Policy makers and community leaders need state-of-the-art tools to anticipate crises arising from our rapidly evolving climate. An answer to this challenge is coming from a marriage of IBM’s latest AI technology with NASA’s world leading observation data on both the Earth and the sun. Or in layman’s terms, it’s about AI analyzing and interpreting the vast mass of data collected by NASA in recent years and then making that knowledge fully accessible through open-source platforms.
Our planet is becoming a lot less predictable. Municipalities, emergency services, scientists, and weather forecasters have been caught on the back foot by events and phenomena they would have once hoped to anticipate. The challenge is to speed up the process of scientific discovery and then, using AI, empower communities to act on this information. IBM and NASA’s AI model, named Prithvi, for the Sanskrit word for Earth, was trained on high resolution Earth observation data. It has the ability to generate maps of floods, fires, and other landscape changes that reveal current trends and reveal likely future events.
Take for example the problem of cyanobacteria, a microscopic, photosynthesizing form of algae that can be found in most bodies of water. Much of the time, it’s harmless. But give it lashings of sunlight, stagnant water conditions, and pollution from fertilizers, and the situation changes. Exploding algal blooms pose a danger to aquatic life, wetland animals, and human beings. The wider economic impact could be felt in sectors like tourism, so this comes with a potentially big price tag.
Enter NASA with its Earth observation data coupled with IBM’s cutting-edge AI technology. As a foundation model, Prithvi can be fine-tuned for specific problems, like detecting and analyzing harmful algal blooms. Applying machine learning and AI to satellite imagery makes it easier to identify areas that could have high concentrations of cyanobacteria. It complements human analysis at a much faster pace and on a huge scale.
Algal blooms in the oceans, and other marine-water-related risks, are being covered by a new fine-tuned AI model, thanks to IBM’s work with the Plymouth Marine Institute and Exeter University. Two thirds of the Earth is covered by oceans, and, like outer space, we are still on a journey of discovery to understand these huge bodies of water. The data is being collected by the Sentinel-3 mission, which is measuring sea surface topography, temperature, and color to improve our future forecasting and monitoring.
NASA not only has eyes trained on planet Earth, but also collects data about the sun. What happens on the fiery ball at the center of our solar system can be incredibly disruptive for our modern infrastructure and communications systems. A solar storm could create a major disturbance in Earth’s magnetic field, causing large-scale blackouts and power outages. It could even endanger our astronauts in space. As early as September 1859, an intense geomagnetic storm, dubbed the Carrington Event, caused fires in telegraph stations. Today, everything from satellites and aviation to smartphones and power grids might be taken down.
.gif)
Lloyd’s of London, the world’s leading insurance market, published some risk scenarios in March 2025. It calculated that the global economy could be exposed to losses of $2.4 trillion over a five-year period from a hypothetical solar storm. North America and Europe would be the most severely impacted. Juan Bernabe-Moreno, Director of IBM Research Europe, UK, and Ireland, explains the importance of Surya:
"Just as we work to prepare for hazardous weather events, we need to do the same for solar storms. Surya gives us unprecedented capability to anticipate what's coming and is not just a technological achievement, but a critical step toward protecting our technological civilization from the star that sustains us."
Building on the harnessing of Earth observation data to artificial intelligence, IBM and NASA have introduced Surya, an open-source foundation model trained on NASA’s heliophysics data. The AI will have no shortage of solar data to absorb, as NASA’s Solar Dynamics Observatory has been watching the sun for the past 15 years. This will help us understand the dynamics of the sun and predict space weather.
It's great having access to all this knowledge, with AI learning from this amazing data to predict future trends, but how to get this into the hands of those who need it? In the past, creating the practical application tools necessitated a high level of computer knowledge. But times are changing. Access is being democratized.
Central to the partnership between IBM and NASA is an open-source ethos. For this reason, Surya has been released on Hugging Face, a huge open-source community building tools, machine learning models, and platforms for working with AI. Surya joins the already existing Prithvi family of AI foundation models for planet Earth, making AI accessible for open science users, startups, and enterprises. This approach is simplifying model training and deployment.
Juan Bernabe-Moreno adds:“We are lowering the barrier to understanding this valuable data that can be used by many different sectors, from insurance to agriculture and habitat restoration. The open-source approach allows users to download the model and fine-tune it to solve their own specific problems.”
This leap forward in AI is much needed. By partnering to unlock insights from valuable data, NASA and IBM are offering more people the opportunity to better understand the Earth and solar system we’re all a part of.
For more information about IBM, visit here.
