Scientists Teach the Webb Telescope to Erase Starlight and Reveal Hidden Planets

Photographing a planet around another star is a bit like spotting a firefly beside a lighthouse. Now a team of scientists has found a smarter way to switch off the glare.

In a paper published in Nature on November 4, researchers from Caltech, Northwestern University, MIT and the Space Telescope Science Institute describe a method that removes starlight from James Webb Space Telescope images with far greater precision than before. They call it FLIP, short for Physics-based Learning for Imaging Planets.

A telescope's digital twin

The team built a detailed digital model, or "digital twin," of Webb's Near Infrared Camera, known as NIRCam. Using optimization techniques borrowed from machine learning, they solved for millions of tiny imperfections and unknowns that affect what the real telescope records. That let them model exactly how a star's glare should look, then subtract it.

According to Caltech, the approach improved detections of known planets with sensitivity two to 15 times better than standard techniques. In archived Webb data of the star 51 Eridani, it produced a clear detection of the young planet 51 Eri b, which an earlier analysis of that same observation had not picked up.

"By combining a physical model of the instrument with optimization techniques from machine learning, we have been able to remove starlight much more precisely," said Katie Bouman, a Caltech professor and co-author of the study.

A student's comeback

Lead author Rodrigo Ferrer-Chavez, a graduate student at Northwestern, first tackled the problem as an undergraduate in Mexico. His early attempts did not work, but they helped the team understand what to avoid. Instead of training a model to guess the glare, the researchers used machine learning's ability to tune millions of settings to carry out physics calculations no person could do by hand.

"We now have a tool that lets us discover and study planets that we were not able to 'see' before, even if that planet was already in an image in the JWST archive," Ferrer-Chavez said.

Why it matters

Because the starlight is removed after the images are taken, the method could work on other observatories too, including NASA's Nancy Grace Roman Space Telescope and ground-based telescopes. The team hopes it will inform NASA's planned Habitable Worlds Observatory, which would search for Earth-like planets around Sun-like stars.

"I feel we just unlocked a new frontier," said co-author Jason Wang of Northwestern.

There is still a long road to photographing another Earth. But this study shows that better software can help a telescope see well beyond its original limits.