Blog

From One, Many

20 October 2025

A photograph of a full-scale model of the James Webb Telescope. The seven hexagonal mirrors of JWST are clearly seen. EADS Astrium
A full-scale model of the James Webb Space Telescope.

The wavelengths of radio light are so large that you can’t capture a high-resolution image with a single dish. To capture an image as sharp as, say, the Hubble telescope, you’d need a radio dish tens of kilometers across. So radio astronomers took a different approach. They used an array of dozens of antennas, each capturing their own signal. Since the antennas not only capture precise data but also the precise timing of that data, astronomers can use a process known as interferometry. Light from a distant radio object reaches each antenna at a slightly different time, and by correlating the arrival times, astronomers can treat the array as a virtual antenna disk the size of the entire array. From many, one, as the saying goes.

Optical astronomy doesn’t need to bother with this sort of thing. The wavelengths of visible light are on the atomic scale rather than millimeters to meters, so even a moderately sized telescope can capture great images. The primary mirror of the Hubble, for example, is only 2.4 meters in diameter. But that’s starting to change. Modern ground-based optical telescopes use multiple hexagonal mirrors rather than a single primary mirror, and even the James Webb Space Telescope has an array of seven mirrors so it wouldn’t be limited by the size of its launch rocket. The mirrors can be focused to a single detector, so we still don’t need to use interferometry. But what if we did anyway?

That’s the question of a new study.1 The authors propose a method known as Kernel Phase Interferometry (KPI), and while it’s not the same as radio interferometry, it has many of the same benefits.

Diagram showing how the JWST mirrors can be treated as an array of small apertures to process data via interferometry. Adelman, et al
The primary mirror of JWST can be virtualized as an array of apetures.

With regular interferometry, individual signals are correlated to create a single image. KPI, on the other hand, starts with a single image and creates a virtual array of individual signals through Fourier transformations. Once the virtual array is created, you can then use it to produce an image through correlation, just like we do with radio signals.

Most of the time this approach wouldn’t gain you anything. While radio interferometry can create high-resolution images, those images have artifacts due to the layout of the antennas. Using KPI on a high-resolution image would just create a different high-resolution image with artifacts. But one thing interferometry is particularly good at is isolating sources. As the authors show, using KPI on observations such as close binaries better distinguishes individual sources. This method would be particularly useful for observing Earth-sized planets closely orbiting Sun-like stars.

What’s neat about this method is that you don’t need to make new observations. The observations we currently have from telescopes such as JWST can be analyzed through KPI to create direct images of exoplanets and close binary stars. From one observation, we can get many observations thanks to this new approach.


  1. Adelman, Chelsea, et al. “First demonstration of kernel phase interferometry on JWST/MIRI: prospects for future planet searches around post main sequence stars.” Techniques and Instrumentation for Detection of Exoplanets XII. Vol. 13627. SPIE, 2025. ↩︎