This AI model ‘studied’ physics — and learnt to forecast extreme weather Combining artificial intelligence with a conventional climate model can predict heatwaves faster than the standard model alone.

#AI models do pretty darn well at forecasting the #weather. But they've struggled to predict weather extremes (e.g. strong storms, heat waves, heavy rains).

I wrote about one of the many teams of scientists looking to change that. 🧪 #AGU25 #wx @pedramh.bsky.social

www.nature.com/articles/d41...

Researchers at the AI for Climate Initiative (AICE) are leveraging AI and machine learning to revolutionize how we predict the weather

Learn how they’re working to explore extreme scenarios and democratize access to state-of-the-art forecasts

🧵

Here is the NotebookLM podcast summarizing the work hassanzadehresearch.blogs.rice.edu/files/2025/0...

The Dubai event had 2x more rain than the max rainfall in the region in the training set

Predicting Beyond Training Data via Extrapolation versus Translocation: AI Weather Models and Dubai's Unprecedented 2024 Rainfall Artificial intelligence (AI) models have transformed weather forecasting, but their skill for gray swan extreme events is unclear. Here, we analyze GraphCast and FuXi forecasts of the unprecedented 20...

Work with @Qiang Sun, @Mohsen Zand, @ashesh6810.bsky.social, @Jon Weare & @Dorian Abbot. Also, see our new preprint for a more in-depth analysis of this question using the 2024 Dubai rainfall event arxiv.org/abs/2505.10241

Can AI weather models predict out-of-distribution gray swan tropical cyclones? | PNAS Predicting gray swan weather extremes, which are possible but so rare that they are absent from the training dataset, is a major concern for AI wea...

Can AI weather models predict out-of-distribution gray swan extremes? We report @pnas.org that the answer is NO for global gray swans, YES for regional ones: AI models can't extrapolate from weaker events but can learn from similar events in other regions during training! doi.org/10.1073/pnas...

Propagation and Periodicity of Mars's Northern Annular Mode Modulates the Dust Cycle Mars's Northern Annular Mode (MNAM), diagnosed from a reanalysis dataset, propagates with a 150-day period, independent of the season Like Earth's, MNAM arises from feedbacks between the two lead...

Cool Mars science thread to celebrate Musk’s defeat yesterday: Mars has irregular global dust storms—sometimes x2 a year, sometimes none for a decade. I, with @pedramh.bsky.social +others, looked at potentially why in @agu.org. ☄️🔭🧪 #planetsci #Mars agupubs.onlinelibrary.wiley.com/doi/full/10....

Machine learning for the physics of climate - Nature Reviews Physics Artificial intelligence techniques, specifically machine learning, are being increasingly applied to climate physics owing to the growing availability of big data and increasing computational power. T...

I am very glad to share this review paper: 📃 "Machine Learning for the Physics of Climate." As stated in the title, the review targets the opportunities and challenges of machine learning for climate physics. www.nature.com/articles/s42...
@pedramh.bsky.social

Research scientist position on AI weather forecasting at UChicago's AI for Climate (AICE) initiative to work with me, Amir Jina, and Michael Kremer. Will work across our new Human-Centered Weather Forecasting Initiative + @dsi-uchicago.bsky.social + Climate Institute

Apply by Dec. 11: dil.uchicago.edu/senior-direc...

Director works closely with PIs (Michael Kremer, Amir Jina & me). Unique opportunity to advance science and its applications in a rapidly evolving field as a part of UChicago’s extensive efforts around AI+climate: AI for Climate Initiative, Climate Institute, @dsi-uchicago.bsky.social

Sr Director position (weather forecast) at UChicago's new Human-Centered Weather Forecasting Initiative. Unique opportunity to lead an interdisciplinary team to generate AI- & physics-based forecasts, particularly to support communities most vulnerable to climate variability.

David Keith and I are looking for a postdoc to work on leveraging the recent advances in ML weather models to develop ML emulators for solar geoengineering at the University of Chicago. Details below geosci.uchicago.edu/postdoc-posi...