New developments in artifical intelligence will actually be able to predict upcoming solar flares.
The Solar Dynamics Observatory gathers data on the Sun and it does it well. As a matter of fact, it collects more data than any satellite NASA has ever sent out into space. Parsing the 1.5 terabytes of data that the satellite collects per day can be a bit much for researchers, however. That’s why researchers at Stanford University are using artificial intelligence to help parse and sort the sea of information.
Using a process called “machine learning”, artificial intelligence can sort data into relevant categories becoming better at sorting as it gains more and more data.
“Machine learning is a sophisticated way to analyze a ton of data and classify it into different groups,” said researcher Monica Bobra.
Bobra and her fellow researcher Sebastien Couvidat wanted to see if the machine’s propensity to find patterns could be used to predict the strength of potential solar flares. This is important information as the massive bursts of solar energy can wreak havoc on communications equipment on Earth.
Using a database of nearly 2000 active regions, the researchers cataloged active and non-active areas of the Sun’s surface, then fed the numbers into their machines. The machines quickly learned to identify relevant features of active regions that threatened to send out a solar flare and dormant regions.
As the machine gets better at predicting which areas will flare, communications workers will be able to prevent damage to their equipment (or at least not be caught by surprise) when a solar flare makes its way toward Earth.
Always hoping to be more efficient, the researchers have pared down even this latest technology that would allow them to predict he coming flares. Before the study was completed, researchers realized that not all 25 of the features used by the machine to discriminate between active and non-active regions were necessary for the machine to make an accurate prediction.