Overview: A distant problem, a local solution
When the James Webb Space Telescope (JWST) launched in 2021, the world watched with awe at humanity’s most powerful eye into the cosmos. Yet even as the telescope floated in space, a stubborn challenge persisted: subtle image blur arising from how Webb’s instruments handle light. For Australian scientists, the task was not only to diagnose the issue but to craft a software-based fix that could be applied from Earth, millions of kilometres away. Their work focused on the aperture masking interferometer (AMI), a high-resolution mode tucked inside one of Webb’s cameras that helps scientists measure and correct distortions in the light reaching the detector.
The heart of the problem: electronic leakage in AMI
Led by PhD student Louis Desdoigts at the University of Sydney, the team identified an electronic leakage where bright pixels bled into dark ones. In infrared cameras, this bleed is known, but Webb’s extreme sensitivity amplified the effect, making faint objects—such as distant exoplanets—hard to distinguish near brighter stars. The blur wasn’t just cosmetic; it limited Webb’s ability to resolve faint companions and intricate structures in distant systems. The Australian researchers treated this as a solvable signal-processing problem rather than a hardware fault to be repaired in space.
A model, a machine learning partnership, and a path to reversal
To counter the leakage, the team built a physical model of AMI’s optical behavior. This model simulated how light propagates through the interferometer and how electronic leakage alters the observed data. But a model alone cannot correct real-world data with the precision Webb demands. So the researchers integrated the model with machine learning techniques to detect the telltale signs of leakage and reconstruct the pristine signal that AMI would have recorded in the absence of the blur.
The result was a calibration pipeline that could reverse the blur in Webb’s data. The new approach, described in online preprint repositories, demonstrated that software alone could restore a portion of the telescope’s high-resolution capability. By validating their method against known targets and simulated datasets, the team showed that AMI could once again deliver crisp, trustworthy measurements of astronomical sources.
What the improved images revealed
With the correction in place, JWST’s images took on a sharper character. The updated data captured by AMI allowed astronomers to spot a faint planet and a reddish-brown dwarf orbiting star HD 206893—an achievement that had previously been obscured by the leakage. The enhanced clarity opened up possibilities for detecting unseen planets by improving the contrast and resolution in Webb’s infrared observations.
Another University of Sydney PhD student, Max Charles, extended the correction to more complex imagery. Applying the calibration to varied scenes, the team brought Jupiter’s volcanically active moon Io into sharper focus, enabling time-resolved observations of eruptions over sequences lasting an hour. They also demonstrated improvements in imaging the central engine of a distant active galaxy, the jet from NGC 1068, and a dusty ribbon around the star WR 137. Each success underscored the broader potential of software-based calibration for current and future space telescopes.
Why this matters: a new paradigm for space instrumentation
JWST sits about 1.5 million kilometres from Earth, far beyond the reach of routine servicing missions. Hubble’s extended career included on-orbit repairs, but Webb’s orbital distance makes physical fixes impractical. The Australian-led AMI calibration work exemplifies a crucial trend in modern astronomy: software and data processing can compensate for hardware imperfections, extending a telescope’s lifetime and scientific yield without costly missions to repair optics.
Beyond Webb, the approach has implications for next-generation instruments. The calibration techniques, and the machine-learning frameworks that accompany them, could inform future missions such as NASA’s Roman Space Telescope, helping researchers push the boundaries in the search for Earth-like planets and other faint phenomena in the universe.
What lies ahead
As more datasets from Webb are released, scientists will apply and refine these corrections, improving image fidelity across the board. Each corrected pixel is a step closer to resolving the atmospheres of distant worlds, mapping black-hole environments with unprecedented clarity, and studying the complex interplay of cosmic dust and starlight. The Australian contribution demonstrates how collaborative, cross-disciplinary science—spanning optics, physics, and machine learning—can transform how we observe the cosmos from far away.
In the words of the researchers, every improvement in pixel-level accuracy magnifies our ability to explore far-off worlds and to decode the faint whispers of the universe with greater certainty.
