A climate model developed by researchers in Seattle and San Diego is set to revolutionize climate projections, delivering results 25 times faster than current state-of-the-art models. By combining generative AI algorithms, like those behind DALLยทE, with physics-based data, the new model – named Spherical DYffusion – can predict climate patterns over the next century in just 25 hours, a task that traditionally takes weeks.

This innovation promises to reduce the computational cost of climate simulations while maintaining high accuracy, potentially transforming the way scientists and policymakers analyze long-term climate scenarios.


Ioana Patringenaru | University of California – San Diego – The algorithms behind generative AI tools like DallE, when combined with physics-based data, can be used to develop better ways to model the Earthโ€™s climate. Computer scientists in Seattle and San Diego have now used this combination to create a model that is capable of predicting climate patterns over 100 years 25 times faster than the state of the art.

Rose Yu Climate Modeling Rotator
From left: Rose Yu, a faculty member in the UC San Diego Department of Computer Science and Engineering, and Ph.D. student Salva Ruhling Cachay examine data. Credit: David Baillot | University of California – San Diego

Specifically, the model, called Spherical DYffusion (.pdf), can project 100 years of climate patterns in 25 hoursโ€“a simulation that would take weeks for other models. In addition, existing state-of-the-art models need to run on supercomputers. This model can run on GPU clusters in a research lab.

โ€œData-driven deep learning models are on the verge of transforming global weather and climate modeling,โ€ the researchers from the University of California San Diego and the Allen Institute for AI, write.

The research team is presenting their work at the NeurIPS conference 2024, Vancouver, Canada.

Climate simulations are currently very expensive to generate because of their complexity. As a result, scientists and policymakers can only run simulations for a limited amount of time and consider only limited scenarios.

One of the researchersโ€™ key insights was that generative AI models, such as diffusion models, could be used for ensemble climate projections. They combined this with a Spherical Neural Operator, a neural network model designed to work with data on a sphere.

The resulting model starts off with knowledge of climate patterns and then applies a series of transformations based on learned data to predict future patterns.

โ€œOne of the main advantages over a conventional diffusion model (DM) is that our model is much more efficient. It may be possible to generate just as realistic and accurate predictions with conventional DMs but not with such speed,โ€ the researchers write.

In addition to running much faster than state of the art, the model is also nearly as accurate without being anywhere near as computationally expensive.

There are some limitations to the model that researchers aim to overcome in its next iterations, such as including more elements in their simulations. Next steps include simulating how the atmosphere responds to CO2.

โ€œWe emulated the atmosphere, which is one of the most important elements in a climate model,โ€ said Rose Yu, a faculty member in the UC San Diego Department of Computer Science and Engineering and one of the paperโ€™s senior authors.

The work stems from an internship that one of Yuโ€™s Ph.D. students, Salva Ruhling Cachay, did at the Allen Institute for AI (Ai2).

More information:
Rose Yu, Salva Ruhling Cachay, Brian Henn, Oliver Watt-Meyer and Christopher S. Bretherton ‘Probabilistic Emulation of a Global Climate Model with Spherical DYffusion’ (2024) – (Read .pdf)

Article Source:
Press Release/Material by University of California – San Diego
Featured image: A snapshot from the climate simulation generated by the Spherical DYffusion model. Credit: University of California San Diego

Image: 3D-render globe (s. monsoons)
Cyclone kills 14 people in France’s Mayotte: security sourceNews

Cyclone kills 14 people in France’s Mayotte: security source

Paris, France | AFP - Cyclone Chido killed at least 14 people on the French Indian Ocean island territory of Mayotte, a security source told…
SourceSourceDecember 15, 2024 Full article
Satellite Image: flooded areas near the town of Aรฑover de Tajo in Castilla-La Mancha (in red tones), Spain
Image of the day: Flooding in Spain after Storm MartinhoNews

Image of the day: Flooding in Spain after Storm Martinho

Storm Martinho has triggered severe flooding across Spain, forcing evacuations and emergency measures in several regions. In Mejorada del Campo, rising waters from the Jarama…
Muser NewsDeskMuser NewsDeskMarch 25, 2025 Full article
Satellite image: Dokkum, Netherlands (s. snow)
Image of the day: Dokkum, a snowy view of FrieslandNews

Image of the day: Dokkum, a snowy view of Friesland

Dokkum, a historic city in Friesland, Netherlands, lies near the Wadden Sea, a UNESCO World Heritage site known for its tidal flats and barrier islands.…
Muser NewsDeskMuser NewsDeskMarch 1, 2025 Full article