Astronomers have developed a new computational method called Physics-based Learning for Imaging Planets (FLIP) that removes starlight from James Webb Space Telescope images to reveal hidden exoplanets. The technique uses a digital simulation of Webb's Near Infrared Camera to model how the instrument's optical system produces starlight patterns, then subtracts those patterns from actual observations. In tests, FLIP achieved two to 15 times greater detection sensitivity than conventional approaches and has already exposed an exoplanet that was concealed in existing data.
The method was developed by researchers at Caltech, Northwestern University, and NASA's Jet Propulsion Laboratory, including Katie Bouman, Rodrigo Ferrer-Chavez, and Jason Wang. Ferrer-Chavez explains the process: "As you go through this iterative process of trying to find what the telescope actually is doing during observations, you basically see the glare of the star disappearing in very bizarre ways. It gets dimmer and very deformed until at the end you get a pretty clean image and, if you're lucky, there's a blob there, which is a glowing planet."
The technique combines physical laws governing light and telescope optics with machine learning optimization methods that adjust millions of variables to refine the simulation until it reproduces actual observations more closely. Bouman notes: "By combining a physical model of the instrument with optimization techniques from machine learning, we have been able to remove starlight much more precisely, making it easier to detect faint exoplanets and even reveal one that was previously hidden in the JWST data."
The approach may allow astronomers to search existing Webb archive data for previously missed planets without requiring new observations. A paper describing the method was published online in Nature on November 4.