JWST's JADES Survey Releases 1,253 Hours of Deep-Field NIRCam Imaging

Two small patches of sky—GOODS-S in the southern sky and GOODS-N in the north—have been observed by many major telescopes for decades. They contain the Hubble Deep Field, the Hubble Ultra Deep Field, and now one of JWST's largest surveys. On 22 January 2026, the JADES collaboration posted a preprint describing Data Release 5, the survey's full NIRCam imaging in these two fields.

JADES stands for JWST Advanced Deep Extragalactic Survey. It combines deep infrared imaging from JWST's Near Infrared Camera, NIRCam, with multiobject spectroscopy from NIRSpec. The new release is about the imaging side: the complete NIRCam dataset from JADES, plus co-reductions of 19 other JWST programs that have pointed at the same fields.

This is not a single-object discovery paper. It is a data release—an archive built to let astronomers study large samples of galaxies at once, including galaxies from the first few hundred million years after the Big Bang.

A 469-Square-Arcminute Map in Up to 18 Filters

The DR5 mosaics cover 469 square arcminutes with at least one NIRCam filter, split between GOODS-S and GOODS-N. Of that area, 250 square arcminutes have at least eight filters of coverage. The survey observed in up to 18 filters, each a slice of infrared wavelength. The eight core JADES filters cover 240 square arcminutes, a region large enough for accurate photometric redshifts—redshift estimates made by comparing an object's brightness across colors rather than by measuring a spectrum.

The color coverage matters. With eight or more filters, astronomers can make more reliable photometric-redshift estimates and, using medium-band filters such as F335M and F410M, help identify strong rest-optical emission lines common in galaxies at redshift greater than 3. The release also includes 65 square arcminutes with more than 12 filters, giving detailed color information in those patches.

The imaging is deep. The JADES guaranteed-time NIRCam imaging alone amounts to more than 800 JWST mission hours. Across all contributing programs, the DR5 mosaics incorporate 1,253 hours of short-wavelength NIRCam exposure and 1,070 hours of long-wavelength exposure in the two fields. JADES itself supplies 578 of those short-wavelength hours, affiliated programs add 211, and other programs supply 464.

Cleaning Artifacts and Tying the Mosaics to Gaia

Raw JWST images are not ready for science. NIRCam data include artifacts: cosmic-ray hits, hot pixels, persistence from earlier exposures, scattered light called wisps, and correlated readout noise. The JADES team describes a pipeline that goes beyond the standard JWST calibration steps.

They correct crosstalk between detector amplifiers. They fit and subtract 1/f noise, a low-frequency pattern that can stripe images. They build custom flat fields from the sky itself for the long-wavelength detectors. They fit and remove wisp templates. They mask artifacts by visually inspecting every exposure, using animations of dither sequences to separate detector-fixed features from real sky objects.

The result is a set of mosaics—large stitched images—that are deep and well characterized. The team validates photometric properties, the point-spread function (how a point source is blurred by telescope and detectors), and astrometric alignment (positions on the sky).

Combining data from many programs is difficult because the images were taken at different times, position angles, and dither patterns. JADES aligns exposures using a common astrometric reference, then groups them into subregions by program, epoch, and position angle for mosaicking onto a common pixel grid.

The astrometric solution is tied to Gaia DR3 stars, whose positions are known precisely. For 341 subregion pairs with more than 20 matched objects, 82 percent have median offsets better than 5 milliarcseconds. In GOODS-S, the rms residual of Gaia stars in the NIRCam data is 2.6 milliarcseconds. These are tiny angles: a milliarcsecond is roughly the width of a coin seen from several thousand kilometers away.

The release includes both individual subregion mosaics and combined mosaics. The individual groups preserve epoch information, which helps astronomers search for transients such as supernovae and for high-proper-motion sources such as brown dwarfs.

What DR5 Does Not Solve

The coverage is not uniform. Depth varies across the fields because different programs used different exposure times and dither patterns, and because artifacts had to be masked. Some rare flaws remain: isolated hot pixels, residual scattered light, diffraction spikes around bright stars, and background oversubtraction around very large galaxies. The paper also warns that multi-epoch stacks can produce misleading colors if an object varied in brightness between observations.

Astrometry is less certain in a few isolated edge regions that lack overlapping pointings. The data release closed in May 2025, so it does not include some later JWST programs in the same fields.

The paper presents DR5 as generally clean and a clear improvement over previous JADES releases, with these rare caveats. The GOODS fields already have decades of Hubble, Spitzer, Chandra, and other multi-wavelength observations. By placing nearly all available NIRCam imaging onto a common, well-tested grid, JADES gives astronomers a deep infrared layer to combine with that legacy.

The earliest galaxies are faint and remote. Finding them requires deep infrared imaging in many filters and a well-understood telescope response. DR5 is that foundation: a 469-square-arcminute map of the sky that turns years of JWST observing into a public tool for studying the first galaxies.