Exoplanets Can't Hide From The JWST's 'Digital Twin'
Finding planets beyond our solar system would be considerably more straightforward if they didn't orbit so frustratingly close to their host stars. Yet that is precisely the challenge facing modern exoplanet science: the overwhelming brilliance of a star routinely drowns out the comparatively faint light reflected or emitted by any planets in its vicinity. A newly published technique, however, is changing the game — and it starts by building a meticulous digital replica of the most powerful space telescope ever constructed.
The challenge of direct imaging of exoplanets is not merely one of raw telescope power. Even the James Webb Space Telescope (JWST), humanity's most sophisticated eye on the cosmos, struggles to tease apart the faint glow of a distant world from the blinding radiance of the star it circles. The brightness contrast between a star and an Earth-like planet can exceed ten billion to one in visible light and around one million to one in infrared — a staggering disparity that demands ingenious engineering and equally ingenious mathematics to overcome.
Coronagraphs: Powerful But Imperfect
Coronagraphs are the astronomer's primary weapon against stellar glare. First invented in 1930 by French astronomer Bernard Lyot to study the solar corona, these optical devices use precisely shaped masks to block the direct light of a star, much like holding your hand up to shield your eyes from the sun. Modern coronagraphs aboard space telescopes can suppress starlight by factors of thousands to millions, revealing objects hiding in the glare that would otherwise be completely invisible.
The JWST is equipped with coronagraphic capabilities embedded in two of its primary instruments: the NIRCam (Near-Infrared Camera) and the MIRI (Mid-Infrared Instrument). These tools have already proven their worth, enabling scientists to study faint debris disks — the dusty remnants of planetary formation — and to scrutinize extraordinarily luminous objects such as quasars whose surrounding environments would otherwise be lost in their intense light. For more on the JWST's coronagraphic instruments, the NASA JWST Instruments page provides detailed technical overviews.
Yet even coronagraphs have hard limits. At the heart of those limits lies a fundamental concept known as the telescope stability limit — the threshold beyond which mechanical, thermal, or optical fluctuations within the observatory itself render pristine images impossible to obtain. The JWST was engineered with extraordinary care to push this threshold as high as possible. It is equipped with fine guidance sensors capable of locking onto guide stars with astonishing precision, and a five-layer sunshield the size of a tennis court that keeps the telescope's optics at an operating temperature near –233°C (40 Kelvin), shielding it from the thermal noise of the Sun, Earth, and Moon simultaneously. And yet, the stability limit remains.
The Hidden Enemy: Nanoscale Instabilities
To appreciate why this matters, consider the scale of the problem. The JWST's primary mirror spans 6.5 meters in diameter and is composed of 18 hexagonal gold-coated beryllium segments. To function as a single unified mirror, those segments must be aligned to within a fraction of a wavelength of light — tolerances measured in nanometers, or billionths of a meter. Any deviation from perfect alignment, no matter how small, distorts the resulting image.
"We try to make telescopes extremely stable, but there are extreme engineering challenges that we haven't quite solved yet. There are known causes of instability and unknown causes. Small thermal changes could cause a mirror to expand or contract by a picometer. Telescopes like JWST have cryocoolers to keep the instrument super cold. Those coolers have little motors that can cause the mirrors to vibrate. So, there are issues like those but also issues that we still don't understand as we learn how to build these complex telescopes." — Jason Wang, assistant professor of physics and astronomy, Northwestern University
Even a shift of a single picometer — one trillionth of a meter, roughly one-hundredth the diameter of a hydrogen atom — can alter the characteristic pattern of starlight that leaks around a coronagraph. These infinitesimal perturbations accumulate into a subtle, constantly shifting background of speckle noise: a mottled haze of residual starlight that masquerades as faint point sources and buries the signal of genuine planets beneath a sea of optical artifact.
The Standard Approach and Its Limitations
For roughly two decades, astronomers have addressed the challenge of residual starlight through a technique called reference differential imaging (RDI). The concept is straightforward in principle: observe a reference star — one that is as similar as possible to the target star but known not to host detectable planets — and then subtract that reference image from the target observation. The process removes much of the structured starlight pattern, in theory leaving behind only genuine astrophysical signals.
"For the past 20 years, astronomers have tackled this problem by observing a different star and using its image as a reference," explained study co-author Jason Wang. "Then, they subtract that reference image from the original observation to remove the unwanted glare to reveal faint planets that otherwise might be hidden."
In practice, however, this approach is hampered by an unavoidable problem: the telescope does not behave identically when pointing at two different stars. The act of slewing the telescope, waiting for it to thermally settle, and re-acquiring a guide star introduces subtle but consequential changes to the optical system. The resulting mismatch between the reference and target observations leaves behind a residual pattern of starlight that can still overwhelm faint planetary signals. Sophisticated algorithms such as KLIP (Karhunen–Loève Image Processing) attempt to statistically model and remove this residue, but they too have their ceilings.
As the authors of the new research state plainly:
"As well as sophisticated instrumentation, effective image processing is necessary for starlight suppression, which continues to be a substantial limitation."
Introducing FLIP: A Physics-Based Revolution
Now, a new study published in the prestigious journal Nature introduces a fundamentally different approach. The research, titled "Imaging exoplanets below the telescope stability limit," is led by Rodrigo Ferrer-Chavez, a graduate student in the Department of Physics & Astronomy at Northwestern University, and co-authored by Jason Wang and Katherine Bouman, among others. The team's new method is called FLIP — Physics-based Learning for Imaging Planets.
Rather than relying on empirical reference images of other stars, FLIP constructs a mathematically rigorous, physics-based model of how light propagates through the JWST's entire optical system — from the moment it enters the telescope to the moment it registers on the detector. The model accounts for the shape of the primary mirror, the properties of the coronagraphic masks, the effects of residual wavefront errors (deviations from a perfect optical wavefront), and even the subtle nanoscale drifts and vibrations that the telescope experiences during an observation.
"Planets are a million times fainter than their stars. So, to image exoplanets, we need to remove that glare. That is a very tricky problem because even the tiniest changes in the telescope can affect how that glare appears. Slight shifts or tilts — even on the scale of just a few nanometers — can change the pattern of light. Our method models how light travels through the telescope and accounts for those tiny changes, allowing us to remove the glare much more precisely." — Rodrigo Ferrer-Chavez, lead author
Understanding the Point Spread Function
Central to FLIP's operation is a concept called the Point Spread Function (PSF). No optical system is perfectly ideal; every lens, mirror, and detector introduces some distortion to incoming light. As a result, what should appear as a perfect mathematical point of light — a distant star, for instance — is instead spread out into a characteristic pattern that reflects all the imperfections of the optical path. For the JWST, this PSF is exquisitely well characterized under stable conditions, but it shifts and evolves subtly as the telescope's optical state changes over time.
Standard image-processing algorithms treat the PSF as essentially fixed or model it statistically. FLIP takes a more sophisticated approach: it uses a differentiable instrument model — meaning the model's parameters can be systematically adjusted using calculus-based optimization — in conjunction with real-time wavefront sensing data collected by the telescope itself. This allows FLIP to track the actual, instantaneous state of the JWST's optics and construct a highly accurate PSF model for each individual observation, rather than relying on a generic average.
FLIP also incorporates ideas borrowed from machine learning, specifically the use of gradient-based optimization to iteratively refine its model until it matches the observed data as closely as possible. This fusion of physical modeling with data-driven optimization is what gives FLIP its exceptional power.
The Digital Twin of JWST
In effect, the researchers built what they describe as a "digital twin" of the James Webb Space Telescope. This virtual replica reproduces the behavior of the actual telescope with extraordinary fidelity, capturing not only its nominal design parameters but also its real-world quirks, imperfections, and moment-to-moment fluctuations. By iteratively tuning this digital twin until its simulated output matches the actual observed data — including the contaminating starlight — the team can then subtract a highly accurate starlight model from the observation, revealing whatever lies beneath.
The concept of a digital twin has gained considerable traction across engineering and manufacturing disciplines, where virtual replicas of complex physical systems are used to predict behavior, diagnose problems, and optimize performance without touching the real hardware. Applying this concept to a space telescope represents a novel and potentially transformative extension of the idea.
Proof of Concept: Recovering 51 Eridani b
To validate FLIP, the team applied it to JWST/NIRCam observations of 51 Eridani b, a directly imaged gas giant exoplanet located approximately 96 light-years from Earth in the constellation Eridanus. First discovered in 2015 using the Gemini Planet Imager at the Gemini South telescope, 51 Eri b is estimated to be roughly two times the mass of Jupiter, making it one of the least massive directly imaged exoplanets known at the time of its discovery.
The choice of 51 Eri b was deliberately strategic. The team analyzed observations taken at a wavelength of 4.3 micrometers (μm), corresponding to the F430M filter of NIRCam. This wavelength is particularly significant because it falls within a strong carbon dioxide (CO₂) absorption band, making it a critical spectral region for characterizing exoplanet atmospheres and searching for potential biosignatures. It is also, as the researchers note, a kind of sweet spot in JWST's capabilities — both scientifically important and technically demanding.
When standard processing methods, including the widely used KLIP algorithm, were applied to these observations, the planet remained stubbornly invisible, buried beneath residual starlight noise. When FLIP was applied to the same data, 51 Eri b emerged with a detection significance exceeding 5-sigma — the conventional threshold for a confident astronomical detection.
"We achieve sensitivity 2–15 times superior to standard techniques, including the first detection of 51 Eridani b at 4.3 μm. This is the first time, to our knowledge, that such an approach has demonstrated substantially improved sensitivity on real, on-sky coronagraphic imaging." — Ferrer-Chavez et al., Nature 659, 58–62 (2026)
"We showed that we can actually recover a planet that people previously couldn't see," said Wang. "This is real data in a real scenario. It was science left on the table because we didn't have the right tools to analyze the data in the most optimal way."
The implications extend far beyond 51 Eri b. The technique is not merely improving the detection of already-known planets; it is demonstrating that existing JWST datasets may harbor hidden planetary detections waiting to be unlocked with better analytical tools. A systematic reanalysis of archival JWST coronagraphic data using FLIP could yield a harvest of new discoveries without the need for a single additional observing hour.
Broader Scientific Implications
A New Census of Exoplanets
The significance of direct imaging as a technique cannot be overstated. While methods such as the transit method — which detects the slight dimming of a star as a planet passes in front of it — and radial velocity measurements — which detect the subtle wobble imparted to a star by an orbiting planet's gravity — have discovered the vast majority of the more than 5,700 confirmed exoplanets known today, these techniques are most sensitive to planets that orbit close to their host stars. Direct imaging, by contrast, is most sensitive to planets in wide orbits, at distances of one astronomical unit (AU) or more — the region analogous to the outer solar system where giant planets like Jupiter and Saturn reside, and where Earth-like planets could in principle exist around other stars.
As Ferrer-Chavez noted: "We know of no planets around other stars that are like those in our own solar system. It's not that they don't exist. It's that we cannot see them. Every time we improve our techniques, we discover new types of planets. By finding more planets, we can better understand how planetary systems form and evolve, and ultimately how typical our own solar system truly is."
Key scientific benefits of FLIP and improved direct imaging include:
- Atmospheric characterization: Direct imaging combined with spectroscopy allows scientists to analyze the chemical composition of exoplanet atmospheres, searching for gases like water vapor, methane, CO₂, and potentially oxygen or ozone as biosignature candidates.
- Orbital dynamics: Directly imaged planets can have their orbits traced over time, providing constraints on the architecture of planetary systems.
- Planetary formation theories: The demographics of wide-orbit giant planets revealed by direct imaging inform competing theories of how planets form — specifically the debate between core accretion and disk instability formation pathways.
- System architecture: Understanding what types of planets exist at what distances from their stars helps astronomers assess whether solar systems like our own — with rocky inner planets and gas giants in the outer regions — are common or rare.
- Unlocking archival data: Reprocessing existing JWST coronagraphic observations with FLIP could reveal previously undetected planets without requiring new telescope time.
Paving the Way for the Habitable Worlds Observatory
Frequently Asked Questions
Quick answers to common questions about this article
