NASA and IBM have presented the most empowering open-source artificial intelligence foundation model, Surya, that is designed to forecast solar weather and secure the technology against space-related dangers. Unlike the traditional method that utilizes the raw Web data alone, AINIPS is the first of its kind to apply the concept of artificial intelligence in the classification of solar images and in space weather forecast studies. The model has been developed using nine years of observation data in high resolution that NASA space researchers were able to get using the Solar Dynamics Observatory, making the possibilities of predicting solar storms unprecedented.
Transformational technology: How Surya makes solar weather forecasts better than ever
Surya breaks previous constraints in heliophysics forecasting because it is trained on the largest, carefully curated high-resolution heliophysics dataset ever created. IBM newsroom states that the model had a 16 percent increase in the accuracy of solar-geophysical flare classification over the other methods, which is a significant step in predictive abilities.
This was a hugely technically ambitious job since the data, in the form of solar images, were 10 times larger than standard training sets in the field of artificial intelligence. This necessitated a multi-architectural solution, which was designed on a large scale that had to be highly efficient. The effect is a model with higher spatial resolution than previous models, able to capture solar scale at rates and contexts not previously visible in large-scale AI training workflows.
That is the way Surya can give visual solar flare forecasts
In addition to classical classification problems, Surya can extrapolate visually to predict solar flares, yielding high-resolution images of the places where flares are predicted to be visible up to two hours before they actually appear. This ability is considered a breakthrough in forecasting space weather, hence providing researchers and operators with some necessary lead time in which to prepare against a potentially disruptive solar event.
Economic effect: Securing trillions of infrastructure on the planet
The economic consequences of predicting the weather make the union of sunshine something too big to comprehend. In a systemic risk scenario developed by Lloyds, the world economy would lose 2.4 trillion over five years due to the occurrence of extreme space weather, and an expected loss of 17 billion on an imaginary solar storm.
Juan Bernabe-Moreno, who holds the position of Director of IBM Research Europe, stated the importance of this technology as being enormous since such a technology allows predicting solar storms as never before. As an example of the application, it can be used in weather forecasting in space, so that the experts can make preparations for the occurrence of events that can damage the technological infrastructure on Earth, similar to weather forecasting on Earth.
Science: Democratization of space weather research
Surya is one among a greater partnership between IBM and NASA, aiming to use AI to explore the planetary and solar systems. The Prithvi family joins a range of foundation models, geospatial, and weather prediction models that have since been issued on Hugging Face. Kevin Murphy, NASA Headquarters chief science data officer, said the partnership fills a gap in the knowledge and experience being leveraged into novel AI models. Such a strategy will enable greater insights into the way solar activity affects vital services and technologies upon which society depends.
The release of Surya by NASA and IBM is a paradigm shift in the space weather forecasting environment by infusing highly evolved Artificial Intelligence and making it open-source. Model capability to make significant advancements in solar flare forecasting, in combination with visual forecasting, provides the potential of saving global infrastructure. By freely publishing Surya on Hugging Face, the initiative eliminates access barriers to advanced tools to predict space weather to make this tool available to researchers in developing countries.