SAN FRANCISCO — The Aerospace Corp. is working with Google Public Sector to apply artificial intelligence and high-performance computing to solar-activity forecasts.
The goal is to improve prediction of geomagnetic storms days in advance “with unprecedented accuracy and speed,” according to a Jan. 28 news release.
“This collaboration is a game-changer for space weather forecasting and a powerful example of how innovation and partnership can drive national security and societal impact,” Kevin Bell, Aerospace Engineering and Technology Group senior vice president, said in a statement. “Together, we’re not just advancing the science of space weather forecasting; we’re also strengthening the resilience of the systems and critical infrastructure that underpin our modern way of life.”
To date, forecasters have struggled to predict solar flares and coronal mass ejections due in part to the complexity of solar activity and the volume of observations.
NASA’s Solar Dynamics Observatory alone provides about 70,000 daily images. NASA’s Parker Solar Probe, which passed within 6.1 million kilometers of the sun in December, is providing additional measurements of the sun’s magnetic field, plasma and high energy particles. More open-source data relevant to space-weather forecasting comes from government agencies including NASA, the National Oceanic and Atmospheric Administration, the National Science Foundation and the European Space Agency.
Training machine-learning models on this vast dataset in ways that minimize bias “is computationally challenging,” Alison Kremer, a senior engineer specialist who leads the Aerospace Corp. data-science team investigating space weather, told SpaceNews by email. “There are multiple data sources, types of data, and machine-learning models to investigate for space weather forecasting. Without [high-performance computing], training time would be prohibitively long; it would be impossible to make meaningful progress.”
Through the partnership, Google will apply technologies like Vertex AI, a Google Cloud machine-learning platform, to the problem. Google Public Sector, a 2022 spinoff from Google Cloud, is based in Reston, Virginia.
“We are proud to partner with The Aerospace Corporation to advance scientific discovery and address critical challenges facing our nation,” Jim Kelly, Google Public Sector Federal Sales vice president, said in a statement. “This collaboration demonstrates the transformative potential of AI to benefit society as a whole.”