Astronomers produce sharpest, deepest radio image yet of a patch of the night sky

An international team of astronomers led by researchers at Leiden University has produced the sharpest and most sensitive image yet made of the sky at low radio frequencies, using 200 hours of accumulated observations from the Low-Frequency Array, or LOFAR, a pan-European network of radio antennas coordinated from the Netherlands. The image covers a patch of northern sky known as ELAIS-N1 and resolves detail roughly 20 times finer than the deepest previous LOFAR image of the same field.

The work is described in a paper titled "Pushing the frontiers of low-frequency radio astronomy: An ultra-deep sub-arcsecond LOFAR image of ELAIS-N1," posted November 4 to the arXiv preprint server and accepted for publication in the journal Astronomy & Astrophysics. The lead author is Jurjen de Jong, a postdoctoral researcher at Leiden Observatory and the Netherlands Institute for Radio Astronomy, known as ASTRON. His co-authors include Leiden astronomer Reinout van Weeren and researchers at Durham University, ASTRON and the University of Edinburgh.

A sharper window on faint radio sources

The new image spans an area of 2.5 degrees by 2.5 degrees — about 25 times the area of the full moon as seen from Earth — and contains roughly 8 billion pixels. It was assembled from 200 hours of LOFAR observations made between 115 and 168 megahertz, producing about 400 terabytes of raw data, and resolves individual radio sources down to sub-arcsecond scales. The resulting catalogue lists 62,916 distinct radio sources, most of them active galactic nuclei — the bright cores of galaxies powered by supermassive black holes — and star-forming galaxies.

According to the preprint, the image reaches a point-source sensitivity of roughly 5.3 microjanskys per beam at sub-arcsecond resolution. For comparison, the standard wide-area LOFAR Two-metre Sky Survey, whose most recent data release covers 27 percent of the northern sky, typically reaches a sensitivity of about 83 microjanskys per beam at a much coarser 6-arcsecond resolution, according to ASTRON's description of that survey. The new ELAIS-N1 image is therefore both far deeper and roughly 20 times sharper than the maps that make up most of LOFAR's existing sky coverage, though the two products serve different purposes: one surveys the whole sky, the other probes a single small field in extreme detail.

The team also reports that it kept instrumental polarization leakage to about 0.12 percent, roughly half the typical level in the wide-area survey, and used 95 quasars from the European Space Agency's Gaia satellite to check and correct the image's astrometric accuracy.

From an 8-hour snapshot to a 200-hour stack

The new result builds directly on earlier work by much of the same team. A 2024 paper by de Jong and colleagues, published in Astronomy & Astrophysics, produced a sub-arcsecond image of the same ELAIS-N1 field using only 32 hours of data and reached a central noise level of about 14 microjanskys per beam — at the time described as the deepest sub-arcsecond LOFAR image yet made. The new paper uses the full 200 hours of data accumulated on the field, pushing the noise down to roughly 5.3 microjanskys per beam.

Achieving that resolution at low radio frequencies is difficult because Earth's ionosphere distorts incoming radio signals in ways that are especially disruptive at the long wavelengths LOFAR observes. Sub-arcsecond imaging additionally requires combining signals from LOFAR's international stations — located as far afield as the United Kingdom, Ireland, Germany, France, Poland and Latvia — to create the long baselines needed for high resolution, on top of the Dutch core stations near Exloo that anchor the network. According to the paper, improved calibration and data-reduction methods cut the computational cost of producing the image by more than a factor of ten, and the full 400-terabyte dataset was processed in a few months on the Spider computing platform at SURF, the Dutch national e-infrastructure cooperative.

"The entire data set could be processed on Spider, the high-performance data-processing platform hosted at SURF in the Netherlands, in just a few months," de Jong said, according to an account of the findings distributed by the Netherlands Research School for Astronomy.

Separating jets from star formation

Radio emission at these frequencies comes from two main sources that are often jumbled together in lower-resolution images: synchrotron jets launched by matter falling onto supermassive black holes, and radio waves generated by star formation. Because radio waves pass largely unimpeded through the dust that can hide star-forming regions from optical and infrared telescopes, the LOFAR data give astronomers a way to see star formation activity that other instruments miss. The sub-arcsecond resolution of the new image lets researchers distinguish compact black-hole-powered jets from the more diffuse emission produced by star formation within the same galaxy, something largely impossible in the standard 6-arcsecond LOFAR survey maps.

"With this sharp and sensitive view of our sky at low radio frequencies, we have a new window to study how these radio jets interact with their environment and how they evolve over cosmic time," said Reinout van Weeren, an associate professor at Leiden Observatory and a co-author of the paper.

The authors note that the data are most valuable in combination with other surveys of the same patch of sky — including optical, infrared and X-ray observations of the ELAIS-N1 field already made by other telescopes — which together let astronomers connect a galaxy's radio emission to its stellar populations, dust content and black hole activity.

A test case for the next generation of radio surveys

The authors frame the result as a proof of concept as much as a scientific catalogue. The reduced computational cost demonstrated in the paper, the team argues, shows that wide-area, sub-arcsecond imaging using many stacked observations is now practical rather than prohibitively expensive — a benchmark relevant to the planned LOFAR2.0 upgrade of the telescope network and, longer term, to the Square Kilometre Array, the next-generation international radio observatory now under construction in South Africa and Australia. Both projects aim to produce deep, high-resolution radio maps over far larger areas of sky than the single ELAIS-N1 field covered here.

The ELAIS-N1 catalogue and image have been made publicly available, following the pattern of earlier LOFAR survey data releases, so other research groups can combine them with their own observations. No independent, non-author expert reaction to the specific November paper was available at the time of writing; broader context on the LOFAR survey program, including the standard-depth data release referenced for comparison above, comes from ASTRON, the Dutch institute that built and coordinates the telescope network.

• 62,916 radio sources catalogued in a single 2.5-by-2.5-degree field
• Point-source sensitivity of about 5.3 microjanskys per beam at sub-arcsecond resolution
• 200 hours of LOFAR observations, producing roughly 400 terabytes of data and an 8-billion-pixel image
• Computational cost of producing the image cut by more than a factor of ten compared with earlier methods