Pedram Hassanzadeh

AI+weather/climate; Prof at UChicago

[Pedram Hassanzadeh @turbulentjet]

Can AI weather models predict out-of-distribution gray swan extremes? We report @PNASNews that the answer is NO for global gray swans, YES for regional ones: AI models can't extrapolate but can learn from similar events in other regions during training! https://t.co/Evqn3E83K5

[LSE USAPP blog @LSEUSAblog]

Listen to Predicting climate change with AI with Tiffany Shaw and @turbulentjet of @UChicago | The Ballpark podcast https://t.co/BHddsvMohP

[Data Science Institute @DSI_UChicago]

The @LaudeInstitute has awarded a $10M multi-year grant to a @UChicago team building AI weather forecasts for low- and middle-income countries

Their work has already reached millions of smallholder farmers across Africa and South Asia, informing them when the rains will come

[Laude Institute @LaudeInstitute]

This summer, @k_tighe and the Laude team headed to @UChicago on our Moonshots // ONE tour. There, @WillettBecca, @ianfoster, @turbulentjet, and Michael Kremer are developing weather forecasting models to help millions of farmers protect their crops and livelihoods. Their approach: build open tools tailored to end users that let partner orgs steer their own stack.

Keep an eye out for more from our visits 🎙️
https://t.co/30bLPlPWkw

[LSE USAPP blog @LSEUSAblog]

Listen to Predicting climate change with AI with Tiffany Shaw and @turbulentjet of @UChicago | The Ballpark podcast https://t.co/BHddsvMohP

[Institute for Climate and Sustainable Growth @UChiClimate]

New study in @ScienceAdvances from @turbulentjet + coauthors shows #AI weather models can predict extreme weather events, even if unprecedented, by learning from similar events in other regions.
https://t.co/Z2XPRW8QmF

@dsi_uchicago

[Pedram Hassanzadeh @turbulentjet]

Work co-led by amazing early-career researchers Amaury Lancelin (ENS-Paris) and Alexander Wikner (UChicago) in collaboration with @Freddy_Bouchet ,@DorianAbbot, Jonathan Weare, Laurent Dubus, and Clément Le Priol.

[Pedram Hassanzadeh @turbulentjet]

Excited to share our new PRL paper, "AI-Boosted Rare Event Sampling to Characterize Extreme Weather," which introduces AI+RES, a framework that hybridizes accurate but expensive physics-based models with fast, skillful AI models using a mathematical algorithm.

[Physical Review Letters @PhysRevLett]

To better understand extreme weather, scientists created a new framework that couples AI weather forecasts with a physics model for rare weather events. The framework boosts sampling of these rare weather events at much lower computational costs: https://t.co/mWSHwWu0wY

[Pedram Hassanzadeh @turbulentjet]

The AI model guides the physics-based model to efficiently sample the rarest "gray swan" weather extremes. The statistics and examples of the events are all generated by the high-fidelity physics-based model.

[Data Science Institute @DSI_UChicago]

Are you passionate about applying AI and machine learning for scientific discovery?

The UChicago Research Computing Center seeks an experienced Computational Scientist – AI/ML Engineer to support computational and AI-driven research!

Apply at https://t.co/iTB8KeYvQ5

[Pedram Hassanzadeh @turbulentjet]

Excited to be a part of this amazing opportunity to accelerate science with AI!

[Laude Institute @LaudeInstitute]

Accelerating Science track: A century of scientific progress in one decade - that's the target.

👑Accelerating the Queen of Sciences, @UCLA
Teaching AI to wonder, conjecture, and discover the way a mathematician does. Every hard science benefits if this works.
@sahaiamit @kaiwei_chang Raghu Meka @VioletNPeng @terrence_tao @WeiWang1973

🌦️Actionable AI Weather Forecasts for Developing Economies, @UChicago
Open-source AI weather forecasting infrastructure for developing economies, with millions of farmers already reached with better monsoon forecasts.
@WillettBecca @ianfoster @turbulentjet Michael Kremer

[Pedram Hassanzadeh @turbulentjet]

Please join us for the workshop "Unpacking AI Weather Emulators: What and How Do They Learn?" at @IMSI_Institute, Sep. 28-Oct. 2. https://t.co/4YpzxOBIZj. Co-organizers: @WillettBecca @atmosbarnes T. Shaw. Supported by IMSI + AI for Climate Initiative @DSI_UChicago @UChiClimate

[Pedram Hassanzadeh @turbulentjet]

Can AI discover something new about turbulence physics? Yes, but only if we didn’t already know Taylor expansion!

Check out our new paper in PRL.

Editors' Suggestion and American Physical Society's Physics: https://t.co/41GJPHz4a2

UChicago News: https://t.co/ScAh18bfoz

[Physical Review Letters @PhysRevLett]

A closed-form closure for 2D turbulence from direct numerical simulation data analyzed with AI tools offers insights into the dynamics of atmospheric and oceanic turbulence
Letter: https://t.co/Y8XJyrHSXq
Synopsis: https://t.co/WdFIhpggFH

[Pedram Hassanzadeh @turbulentjet]

Check out this Nature news story about our recent work on AI and weather extremes, summarizing three papers on "extrapolation", "translocation", and "rare-event sampling" mathematical algorithms coupling AI and physics-based models to efficiently sample the most extreme events.

[nature @Nature]

Until now, AI models have struggled to forecast extreme weather events that they've never seen before

https://t.co/PQzT9NBWzi

[SIAM @TheSIAMNews]

On the SIAM News Blog, Y. Qiang Sun, @turbulentjet, Jonathan Weare, and Dorian Abbot explain how AI #weather models need augmented training data—beyond historical—to accurately predict extreme weather like #tropicalcyclones. Read more here! #SIAMDS25 https://t.co/VK1mmMl0yE

[Mark Suzman @MSuzman]

In India, the rainy season makes or breaks farmers. The quality of their harvest hinges on how much time they have to plan, and as monsoon season becomes more erratic, intense, and geographically variable, their work becomes more difficult.

This is why I’m excited for the @AgriGoI and its latest advancements in using AI to send hyper-local weather forecasts to 38 million farmers across the country. They can predict the start of the rainy season up to 30 days in advance.

Ahead of @Cop30noBrasil, it's hard to overstate how critical agriculture is to protecting and advancing human progress. AI systems like India’s have very low marginal costs and are highly scalable. Every $1 spent on early preparedness saves up to $7 in response costs.

India’s climate resilience is safeguarding crop yields, protecting farmers’ incomes, and securing the future for millions of people. More from @UChiClimate: https://t.co/8eSfdkny10

[Institute for Climate and Sustainable Growth @UChiClimate]

Can better data close the climate survival gap?

In a new podcast episode of Shocked, @amyaharder, @fburlig, @turbulentjet, & @amirjina explore how #AI is transforming weather prediction in places where stakes are highest and resources are few. https://t.co/TjAWTNCsV9

[Karan Jakhar @Jakharkaran]

Glad to have defended my PhD and submitted the third chapter! Can AI discover accurate and stable SGS closure?

Check the preprint: https://t.co/3bi5Lb5vNV
@turbulentjet @usernomeTaken

[Pedram Hassanzadeh @turbulentjet]

@shoyer @amirjina This was an incredible multidisciplinary collaboration, and it was great to work with you and your wonderful colleagues to accelerate the translation of cutting-edge science to real-world impact!

[Institute for Climate and Sustainable Growth @UChiClimate]

The AI for Climate Initiative (AICE), a joint program of our Institute and @DSI_UChicago, aims to leverage the power of #AI and #MachineLearning to pioneer new models for climate and weather prediction.

Explore AICE's recent accomplishments: https://t.co/HqNe4Rbm7X

[Gokul Tamilselvam @Gokul46978057]

Monsoon Trough (Dynamic Low) gets established for the first time this monsoon season:
-------------------------------------
The Monsoon Trough or the Dynamic low is getting established for the first time this monsoon season over the Indo Gangetic Plains. The mechanism is an indication of the arrival of moist monsoonal air mass and related physical processes over the northern boundary of the Indian subcontinent. With this, the advancing phase of summer monsoon rains is nearing completion. The latest update from IMD indicates that the monsoon isochrone has covered most parts of Uttar Pradesh and parts of north western Himalayas in the past 24 hours. Satellite observations indicate formation of persistent low to mid level cloud layers on a large scale to the south of the Monsoon Trough axis. Rainfall observations over the core monsoon zone show that the region has been experiencing huge positive rainfall anomalies in the past 4-5 days with area averaged rainfall being close to + 2 standard deviations. Meanwhile, NWP model forecasts continue to indicate an early advance of Monsoon over the north western regions of the subcontinent.
Images sourced from @Windycom
@eumetsat @Indiametdept @ClimateImd . Analysis and interpretations here are mine alone in an attempt to understand tropical weather and related processes. For official forecasts please follow the respective national weather forecasting agencies @Indiametdept