PhD Students’ Groundbreaking New Technique Rescues JWST’s Highest Resolution Data

It’s hard to believe the Hubble Space Telescope was considered a disaster shortly after launch, with misshaping of the primary mirror smearing its images badly. A mission to fit corrective mirrors fixed that problem and gave us 30 years of priceless visions of the universe. Before JWST was launched, astronomers were painfully aware that its far more distant location meant there could be no such rescue mission. The knowledge it had to work first time was one of the reasons it suffered so many delays and cost overruns. 

The triumphant success of the first JWST images overshadowed the fact that there was an imperfection, albeit a much more limited one than for Hubble. Electrical charge was bleeding from some pixels to their neighbors, known as the brighter-fatter effect. 

“The problem was in the infrared detectors,” University of Sydney PhD student Max Charles told IFLScience. “It’s common for almost all the observing modes, but the one [head of Charles’ lab, Professor Peter Tuthill] created is a very high-resolution mode, and this was where it was particularly damaging,” Charles added.

Tuthill designed the Aperture Masking Interferometer (AMI), which was originally intended as a backup if other methods to make JWST operate failed, but also allows the telescope to capture small, faint objects overshadowed by nearby, bright ones. That would be invaluable for observing planets forming around young stars, such as the PDS 70 system, or it would if it wasn’t for the brighter-fatter problem.

In cases like this, the bleeding pixels affect image quality, but mostly when the target covers a particularly tiny area of sky. The effect is made worse by the techniques used to combined the data from JWST’s 18 mirrors into one cohesive picture.

Some astronomers who had been delighted at getting precious time on JWST to image their chosen targets struggled to get anything publishable out of the observations. One of Charles’ colleagues managed to publish, but Charles told IFLScience: “Only by throwing away 80 percent of the data.” 

Without a fix, there was no prospect for getting more JWST time for such projects, and many other targets were being affected to a lesser degree. Louis Desdoigts had already made finding a solution the subject of his thesis when Charles started his PhD, and he joined the project.

With so many astronomers affected, some perhaps with their careers on the line, Desdoigts told IFLScience he felt considerable pressure. Indeed, he added: “I shied away from it at the start.” However, after Desdoigts developed the necessary background processing for the first half of his PhD, he realized how distraught his supervisor, Dr Benjamin Pope of Macquarie University, and some other astronomers were. “We’ll never get any data,” Pope told him. Although the instrument was not to blame, it was projects involving Tuthill’s AMI that were most affected. 

Desdoigts talked with other teams around the world who were looking for fixes, but told IFLScience that while these approaches might have resolved individual cases, “they could not provide a general solution. Ours was a fundamentally different approach.” If he and Charles couldn’t make it work, no one would.

“It had to be a software fix,” Charles told IFLScience, since a mission to change the hardware was out of the question. “If there was something else that could have been done, we would have tried it. We tried a lot of software fixes, and this was the only one that stuck to the wall.”

Desdoigts is the lead author of a paper that describes AMIGO (Aperture Masking Interferometry Generative Observations), the system they built using simulations and neural networks to understand the electronic bleeding and reverse it.

Charles led the work to test AMIGO’s success, running it over the faulty data for a variety of targets. A preprint still undergoing peer review describes its success on neutrino-rich galaxy NGC 1068, WR 137, a colliding wind-binary star system, and other objects particularly vulnerable to these effects.

Possibly the most striking evidence of success comes from Io. JWST was never going to be able to match the quality of images provided by the Galileo and Juno spacecraft at much closer range, so from a scientific perspective, the innermost of Jupiter’s large moons was not the telescope’s priority.

However, Io represented a particularly strong example of an affected target, and one where those same Juno images could be used to check the validity of AMIGO’s unscrambling of existing JWST data. It proved a huge success, with Charles telling IFLScience: “We could make out volcanoes as lighter dots against the Io background and track them as they rotated.”

“Instead of sending astronauts to bolt on new parts, they managed to fix things with code,” Tuthill said in a statement. “It’s a brilliant example of how Australian innovation can make a global impact in space science.”

Researchers have started taking the opportunity to reassess their data using AMIGO, and the team expect many positive results. The announcement coincides with JWST Cycle 5 applications closing, the deadline for astronomers to submit their pitches for time on the space telescope, with many who would once have been excluded now back in the contest.

Desdoigts and Charles celebrated the news, and the awarding of Desdoigt’s doctorate, by getting matching tattoos of the instrument they have helped reach its full potential.

A description of AMIGO’s methods, and why other approaches have failed, has been accepted by Publications of the Astronomical Society of Australia, while the preprint of the work Charles led, demonstrating its success, can be read on arXiv.