Rubin Observatory DP1 Commissioning Data

• Rubin Observatory DP1 is a commissioning-era data release offering raw and calibrated images, coadds, and catalogs to validate LSST science pipelines.
• It spans ~15 deg² over seven fields with 2.3 million objects from 1792 exposures, supporting investigations from galaxy evolution to variable star studies.
• DP1 serves as a pathfinder for LSST operations by testing data processing, calibration, and classification methods in a realistic early-science environment.

Rubin Observatory Data Preview 1 (DP1) is the first data release based on real observations from the NSF DOE Vera C. Rubin Observatory. It comprises raw and calibrated single-epoch images, coadds, difference images, detection catalogs, and ancillary products derived from 1792 optical and near-infrared exposures acquired over 48 distinct nights with the Rubin Commissioning Camera, LSSTComCam, in late 2024. DP1 covers approximately 15 deg$^2$ across seven non-contiguous fields in the six broad bands $ugrizy$, contains approximately 2.3 million distinct astrophysical objects, is approximately 3.5 TB in size, and is served to data-rights holders through the Rubin Science Platform (RSP) (Team et al., 24 Mar 2026).

1. Programmatic context

DP1 belongs to Rubin’s Early Science Program and occupies a specific position in the observatory’s staged preview sequence. In the Google-cloud-hosted Interim Data Facility, DP0 used simulated DESC DC2 data, DP1 was planned to offer commissioning-camera data, and DP2 was planned to offer LSSTCam commissioning data. Within that sequence, DP1 is the first preview based on real Rubin commissioning observations rather than simulations, and it was explicitly intended both for early science and for exercising the Rubin Science Platform, Qserv, and Butler under realistic conditions (O'Mullane et al., 2021).

The role of DP1 is simultaneously scientific, technical, and pedagogical. It provides a miniature commissioning-era analogue of a future LSST Data Release, while also exposing the LSST Science Pipelines, the Rubin data model, and the RSP interfaces to science users. The official DP1 overview presents it as a small dataset relative to future LSST releases, but one with sufficient quality and diversity to support early investigations in galaxy evolution, stellar populations, variability, and Solar System science (Team et al., 24 Mar 2026).

This institutional positioning is important because DP1 is not simply a small survey. It is a commissioning dataset deliberately used to validate end-to-end processing, catalog serving, and interactive analysis workflows before full operations. A plausible implication is that DP1 should be understood less as a uniform legacy survey product than as a calibrated pathfinder for the future LSST ecosystem.

2. Observing campaign and field layout

DP1 is based on LSSTComCam, a 144 Mpix commissioning camera occupying the central raft position of the future LSSTCam focal plane. It contains one $3\times3$ raft of nine $4\mathrm{k}\times4\mathrm{k}$ CCDs, has the same plate scale as LSSTCam, and covers about $0.5$ deg$^2$ per pointing. The selected DP1 observations were acquired between 2024-10-24 and 2024-12-11, with 30 s single exposures in the Rubin $ugrizy$ filter system (Team et al., 24 Mar 2026).

The seven DP1 fields were chosen to stress different scientific and processing regimes, including deep extragalactic imaging, crowded stellar systems, low-latitude Galactic structure, low-ecliptic-latitude Solar System work, and diffuse nebulosity.

Field	Exposures	Primary use case
47_Tuc	72	Crowded globular cluster
ECDFS	855	Deep extragalactic field
EDFS_comcam	272	Deep extragalactic field
Fornax_dSph	42	Crowded dwarf spheroidal
Rubin_SV_095_-25	292	Low Galactic latitude
Rubin_SV_38_7	159	Low ecliptic latitude
Seagull	100	Nebulosity and complex backgrounds

The field strategy was not identical to the planned LSST Wide-Fast-Deep cadence. Most fields used translational dithers and small rotator dithers, while Rubin_SV_38_7 used a $2\times2$ grid of LSSTComCam pointings separated by $0.65^\circ$ to optimize multi-night moving-object linking. The filter exchanger could hold only three filters at a time, so band coverage is field-dependent and temporally non-uniform (Team et al., 24 Mar 2026).

ECDFS became the deepest and most densely sampled DP1 field. Several later studies therefore concentrated on it, either because it had the most observations for time-domain work or because it was the deepest field for low-surface-brightness and classification studies (Carlin et al., 30 Jun 2025, Gatto et al., 26 Mar 2026).

3. Processing architecture, calibration, and delivered products

DP1 was processed with LSST Science Pipelines v29.1 and delivered through both the Butler and VO-style services in the RSP. The image hierarchy includes raw, visit_image, deep_coadd, template_coadd, and difference_image. Each calibrated image product carries science, variance, and mask planes together with PSF and WCS metadata (Team et al., 24 Mar 2026).

The single-epoch processing sequence includes Instrument Signature Removal, background subtraction, PSF modeling, astrometric calibration, and photometric calibration. The major calibration and modeling components are explicit in the DP1 documentation: Piff for PSF modeling, gbdes for global astrometry, FGCM for global photometric calibration, the Monster calibration catalog for reference photometry and astrometry, SCARLET Lite for coadd deblending, GAaP for PSF-homogenized color measurements, MultiProFit for Sérsic fitting, and decorrelated Alard-Lupton image differencing for DIA products. Solar System processing further uses Sorcha, mpsky, HelioLinC3D, link_purify, and orbit fitting with find_orb (Team et al., 24 Mar 2026).

The catalog system is correspondingly layered. Source contains single-visit detections. Object contains coadd-based astrophysical objects merged across bands. ForcedSource provides forced PSF photometry at Object positions on direct and difference images. DiaSource contains $\ge 5\sigma$ detections on difference images, while DiaObject groups non-moving, stationary associations of DiaSource within $ugrizy$0. ForcedSourceOnDiaObject supplies visit-level forced photometry for time-variable sources. SSObject and SSSource serve moving-object analyses, while Visit and CcdVisit record visit- and detector-level metadata (Team et al., 24 Mar 2026).

DP1 also includes survey property maps in HealSparse format and HiPS image products for visual exploration. The RSP exposes these holdings through the Portal, Notebook, and API aspects, with TAP/Qserv for catalog access, SIAv2 and SODA for image discovery and cutouts, Butler for dataset retrieval, and WebDAV for user storage. This architecture places DP1 squarely inside the intended Rubin operational model rather than treating it as a stand-alone file release (Team et al., 24 Mar 2026).

4. Data quality, performance, and systematic limits

Across all bands and fields, the median point-spread-function FWHM in DP1 is approximately $ugrizy$1, with the sharpest images reaching about $ugrizy$2. In the deepest field, ECDFS, the $ugrizy$3 coadd point-source depths are $ugrizy$4, $ugrizy$5, $ugrizy$6, $ugrizy$7, $ugrizy$8, and $ugrizy$9. Other fields are no more than 2.2 magnitudes shallower in any band where they have nonzero coverage (Team et al., 24 Mar 2026).

Photometric repeatability for withheld bright unresolved stars is at the millimagnitude level in $3\times3$0, and internal astrometric repeatability is described as $3\times3$1 mas for isolated stars, with absolute offsets versus Gaia mostly $3\times3$2 mas in $3\times3$3-band in ECDFS. Injection tests in ECDFS show approximately $3\times3$4 completeness for galaxies at $3\times3$5–25.2 and $3\times3$6 completeness around $3\times3$7–26.0, while difference-image injections show $3\times3$8 completeness for point sources at $3\times3$9 and $4\mathrm{k}\times4\mathrm{k}$0 completeness at $4\mathrm{k}\times4\mathrm{k}$1 (Team et al., 24 Mar 2026).

The release also exposes commissioning-era limitations. Documented LSSTComCam artifacts include vampire pixels, phosphorescent patches, crosstalk residuals, severe bleed trails, and ITL dips. Differential chromatic refraction is visible but not yet corrected in the delivered products. In crowded environments, deblending can fail or be skipped. In 47 Tuc, the coadd-object catalog shows a sharp falloff inside approximately 28 pc of the cluster center, while forced photometry on DIAObjects recovers sources down to approximately 14 pc, albeit with significant systematics in the innermost region (Wainer et al., 4 Jul 2025).

Star-galaxy separation illustrates both the strengths and the limits of the early catalogs. In ECDFS, the DP1 refExtendedness morphology flag degrades seriously at faint magnitudes, whereas Random Forest models using multi-band photometry outperform morphology-only classification; colors involving the $4\mathrm{k}\times4\mathrm{k}$2-band are especially informative, and configurations using all six LSST bands keep galaxy contamination negligible across almost the full range $4\mathrm{k}\times4\mathrm{k}$3 mag probed in that study (Gatto et al., 26 Mar 2026). This suggests that DP1 photometry is already strong enough for sophisticated classification, but that morphology-only use of the coadd catalogs should be treated cautiously in the faint regime.

5. Time-domain and moving-object science

A central function of DP1 is to expose Rubin’s time-domain data model in a realistic commissioning setting. The cadence differs from the expected LSST survey cadence, but the concentration of many visits over less than two months makes some DP1 fields, especially ECDFS, unusually well suited to short-period variability searches. One ECDFS study used 217 epochs of time-series photometry from DP1 to discover the SX Phoenicis variable LSST-DP1-O-614435753623041404, with period $4\mathrm{k}\times4\mathrm{k}$4 days, amplitudes $4\mathrm{k}\times4\mathrm{k}$5 and $4\mathrm{k}\times4\mathrm{k}$6 mag, and a distance estimate of 16.6 kpc; its location and Gaia proper motion were interpreted as consistent with Monoceros Ring membership (Carlin et al., 30 Jun 2025).

Extragalactic transient work has used the delivered difference-imaging products directly. A search over three DP1 extragalactic fields started from 369,644 DIA objects and, after automated cuts and visual vetting, identified 11 likely supernovae: six Type Ia, two Type II, two Type Ibc, and one Type IIn. That study explicitly found slight tension with expected detection counts and attributed the discrepancy primarily to template contamination, since DP1 templates were built from the same commissioning campaign (Freeburn et al., 30 Jul 2025).

A parallel line of work used DP1 as a testbed for alternative subtraction strategies. The SLIDE pipeline operated on DP1 calexp images inside the Rubin Science Platform, using archival DECam coadds as templates. In ECDFS and EDFS, the main search produced 46 transient candidates, including 34 previously unreported candidates, and 12 of those 34 had no DP1 DiaObject match. This demonstrated that early Rubin-like data could benefit materially from external templates when LSST-native templates are shallow, incomplete, or contaminated (Dong et al., 29 Jul 2025).

DP1 also includes a fully operational Solar System branch. The release contains 431 Solar System objects, of which 93 are new discoveries. These products were assembled through positional association to known objects, tracklet construction, multi-night linking with HelioLinC3D, and orbit filtering. Within DP1, Rubin_SV_38_7 provided the principal low-ecliptic-latitude test field for this functionality (Team et al., 24 Mar 2026).

6. Early-science demonstrations across astrophysics

Although small in area, DP1 has already supported a wide range of statistically or methodologically demanding science. In low-surface-brightness work, deep $4\mathrm{k}\times4\mathrm{k}$7 coadds in ECDFS revealed a stellar stream around the $4\mathrm{k}\times4\mathrm{k}$8 galaxy LEDA 751050. The feature spans roughly $4\mathrm{k}\times4\mathrm{k}$9, has mean surface brightnesses $0.5$0, $0.5$1, and $0.5$2, and remains detectable to $0.5$3. The analysis showed that DP1 background subtraction, though not yet optimized for very extended galaxies, is already adequate for characterizing low-surface-brightness structures on scales $0.5$4 (Johnson et al., 6 Jan 2026).

In circumgalactic-dust work, only 4.6 deg$0.5$5 of DP1 ComCam imaging was sufficient to detect a chromatic reddening profile by stacking background-galaxy colors around foreground galaxies. Interpreting the average $0.5$6 with a Milky Way extinction curve yielded $0.5$7 within 120 kpc. The authors emphasized that the result used only approximately $0.5$8 of the final LSST footprint, making it a direct demonstration of Rubin’s pathfinding capacity for galaxy-dust statistics (Crenshaw et al., 23 Jun 2026).

Galactic stellar-population studies have likewise treated DP1 as a realistic but intentionally imperfect LSST precursor. A cross-match with public variable-star catalogs produced approximately 600 RR Lyrae with adequate light-curve sampling in five DP1 fields. Using LSST-band pulsation models, period-luminosity-metallicity and period-Wesenheit-metallicity relations, that work found that PWZ-based distances agree with the literature with a mean offset of $0.5$9 mag, while also showing that sparse phase coverage in DP1 makes metallicity and distance estimates sensitive to template fitting (Ngeow et al., 1 May 2026). A separate stellar-halo analysis used DP1 $^2$0 photometry in three southern fields to infer distances and metallicities for blue main-sequence stars brighter than $^2$1, and reported a significant deficit of blue turnoff stars with $^2$2 relative to TRILEGAL, interpreting it as evidence for a steeper halo density profile than the canonical $^2$3 model (Palaversa et al., 30 Dec 2025).

Cluster and stellar-classification studies further illustrate DP1’s breadth. In the outer field of 47 Tuc, deep $^2$4 versus $^2$5 color-magnitude diagrams from DP1 identified unresolved main-sequence binaries with mass ratios $^2$6, yielding $^2$7 beyond the half-light radius (Cordoni et al., 4 Sep 2025). In the ultracool-dwarf program USMILE, LSST DP1 $^2$8 photometry cross-matched to VHS and CatWISE supported the first ultracool-dwarf search with DP1, leading to 15 M6–L2 spectroscopically confirmed discoveries and 25 additional M6–L9 photometric candidates after Euclid-based validation (Zhang et al., 17 Oct 2025).

7. Relation to future LSST releases

DP1 is explicitly smaller and less mature than future LSST data releases. It covers about 15 deg$^2$9, uses LSSTComCam rather than LSSTCam, and omits many higher-level products expected later, including the full long-baseline time-domain and astrometric content of LSST Data Release 1 and beyond. Its seeing is commissioning-era rather than survey-specification, some instrumental artifacts are LSSTComCam-specific, and some calibrations remain intentionally provisional (Team et al., 24 Mar 2026).

At the same time, the release has already been used as a sandbox for scalable Rubin-native analysis. HATS and LSDB conversions of DP1 object and diaObject tables, for example, were used to prototype variability-finding pipelines over hundreds of millions of forced-photometry points, yielding both periodic-variable recoveries and novel discoveries. This suggests that DP1 is not merely a small data sample, but also a realistic environment for testing LSST-scale data infrastructure and workflow design (Malanchev et al., 30 Jun 2025).

The most defensible general conclusion is therefore twofold. Factually, DP1 is a commissioning-era, science-grade release that already exposes the essential Rubin image, catalog, time-domain, and access architecture. Interpretively, it functions as a deliberately constrained precursor: small enough to inspect in detail, but rich enough to reveal the practical strengths, failure modes, and scientific scope of the Rubin/LSST system before full operations.