Erasing the stars to find exoplanets

Outside of our solar system, scientists estimate that billions of worlds exist in the Milky Way galaxy alone. Called exoplanets, these celestial objects represent a chance to find other habitable worlds and learn more about our intriguing galactic neighbors. But imaging exoplanets is hard; most orbit a star just like Earth, and they are very faint compared to the blazing masses that appear near them in astronomical images. 

Now, a team of researchers has developed a new way to remove starlight from telescope data to reveal exoplanets hiding amongst the bright flares. By building a "digital twin" of the James Webb Space Telescope (JWST) Near Infrared Camera (NIRCam), they were able to manipulate physical parameters to model the glare of stars and essentially stamp out the stellar light in existing JWST images to find concealed worlds. Their method improves detections of known planets with a sensitivity two to 15 times superior to standard techniques.

"Directly imaging exoplanets is incredibly challenging because even the tiniest imperfections in a telescope can make it difficult to separate the faint light of a planet from the overwhelming glare of its star," explains Caltech's Katie Bouman, professor of computing and mathematical sciences, electrical engineering and astronomy, a Rosenberg Scholar, and a Heritage Medical Research Institute Investigator, who is a co-author of a Nature paper about the new method published online November 4.

"What's exciting is that we're showing how computation can help overcome limitations of the telescope itself," Bouman continues. "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.”

After building a digital model of JWST's NIRCam, the researchers used computational methods to solve for millions of possible physical imperfections and other unknowns that could affect data collected by the actual telescope. By sorting out potential defects, they were able to get a perfect model of what a given star's glare looks like, which enabled them to remove it from test images. Their approach is called Physics-based Learning for Imaging Planets (FLIP).

"This is the first time that we have demonstrated that these little imperfections or instabilities are not a limiting factor in the kinds of planets we can image," says lead author Rodrigo Ferrer-Chavez, a graduate student at Northwestern University in the lab of astronomer Jason Wang, who is also a co-author of the paper. "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."

 Ferrer-Chavez first tried to solve for a way to suppress starlight as an undergraduate at the Autonomous University of Yucatán in Mexico. He had previously completed a Summer Undergraduate Research Fellowship project with Wang, then a 51 Pegasi b Postdoctoral Fellow (named after the first exoplanet discovered orbiting a Sun-like star) in the lab of Dimitri Mawet, the David Morrisroe Professor of Astronomy at Caltech and a senior research scientist at the Jet Propulsion Laboratory, which Caltech manages for NASA. Mawet is also the director of instrumentation for the Caltech Optical Observatories.

Ferrer-Chavez was hungry for more research opportunities, and Wang was in conversations with Bouman about how her expertise in computational imaging might be combined with his knowledge in instrumentation to get better exoplanet image results. So, they decided to mentor Ferrer-Chavez on a collaborative project. But, like many experiments, his initial attempt failed.

"Rodrigo's senior thesis was an exploration of possible approaches, none of which ended up being what we did for this paper, but it helped rule out what not to do and also helped us understand the problem better," Wang says. "I believe our recent success was only possible because we did that exercise first."

Ferrer-Chavez initially thought he could train machine learning models to predict what the glare of a star would look like in the data, but it didn't work. Instead, the team had to turn that idea on its head, leverage technologies from machine learning—specifically the ability to have millions of different "knobs" or controls you can tweak—to do the physics calculations that would be impossible for a human to achieve in any reasonable amount of time.

"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 explains. "We can try to find planets that may have been missed there, and also planets that perhaps we tried to image, but we could not actually detect at the time."

Recent developments in telescope technology have improved researchers' ability to adjust for or block out stellar glare. For example, NASA's recently launched Nancy Grace Roman Space Telescope has a Coronagraph Instrument designed by scientists at Caltech, JPL, and IPAC (a science and data center for astronomy at Caltech) that is specifically designed to block out starlight to capture exoplanets. But a physical telescope could never achieve FLIP’s level of precision to reveal hidden exoplanets in images, Wang says.

"The James Webb Space Telescope, which we demonstrated on, is optically stable to the nanometer level, which just sounds pretty awesome, but that's still limiting," he says. "We're excited because this technique is not limited specifically to JWST. Because of some very specific design, this technique could be used on any telescope."

The team plans to work on improving their FLIP technique and applying it to other instruments, including the Roman Space Telescope and ground-based telescopes. They hope it could help inform future missions as well, such as NASA's planned Habitable Worlds Observatory (HWO), which would be the first telescope designed to search for signs of life on exoplanets.

"The goal of HWO is to find Earth-like planets and the habitable zones of Sun-like stars through a direct imaging method," Ferrer-Chavez says. "The engineering challenge of trying to image something that is around 10 billion times fainter than something right next to it is a brutal—not just to clean the image but to build a telescope that can take that image."

By having the team's new FLIP approach to remove starlight after the images are taken, engineers may not have to solve for instability issues in the same way.

"I feel we just unlocked a new frontier," Wang says. "We've now gone beyond the telescope stability limit for the first time with this new technique, and this new paradigm, I think, can be applied to every instrument that's trying to image exoplanets."

The ultimate dream, Bouman says, is to take pictures of Earth-like planets around other stars and to, perhaps, one day find signs of life.

“There's still a long way to go, but this work shows how much more is possible when we combine advances in telescope technology with computational imaging," Bouman says. "Sometimes, the key to seeing something new isn't just building a better instrument but finding a better way to make sense of the light it collects.”

The Nature paper is titled "Imaging exoplanets below the telescope stability limit." Additional co-authors are William T. Freeman and Brandon Y. Feng from MIT and Laurent Pueyo and Marshall D. Perrin the Space Telescope Science Institute in Baltimore. The work was supported by funding from the Space Telescope Science Institute, the Sloan Foundation, an Amazon AI4Science Discovery Award, and the National Science Foundation.

 

---

---