---
title: JADES Online Database Evolution
url: https://www.emergentmind.com/topics/jades-online-database
type: topic
---

# JADES Online Database Evolution

Searching arXiv for the cited JADES data release papers to ground the article in the current literature.
First, I’ll locate the initial JADES imaging release and later DR4/DR5 releases that describe the online database and related data products.
The **JADES Online Database** is the public online access layer associated with successive data releases of the **JWST Advanced Deep Extragalactic Survey (JADES)**. Across releases, it has denoted related but not identical interfaces: an initial **interactive FitsMap** environment for visually browsing deep JWST imaging and overlaid catalog data in the Hubble Ultra Deep Field and GOODS-South, a later **graphical interface** for rapid selection and inspection of the complete JADES DR4 NIRSpec spectroscopic sample, and, in DR5, the public-facing repository and interactive web environment for large imaging, photometric, and value-added stellar-population products in GOODS-S and GOODS-N [2306.02466; 2510.01034; 2601.15954; 2601.15956; 2605.21599]. In that sense, the term refers less to a single fixed database architecture than to a progressively expanded discovery layer spanning mosaics, spectra, catalogs, posterior summaries, and download services.

## 1. Release evolution and scope

The role of the JADES Online Database changed substantially between the initial imaging release, DR4 spectroscopy, and DR5 imaging and catalogs. In the initial release, it was not described as a searchable database in the classic query-engine sense, but as a public online access resource built around an interactive **FitsMap** website for browsing deep imaging and catalog overlays in multiple filters [2306.02466]. In DR4, it became a **new graphical interface** intended to enable “quick selection and browsing” of the final JADES spectroscopic release [2510.01034]. In DR5, the online database became the public-facing repository or interactive access environment for the full NIRCam imaging products, the superseding photometric catalogs, and the stellar-population catalogue [2601.15954; 2601.15956; 2605.21599].

| Release context | Online-database role | Principal content |
|---|---|---|
| Initial JADES imaging release | Interactive FitsMap visualization | 9-band NIRCam mosaics, 5 additional JEMS bands, 23-band photometric catalogs, photometric redshifts |
| JADES DR4 spectroscopy | Graphical search, filter, inspect, and download interface | 5,190 NIRSpec/MSA targets, spectra, redshifts, \(S/N>5\) emission-line flux catalogues |
| JADES DR5 imaging and catalogs | Public-facing repository and interactive web interface | NIRCam mosaics, DR5 photometric catalogs, DR5 stellar population catalogue |

A common misconception is to treat the database as a single monolithic product. The release history indicates instead that the “database” is release-dependent. In the earliest phase, it primarily exposed imaging and associated source overlays. By DR4 and DR5, it functioned as a search-and-discovery layer over full science archives, with links to HLSP products and catalog exports [2306.02466; 2510.01034; 2601.15956].

## 2. Initial FitsMap implementation in the first JADES imaging release

The first public incarnation of the JADES Online Database accompanied the initial NIRCam imaging data release for the Hubble Ultra Deep Field and portions of GOODS-South. That release provided **9 filters of infrared imaging** over approximately **25 arcmin\(^2\)**, based on **87 on-sky dual-filter hours of exposure time**, and supplied carefully constructed **9-band mosaics** of the JADES bands together with matching reductions of **5 additional bands** from the **JWST Extragalactic Medium-band Survey (JEMS)** [2306.02466].

Combined with existing HST imaging, the release provided **23-band space-based photometric catalogs** and **photometric redshifts for \(\approx 47{,}500\) sources**. The public release was hosted at **https://archive.stsci.edu/hlsp/jades**, which also included details of the catalog contents. The associated online interface was described explicitly as an interactive **FitsMap** website created “to promote broad engagement with the JADES survey” and to provide an interface for professional researchers and the public to experience the JWST datasets [2306.02466].

The initial functionality was visual rather than database-like in the classical sense. At **http://jades.idies.jhu.edu/**, the release provided a link to a FITSmap visualization “where one can pan and zoom in multiple filters, and use overlays to present the associated catalog data” [2306.02466]. The website therefore served as an intuitive front end to the public imaging and catalog products. It enabled multi-filter inspection of the JWST mosaics and source-level contextualization through overlays, but it was not presented as a metadata-driven query system. This distinction matters historically, because later JADES releases introduced explicit filtering, table export, and browsing of complete spectroscopic and photometric samples.

## 3. DR4 transformation into a spectroscopic search and discovery interface

In **JADES Data Release 4**, the JADES Online Database was launched as a **new graphical interface** designed to make the final NIRSpec spectroscopic release easy to **search, filter, inspect, and download** [2510.01034]. It was intended to sit alongside the HLSP products and the JADES website, and it provided rapid access to the **complete spectroscopic sample** rather than only a small subset.

The underlying release comprised **5,190 NIRSpec/MSA targets** observed across the GOODS fields. The observations covered **\(0.6\text{--}5.5\,\mu\mathrm{m}\)** using both the low-dispersion **prism** mode, with **\(R=30\text{--}300\)**, and the **medium-resolution gratings**, with **\(R=500\text{--}1{,}500\)**. The paper reported **3,297 robust redshifts** from these 5,190 targets, spanning **\(z=0.5\)** up to **\(z=14.2\)**, including **974 galaxies at \(z>4\)** [2510.01034].

The database was organized around the main DR4 data products:

- fully reduced **1-D spectra**
- fully reduced **2-D spectra**
- **3-pixel** and **5-pixel** 1-D extractions
- products with **slit-loss corrections**
- products with **background subtraction**
- **redshift catalogues**
- **\(S/N>5\)** emission-line flux catalogues for prism and grating spectra

The release emphasized consistent ordering across catalog products, so that the same objects appeared in the same order in target metadata, redshift tables, line-flux measurements, and spectrum files. A plausible implication is that the database was designed not merely for visual inspection but for low-friction traversal between heterogeneous but linked spectroscopic products.

The database interface supported search by **sky coordinates** with a user-defined search radius, **redshift range**, **redshift quality flags**, **absolute UV magnitude**, and **emission-line fluxes**, specifically including **H\(\alpha\)** and **H\(\beta\)** [2510.01034]. For matched targets, it displayed target-selection information, available NIRCam images, **R1000 fits of H\(\alpha\) and H\(\beta\)**, the prism spectrum, a link to the **FITS_map** / NIRCam image visualization, and a link to the HLSP portal for downloads. Users could export the displayed results as a **.csv** file or download the **entire JADES multi-HDU FITS table** using a “Download table” button.

The redshift catalogues incorporated explicit quality flags:

- **A**: Redshift from at least one emission line in the medium-resolution grating.
- **B**: Redshift from two or more prism emission lines.
- **C**: Redshift from the continuum, or from the continuum and a single prism emission line.
- **D**: Tentative, from visual inspection.
- **E**: No redshift.

For the final redshift estimate, strong grating detections took precedence, and the final redshift combined prism and grating information when available [2510.01034]. This gave the database an internal quality hierarchy that is scientifically consequential for subsample construction.

The DR4 release also exposed concrete file conventions, including

```text
hlsp_jades_jwst_nirspec_<TIER>-<NIRSpec_ID>_<filter>-<grating>_v1.0-spec1d.fits
hlsp_jades_jwst_nirspec_<TIER>-<NIRSpec_ID>_<filter>-<grating>_v1.0-spec2d.fits
```

and described the main catalogue as the **Obs_info** master table, with columns including unique source identifiers, programme ID, tier names, NIRSpec and NIRCam IDs, observation dates, sky coordinates, intra-shutter offsets, field, selection method, target priority, gold-sample flags, observation assignments, exposure times, redshifts and redshift flags, previous literature redshifts, and photometric quantities such as \(M_{\rm UV}\) [2510.01034].

The same paper also gave exact calibration relations relevant to the spectroscopic products, including

$$
\lambda^{corr} = \lambda+[S(\lambda,x)+S^{ref}(\lambda,x)]\delta\lambda(\lambda)
$$

$$
S(\lambda,x)= - \frac{\Delta \lambda (\lambda,x)}{\delta\lambda_{prism}(\lambda)-\delta\lambda_{grat}(\lambda)}
$$

and

$$
\Delta v = c \left(\frac{\lambda_{\rm [OIII]}}{ \lambda^{\rm rest}_{\rm [OIII]}(z_{\rm HeI}+1)}- 1\right).
$$

These expressions belong to the reduction and calibration framework rather than to the interface itself, but they define the technical provenance of the spectra exposed through the database.

## 4. DR5 NIRCam imaging archive and data model

In **JADES Data Release 5**, the online database served as the public-facing repository for the JADES NIRCam imaging products in GOODS-S and GOODS-N, with the DR5 imaging paper providing the technical foundation for what those products contain and how they should be used [2601.15954]. The release described deep and well-characterized mosaics in up to **18 filters** covering **469 arcmin\(^2\)**, with **245 arcmin\(^2\)** in GOODS-S and **224 arcmin\(^2\)** in GOODS-N. Multi-band coverage included **240 arcmin\(^2\)** with the 8 “core JADES” filters, **323 arcmin\(^2\)** with at least the 6 wide bands, **65 arcmin\(^2\)** with more than 12 filters, and the paper also noted that **250 arcmin\(^2\)** had at least 8 filters of coverage [2601.15954].

This release contained the full NIRCam imaging of JADES, over **800 JWST mission hours**, together with co-reductions of **19 other programs** in the same framework. A central design feature was the release of two complementary classes of mosaics: mosaics for **individual programs or epochs**, and **combined mosaics** that coadded all available programs into the deepest possible image for each band. The per-program or per-epoch mosaics preserved time-domain information and observing-uniformity, while the combined mosaics maximized depth for source detection and photometry [2601.15954].

The data products were the output of a custom reduction pipeline built on top of the **JWST Calibration Pipeline**, using pipeline version **1.14.0** with the **May 2024 context map**. Stage 1 fit raw ramps into count-rate images, Stage 2 performed photometric calibration, WCS assignment, and background subtraction, and Stage 3 resampled and combined the data into mosaics. Additional custom steps included **crosstalk subtraction**, **custom bad-pixel masking**, **persistence mitigation**, **wisp subtraction**, **custom LW sky flat fields**, **1/f noise removal**, and carefully tuned **outlier rejection** [2601.15954].

Several of these calibrations are directly relevant to scientific use of the online archive. The paper stated that crosstalk-induced signal is of order \(10^{-5}\) to \(10^{-4}\) of the original signal, but can matter near the detection limit in deep mosaics. Saturated-pixel persistence was tracked by propagating saturation masks into later exposures within **1800–3600 s** depending on channel. Residual flat-field structures were reduced to the \(\sim 1\%\) RMS level, and in the most affected medium bands the remaining background pattern was estimated at \(\lesssim 2\%\) on **0.3″** scales [2601.15954].

The 1/f-noise correction was formalized as

$$
\delta_{m, x, y} = a_{m, y} + b_{x} + c_m + \epsilon,
$$

where \(\delta_{m, x, y}\) is the residual in pixel \(x, y\) of amplifier \(m\), \(a_{m,y}\) is a row term, \(b_x\) is a column term, \(c_m\) is an amplifier pedestal, and \(\epsilon\) is uncorrelated Poisson noise. The model has **10,244 parameters** and was fit with stochastic gradient descent in **TensorFlow** on GPUs [2601.15954].

Astrometric alignment was tied to a JWST-based astrometric reference catalog anchored to **Gaia DR3** rather than to the older HST-based frame used in earlier JADES releases. The reference catalog was built from **1398 GOODS-S exposures** and more than **200,000 input detections**, yielding **11,436 unique objects** with multiple detections plus **477 Gaia detections**. The fitting used the robust loss function

$$
L = \frac{c^2}{1+c^2/s^2}, \quad c^2 = \frac{(\Delta x)^2 + (\Delta y)^2}{\sigma^2},
$$

with \(s=4\), an initial \(\sigma = 15\) mas that later dropped to **7.5 mas**, and increasing weight on the Gaia stars. The reported Gaia residuals were about **2.6 mas rms**, with typical internal alignment better than a few mas and residual distortion maps no worse than **2.5 mas rms** within each detector [2601.15954].

The mosaics were packaged as multi-extension FITS files with key image extensions **SCI**, **ERR**, **WHT**, **EXP**, and **NIM**. **SCI** is the science image in MJy/sr; **ERR** contains the uncertainty estimate, including read noise, background and source Poisson noise, and flat-field uncertainty, but not correlated noise from resampling; **WHT** gives the inverse-variance-like weight image; **EXP** stores the summed exposure time per pixel; and **NIM** stores the number of exposures contributing to each pixel. The subregion mosaics used special **NIM** values, with \(NIM=-2\) for diffraction spike masks and \(NIM=-4\) for artifact masks [2601.15954].

A 32-bit **bithash** image encoded which JWST programs contribute to each sky pixel:

$$
1 \ll b
$$

was added for each contributing program bit \(b\), and the corresponding decoder table was included in the FITS file. This provenance-tracking mechanism is central to the DR5 archive, because it lets users reconstruct which subregions or programs contribute to any given source position [2601.15954].

## 5. DR5 photometric catalog and interactive source-level access

The **JADES Data Release 5 photometric catalogs** were publicly released through the Mikulski Archive for Space Telescopes and an **interactive web interface**, and the paper explicitly described them as superseding previous JADES photometric releases [2601.15956]. The public photometric release was designed to provide a unified, source-by-source and band-by-band database for GOODS-North and GOODS-South.

The catalogs were constructed from **35 space-based imaging mosaics** obtained with **JWST/NIRCam**, **JWST/MIRI**, **HST/ACS**, and **HST/WFC3**, combining approximately **1250 hours of JADES imaging** with extensive additional public JWST and HST observations in the GOODS fields [2601.15956]. The full filter set included HST/ACS bands **F435W, F606W, F775W, F814W, F850LP**; HST/WFC3 bands **F105W, F125W, F140W, F160W**; JWST/NIRCam bands **F070W, F090W, F115W, F150W, F162M, F182M, F200W, F210M, F250M, F277W, F300M, F335M, F356W, F410M, F430M, F444W, F460M, F480M**; and JWST/MIRI bands **F560W, F770W, F1000W, F1280W, F1500W, F1800W, F2100W, F2550W** [2601.15956].

Source detection and deblending used custom signal-to-noise-based algorithms optimized for the depth, resolution, and complex PSF structure of JWST imaging. The workflow consisted of building SNR mosaics for detection and deblending, detecting blended segments, deblending them into individual sources, interactively curating the catalog, and measuring centroids, shapes, apertures, and photometry [2601.15956]. Detection used inverse-variance-weighted SNR mosaics from long-wavelength NIRCam filters, while deblending used a higher-resolution short-wavelength NIRCam SNR stack combining **F182M, F200W, F210M** where available.

The segmentation and deblending algorithm used `photutils.detect_sources`, morphological cleanup, and local peak assignment. The initial detection threshold was **SNR > 1.5** over at least **1 pixel**, followed by a second `detect_sources` pass at **SNR > 3.5** over **4 pixels**. Deblending used local maxima in the deblending SNR image with a disk filter of radius **3 pixels**, and the dominance criterion was defined by

$$
\xi = 0.5.
$$

Pixels were then assigned deterministically in descending SNR order to the dominant peaks. A second simplified detection pass in isolated unmasked regions recovered faint compact sources using `detect_sources` with **SNR > 3** and minimum area **5 pixels** [2601.15956].

Manual curation was performed interactively using **FitsMap**, with a custom selection tool that let collaborators draw polygons around problematic fragments and mark them for merging or deletion. The paper stated that this affected **less than 1%** of the originally detected objects by number [2601.15956]. This is an important operational detail: the online environment was not only a public exploration interface but also part of catalog production and quality control.

The per-source catalog structure included a **FLAG** HDU and a **SIZE** HDU. The **FLAG** table included source ID, RA, DEC, per-band bad-pixel flags, exposure time, inverse-variance of the sky background, bright-neighbor flag, parent-segmentation ID, and program bithash. The **SIZE** table included centroid and barycenter positions, bounding-box limits, Gaussian shape parameters, Kron radius, FWHM, and Gini coefficient [2601.15956].

Source shapes were estimated through a new fast **2D Gaussian regression** method. The fitted profile was

$$
z(x,y) = K \exp[-(ax^2 + 2bxy + cy^2)]
$$

or equivalently

$$
z(x,y) = \exp[\alpha x^2 + 2\beta xy + \gamma y^2 + \delta].
$$

The paper then gave the iterative weighted-regression system used to solve for \(\alpha_k,\beta_k,\gamma_k,\delta_k\), from which the Gaussian semimajor axis \(A\), semiminor axis \(B\), and position angle \(\theta\) were derived for Kron apertures [2601.15956].

The photometric products included forced circular-aperture photometry at radii

$$
r = [0.1'', 0.15'', 0.25'', 0.3'', 0.35'', 0.5'']
$$

labeled **CIRC1** through **CIRC6**, plus **CIRC0**, the aperture that encloses **80% of the PSF energy** in that band. Background-subtracted variants used a circular annulus at **\(1.5'' < r < 1.55''\)**. The release also included native and common-PSF **Kron photometry**, with standard \(k=2.5\) and short \(k=1.4\) apertures, and a dedicated **curve-of-growth** product with one HDU per band in **GROWTH** and **GROWTH_CONV** [2601.15956].

A major methodological contribution was the treatment of correlated noise in heterogeneous JWST mosaics. The sky-noise component of the flux uncertainty was modeled as

$$
\sigma = \sigma_1 N^\beta,
$$

where \(N\) is aperture linear size in pixels, \(\sigma_1\) is the single-pixel uncertainty, and \(1 \le \beta \le 2\). The release built uncertainty mosaics from **4,000,000 random apertures per filter**, computed on a low-resolution grid at **50:1** with the full mosaic resolution. For NIRCam, the pixel-level regression model was written as

$$
\sigma_{ij} = A P_j N_{ij}^{\beta_j}
$$

or

$$
y_{ij} = p_j + \alpha + \beta_j x_{ij},
$$

with the solution expressed as

$$
\vec{\phi} = \mathbf{M}^{-1}\vec{z}.
$$

The catalogs reported two uncertainty estimates: **`_e`**, the model-based uncertainty from the random-aperture regression, and **`_ei`**, the direct uncertainty from the JWST pipeline ERR image. The paper emphasized that **`_ei`** underestimates the real error because it does not account for correlated noise, while **`_e`** does [2601.15956].

Photometric redshifts were computed with **EAZY**, with two separate products: **PHOTOZ**, based on **CIRC1** \((r=0.1'')\) circular photometry on the native mosaics, and **PHOTOZ_KRON**, based on **KRON_CONV** ellipsoidal photometry on the common-PSF mosaics. The redshift PDF was defined as

$$
p(z) = \exp[-\chi^{2}(z)/2]
$$

renormalized so \(\int p(z)\,dz = 1\). Performance against spectroscopic compilations for about **12,500 sources** gave NMAD scatters of **0.028–0.029** for GOODS-S and **0.044–0.045** for GOODS-N, with

$$
\sigma_{\mathrm{NMAD} = 1.48 \times \mathrm{median}\left(\left|\frac{\delta z - \mathrm{median}(\delta z)}{1+z_s}\right|\right)
$$

and \(\delta z = z_s - z_p\) [2601.15956].

Public access was through MAST and the interactive **FitsMap-based** JADES online database at **https://jades.idies.jhu.edu**, with DOI **10.17909/8tdj-8n28** for the MAST High Level Science Product [2601.15956].

## 6. DR5 stellar-population catalogue and value-added physical inference

The DR5 **stellar population catalogue** extended the JADES Online Database beyond directly observed imaging and photometry into a value-added, Bayesian catalogue of inferred physical galaxy properties for GOODS-N and GOODS-S [2605.21599]. The modelling was run on approximately **500,000** detected galaxies or sources in the DR5 photometric catalogue, with the scientifically most useful galaxy sample being roughly **350,000** galaxies at **\(z = 1–9\)** where the depth and wavelength coverage yield robust stellar-population constraints.

The catalogue provided Prospector-based posterior constraints and summary quantities, including observed photometric measurements and band-coverage metadata, inferred physical parameters, posterior distributions and summary percentiles, rest-frame magnitudes and colors, and flags for spectroscopic-redshift availability, MIRI coverage, AGN/LRD classifications, and fit quality [2605.21599]. The physical parameters included **stellar mass**, **star-formation rate**, **full star-formation history**, **dust attenuation**, **stellar metallicity**, **gas-phase metallicity**, **mid-IR dust emission parameters**, **AGN contribution**, **photometric redshift**, and derived quantities such as rest-frame magnitudes, colors, \(M_{\rm UV}\), and the UV slope \(\beta\).

The underlying data combined deep JWST/NIRCam and MIRI imaging with ancillary multi-wavelength data. The release emphasized contiguous coverage from the UV through the mid-IR, with up to **35 filters in GOODS-S** and **26 filters in GOODS-N** [2605.21599]. This broad baseline was identified as critical for breaking redshift, age, dust, and metallicity degeneracies in SED fitting.

The inference used **Prospector**, built on **FSPS**, with **nested sampling** via `dynesty` and accelerated SPS evaluation using the **Parrot** neural-network emulator. The adopted stellar-population ingredients were **MIST isochrones**, the **MILES stellar library**, and a **Chabrier IMF**. The model had **18 free parameters**, grouped into redshift and stellar populations, dust emission, dust attenuation, and AGN emission [2605.21599].

The non-parametric star-formation history used Prospector’s **continuity** model with **7 age bins** and piecewise-constant SFR in each bin. Nebular emission was included self-consistently through **FSPS/Cloudy-style prescriptions**. Dust attenuation used a two-component **Charlot & Fall–style** model with a variable diffuse-law slope \(n\), and mid-IR dust emission was modeled with the **Draine et al.** templates parameterized by \(Q_{\rm PAH}\), \(U_{\rm min}\), and \(\log_{10}(\gamma_e)\). AGN mid-IR emission used **CLUMPY torus** templates with \(\log_{10}(f_{\rm AGN})\) and \(\log_{10}(\tau_{\rm AGN})\) [2605.21599].

A major methodological feature was the evolving **star-forming main sequence prior** for the star-formation history. Prospector’s continuity SFH uses adjacent-bin SFR ratios,

$$
\log_{10}\left(\frac{\mathrm{SFR}_i}{\mathrm{SFR}_{i+1}}\right),
$$

assigned a Student’s-\(t\) prior,

$$
\mathrm{PDF}(x\mid \nu,\mu,\sigma)= \frac{\Gamma\!\left(\frac{\nu+1}{2}\right)} {\sqrt{\nu\pi}\,\sigma\,\Gamma\!\left(\frac{\nu}{2}\right)} \left[ 1+\frac{1}{\nu}\left(\frac{x-\mu}{\sigma}\right)^2 \right]^{-\frac{\nu+1}{2}. \right]
$$

The star-forming main sequence itself was parameterized as

$$
\log_{10}(\mathrm{SFR}\,[\mathrm{M_\odot/yr}]) = m - m_0 + a_0 r - a_1 \left[\max(0, m - m_1 - a_2 r)\right]^2,
$$

with \(r = \log_{10}(1+z)\), \(m = \log_{10}(M_*/10^9\,\mathrm{M_\odot})\), and best-fit parameters \(m_0 = 0.5\), \(m_1 = 0.4\), \(a_0 = 2.0\), \(a_1 = 0.3\), and \(a_2 = 2.5\). The adopted prior-strength parameters were \(\nu = 2\) and \(\sigma = 0.5\) [2605.21599].

Validation used **16,320 galaxies** with secure spectroscopic redshifts, yielding \(\sigma_{\rm NMAD} = 0.026\) and outlier fraction \(\eta \approx 7\%\). Comparison of **10 Myr-averaged** Prospector SFRs against H\(\alpha\)-based SFRs from NIRSpec grating spectroscopy gave a median residual of \(-0.017^{+0.380}_{-0.349}\) dex. The paper further reported that adding MIRI has negligible impact on stellar masses but improves AGN and dust constraints, while adding NIRCam medium bands improves redshifts and reduces stellar-mass uncertainties by about **0.11 dex** on average [2605.21599].

For practical use, the paper recommended the fiducial selection **`use_phot = 1`**, corresponding to galaxies satisfying \(\mathrm{SNR}_{\mathrm{F444W} \ge 3\), \(\mathrm{N}_{\mathrm{bands,NIRCam} \ge 6\), `flag_good_redshift_50`, and `flag_good_mass_dex08` [2605.21599]. The catalogue columns were grouped into observed properties and coverage, inferred physical parameters, rest-frame magnitudes and colors, quality and data-coverage flags, and AGN/LRD identifiers.

## 7. Access routes, scientific use cases, and interpretive boundaries

The JADES Online Database is distributed through multiple public access points. For the initial imaging release, the public products were at **https://archive.stsci.edu/hlsp/jades**, with FitsMap visualization linked from **http://jades.idies.jhu.edu/** [2306.02466]. For DR4 spectroscopy, the release provided the **JADES website data page** at `https://jades-survey.github.io/scientists/data.html`, the **DR4 page** at `https://jades.herts.ac.uk/DR4`, the **online database/search page** at `https://jades.herts.ac.uk/search/`, the **HLSP archive** at `https://archive.stsci.edu/hlsp/jades`, the **MAST HLSP portal** at `https://mast.stsci.edu/hlsp`, the **general JADES site** at `https://jades-survey.github.io`, and the MAST portal URL `https://mast.stsci.edu/portal/Mashup/Clients/Mast/Portal.html` [2510.01034]. For DR5 photometry, the interactive FitsMap web interface remained available at **https://jades.idies.jhu.edu** alongside the MAST HLSP distribution [2601.15956].

Scientifically, the database supports several distinct but linked workflows. In spectroscopy, it supports rapid selection of subsamples by sky position, redshift, quality, luminosity, and line fluxes, with immediate inspection of associated spectra and imaging [2510.01034]. In imaging, it supports deep photometry, high-redshift source selection, variability studies, transient searches, proper-motion analyses, and provenance-aware interpretation of stacked mosaics [2601.15954]. In photometric and stellar-population work, it supports source-level measurement, uncertainty propagation, photometric-redshift estimation, posterior-based physical inference, and large-sample statistical studies of galaxy growth, quenching, and stellar-mass assembly [2601.15956; 2605.21599].

Several interpretive boundaries recur across the releases. First, the initial online database was a visualization interface rather than a classical searchable database [2306.02466]. Second, the DR4 slit-loss corrections were optimized for compact sources and may be inaccurate for extended galaxies [2510.01034]. Third, the DR5 1/f-noise subtraction was described explicitly as a pragmatic approximation that can oversubtract around very large bright galaxies or residual diffuse structure [2601.15954]. Fourth, the DR5 photometric catalogs report both `_e` and `_ei` uncertainties, with only `_e` intended to capture correlated-noise behavior in resampled mosaics [2601.15956]. These caveats delimit responsible use of the database products and are part of the scientific meaning of the archive itself.

Taken together, the JADES Online Database constitutes an evolving research infrastructure for the GOODS-S and GOODS-N deep fields. Its core function is to connect calibrated JWST and HST imaging, NIRSpec spectroscopy, photometric catalogs, and posterior-based stellar-population inferences through public interfaces that support both interactive inspection and direct archive access. The progression from a visual FitsMap browser to a spectroscopic search interface and then to a broader imaging and value-added catalog ecosystem suggests an increasingly integrated model of survey dissemination, in which discovery, provenance, and download are treated as parts of the same scientific workflow [2306.02466; 2510.01034; 2601.15954; 2601.15956; 2605.21599].

Source: https://www.emergentmind.com/topics/jades-online-database