Astronomers at Northwestern University have developed a new computational method that makes it possible to cope better with the dazzling light of stars when searching for exoplanets in images from the James Webb Space Telescope.
The method, called FLIP (Physics-based Learning for Imaging Planets), creates a digital twin of the telescope and precisely models how light passes through its mirrors and detectors. This helps account for tiny imperfections and vibrations that interfere with conventional image-processing methods.
By applying FLIP to existing JWST observations, the researchers increased the sensitivity for detecting faint planets by a factor of 2–15. In two-year-old data on the 51 Eridani system, the technique made it possible for the first time to see the giant exoplanet 51 Eri b at a wavelength where carbon dioxide in the planet's atmosphere strongly absorbs infrared light.
Conventional methods, including the standard KLIP, could not cope with the residual glare of the star and did not notice the planet. FLIP, by contrast, produced a confident detection above 5 sigma precisely where it was expected.
The study's lead author, Rodrigo Ferrer-Chávez, and co-author Jason Wang note that telescopes like JWST cannot be absolutely stable: thermal changes and vibrations of the cryocoolers cause shifts on the order of picometers. The new technique compensates for these effects without modifying the telescope's hardware.
The team tested FLIP on observations of the stars HIP 65426, 51 Eridani and TYC 5899. In every case the result was markedly better than with traditional approaches.
The study was published on 4 November 2026 in the journal Nature. How will this affect the search for colder and fainter worlds in the future?
The scientists hope that FLIP will help extract more information from the JWST data already collected and will prove useful in building the next generation of telescopes, such as the proposed Habitable Worlds Observatory.
