Trending Articles
Trending Topics
Trending
IBM and NASA have released a new AI model to help scientists better predict the weather in space.
The model, named Surya after the Sanskrit word for sun, predicts solar activity using information gathered from high-resolution images of the sun’s surface, along with seismic and magnetic data.
Solar activity can have devastating consequences for life on Earth. Particularly solar storms, where the sun releases a a burst of radiation and charged particles. Such events can interfere with Earth’s magnetic field, IT infrastructure and communications networks.
As well as knocking out satellite-based systems such as GPS, solar events can interfere with national power grids, causing power outages and blackouts. The result: disruption to transport, financial systems and supply chains.
Recent research by insurance group Lloyds of London predicted that a solar flare could cause global economic losses of between $1.2 trillion and $9.1 trillion over a five-year period.
IBM and NASA hope Surya can help the way such events are forecast, making predictions of solar activity more reliable.
Previous solar weather prediction systems used satellite imagery of parts of the sun’s surface to make their forecasts. Surya, however, has been trained using nine years of high-resolution solar observation data from NASA’s Solar Dynamics Observatory. According to IBM, the dataset was around ten times larger than the average used to train AI.
Surya is now being used to forecast space weather events, including “predicting solar flares, the speed of solar winds, solar EUV spectra prediction and the emergence of active regions on the Sun”, according to IBM.
The AI model is available on Hugging Face under an open-source licence.
Surya is the latest AI collaboration resulting from the partnership between Big Blue and the US space agency. Last year, the pair released a weather model called Prithvi, a 2.3 billion parameter model that uses 40 years of climate data to predict severe weather and natural disasters. This, they promise, forecasts how global warming will affect particular regions with greater accuracy.
