AI Unlocks 100 Years of Sun Records to Reveal Solar Activity

Scientists have used artificial intelligence to unlock more than a century of hand-drawn solar observations from India's Kodaikanal Solar Observatory, creating one of the longest continuous records of the Sun's magnetic activity and offering fresh insights into how solar cycles have evolved over time.

Century-old Sun drawings become digital science

Researchers led by Dibya Kirti Mishra from the Aryabhatta Research Institute of Observational Sciences (ARIES), an autonomous institute under the Department of Science and Technology (DST), worked with experts from the Indian Institute of Space Science and Technology, the Southwest Research Institute in the United States, and the Indian Institute of Astrophysics to transform historical solar records into machine-readable scientific data. The Kodaikanal Solar Observatory preserves an exceptional archive of daily "suncharts" dating from 1904 to 2022. Before digital imaging became available, astronomers carefully drew features visible on the Sun, including sunspots, filaments, prominences and plages, creating an invaluable historical record. Differences in drawing styles, ageing paper and varying scan quality made it difficult to analyse these observations consistently using conventional techniques.

Machine learning maps magnetic activity across nine solar cycles

The research team used a supervised machine learning model known as U-Net to process the scanned drawings. The system first identified the Sun's disk in every image, accurately determining its centre, size and orientation before locating and outlining plages, which are bright magnetically active regions on the Sun that provide important clues about solar magnetic behaviour.

The AI model successfully traced these magnetic regions across records covering nine solar cycles between 1916 and 2007. The resulting data allowed scientists to build a detailed "butterfly diagram," a visual map showing how magnetic activity gradually shifts in latitude as each solar cycle progresses. The researchers also found that the measurements closely matched data obtained from the observatory's Ca II K full-disk observations, confirming the reliability of the reconstructed dataset.

Better solar history could strengthen space weather research

The newly reconstructed record provides scientists with a much longer and more consistent view of the Sun's magnetic activity than was previously available. This extended dataset can improve studies of how solar cycles differ in strength and structure while helping researchers better understand changes in the Sun's energy output over many decades.

A clearer picture of past solar activity also supports efforts to improve space weather research, which plays an important role in protecting satellites, navigation systems, communication networks and power infrastructure from solar disturbances. The study demonstrates how modern artificial intelligence can breathe new life into historical scientific archives, turning fragile handwritten observations into valuable digital resources for future research.