonal and subseasonal timescales**. In contrast, HybridIO performs poorly due to unanticipated feedbacks. These results highlight the **importance of training ML models within coupled climate systems** for reliable predictions.
12
+
**Gregory** et al. present in this [preprint](https://doi.org/10.48550/arXiv.2505.18328), a hybrid modeling approach that **integrates machine learning (ML) into the GFDL SPEAR climate model to correct sea ice biases in real time**. Two versions are tested: one that includes coupled feedbacks (HybridCPL) and one that does not (HybridIO). HybridCPL **significantly improves Arctic and Antarctic sea ice forecasts on seasonal and subseasonal timescales**. In contrast, HybridIO performs poorly due to unanticipated feedbacks. These results highlight the **importance of training ML models within coupled climate systems** for reliable predictions.
**Perezhogin** et al. propose in this [preprint](https://doi.org/10.48550/arXiv.2505.08900), a **physics-informed neural network to improve the generalization of data-driven mesoscale eddy parameterizations** in ocean models. By applying local input-output scaling based on dimensional analysis, **their method adapts to different grid resolutions and depths**. This approach enhances energy representation and affects biases in both idealized and global ocean simulations. The scaling framework is broadly applicable and robust across configurations. Results show **competitive performance compared to traditional parameterizations**.
12
+
**Perezhogin** et al. propose in this [preprint](https://doi.org/10.48550/arXiv.2505.08900), a **physics-informed neural network to improve the generalization of data-driven mesoscale eddy parameterizations** in ocean models. By applying local input-output scaling based on dimensional analysis, **their method adapts to different grid resolutions and depths**. This approach enhances energy representation and affects biases in both idealized and global ocean simulations. The scaling framework is broadly applicable and robust across configurations. Results show **competitive performance compared to traditional parameterizations**.
This **[APS news article](https://www.aps.org/apsnews/2025/06/ai-could-shape-climate-science?utm_source=jun-amn&utm_medium=email&utm_campaign=jun-amn)** highlights **how AI is reshaping climate science**, offering faster, smarter ways to model Earth’s complex systems. It features highlights from the Global Physics Summit, where researchers presented how machine learning is accelerating simulations, uncovering new physics, and helping build more precise climate models, all while keeping physics at the core. In particular, it spotlights interview exerts with Laure Zanna presenting **[Samudra](https://doi.org/10.1029/2024GL114318), the M²LInES created AI emulator**.
12
+
This **[APS news article](https://www.aps.org/apsnews/2025/06/ai-could-shape-climate-science?utm_source=jun-amn&utm_medium=email&utm_campaign=jun-amn)** highlights **how AI is reshaping climate science**, offering faster, smarter ways to model Earth’s complex systems. It features highlights from the Global Physics Summit, where researchers presented how machine learning is accelerating simulations, uncovering new physics, and helping build more precise climate models, all while keeping physics at the core. In particular, it spotlights interview exerts with Laure Zanna presenting **[Samudra](https://doi.org/10.1029/2024GL114318), the M²LInES created AI emulator**.
The **[Ai2 Climate Modeling](https://allenai.org/climate-modeling)** team has released a new version of their climate emulator. In this **[latest release](https://github.com/ai2cm/ace/releases/tag/2025.7.0), Samudra**, the AI global ocean emulator developed by M²LInES, is now **integrated into Ai2's full model framework**. We encourage you to check out the **[Ai2 codebase](https://github.com/ai2cm/ace)** and Samudra's **[original code](https://github.com/m2lines/Samudra)** to create your own ocean simulations.
12
+
The **[Ai2 Climate Modeling](https://allenai.org/climate-modeling)** team has released a new version of their climate emulator. In this **[latest release](https://github.com/ai2cm/ace/releases/tag/2025.7.0), Samudra**, the AI global ocean emulator developed by M²LInES, is now **integrated into Ai2's full model framework**. We encourage you to check out the **[Ai2 codebase](https://github.com/ai2cm/ace)** and Samudra's **[original code](https://github.com/m2lines/Samudra)** to create your own ocean simulations.
0 commit comments