DESI Legacy Imaging Surveys Overview

• DESI Legacy Imaging Surveys are a set of public, multi-program campaigns (DECaLS, BASS, and MzLS) that provide uniform, deep optical imaging over ~14,000 deg².
• They employ dynamic observing strategies and forward-modeling photometry via The Tractor to achieve near-uniform depth (±0.13 mag) and consistent astrometric calibration.
• The surveys underpin DESI spectroscopic target selection, enabling robust galaxy, quasar, and cosmological analyses with comprehensive multi-wavelength catalogs.

The DESI Legacy Imaging Surveys comprise a coordinated set of public, wide-area optical imaging campaigns designed as the foundational imaging dataset to select spectroscopic targets for the Dark Energy Spectroscopic Instrument (DESI). These surveys integrate observing resources across multiple telescopes and imaging systems, employ a dynamic exposure strategy to achieve near-uniform depth, and deliver calibrated multi-band imaging and catalogs over a large extragalactic footprint—currently approximately 14,000 deg² visible from the northern hemisphere. By combining data from bespoke optical imaging programs with forced photometry from the WISE mid-infrared survey, the DESI Legacy Imaging Surveys establish the photometric backbone for DESI and provide an unparalleled resource for astrophysical and cosmological research.

1. Survey Structure and Components

The DESI Legacy Imaging Surveys are the union of three interrelated, public imaging programs, each targeting a different combination of sky area and optical bands:

• DECaLS (Dark Energy Camera Legacy Survey) utilizes DECam on the CTIO Blanco 4-m telescope to observe both the South Galactic Cap (SGC) and a substantial part of the northern sky (declination < +32°). With a field of view of ~3.2 deg² and advanced CCDs, DECaLS delivers deep, high-quality imaging in g, r, and z bands.
• BASS (Beijing-Arizona Sky Survey) is performed with the Bok 2.3-m telescope's 90Prime camera at Kitt Peak. BASS provides g and r band coverage of the North Galactic Cap for declinations above +32°, with a survey footprint designed to complement DECaLS below this declination.
• MzLS (Mayall z-band Legacy Survey) employs the Mosaic-3 camera on the KPNO Mayall 4-m telescope, delivering z-band imaging over the same regions observed by BASS. MzLS images combine high-quality image characteristics (median FWHM ≈ 1″) with a wide northern extragalactic coverage.

The summed imaging footprint spans approximately 14,000 deg², divided into two major contiguous regions by the Galactic plane (roughly 9,900 deg² in the North Galactic Cap and 4,400 deg² in the South Galactic Cap). This partitioning avoids the Milky Way’s high stellar densities and extinction, optimizing the sample of extragalactic objects.

2. Dynamic Observing and Survey Uniformity

A core innovation in the Legacy Surveys is a "dynamic observing" strategy. Each sky region ("brick") is imaged in three passes, filling detector gaps and providing redundancy. Observing conditions are assessed in real time: exposure times are dynamically adjusted using the scaling factor

$$f = T^{-2} \cdot 10^{0.8 k_i (X-1)} \cdot 10^{0.8 A_i E(B-V)} \cdot \frac{N_\mathrm{eff}}{N_{\mathrm{eff}, \mathrm{fid}}}$$

where $T$ is transparency, $k_i$ is the atmospheric extinction coefficient for band $i$, $X$ is airmass, $A_i$ is the Galactic extinction coefficient, $E(B-V)$ the line-of-sight reddening, and $N_\mathrm{eff}$ the effective pixel count.

At least one pass on each area is taken under strictly photometric conditions and with optimal seeing to anchor both photometric and astrometric solutions. This combination of dynamic exposure allocation and multiple passes enables imaging of almost uniform depth (within ±0.13 mag) across the large survey footprint, a critical requirement for homogeneous target selection and photometric measurements.

3. Image Processing, Photometry, and Catalog Construction

The raw Legacy Surveys exposures are processed through the NOAO Community Pipelines, producing bias-corrected and flat-fielded images. These are then assembled into coadded "brick" images using inverse-variance weighting. Weight maps, bad pixel masks, and variance images are propagated to facilitate quality assessment and science analysis.

Source detection, modeling, and measurement are performed by the Tractor forward-modeling algorithm. The Tractor simultaneously fits parametric light profiles (PSF, exponential disk, de Vaucouleurs, and composite models) across all overlapping exposures and bands, convolving each model with the corresponding PSF of each epoch and detector. Model parameters are optimized to yield robust, PSF-matched positions, shapes, and fluxes per source. Importantly, a single morphological model is fit per source across all optical bands to minimize systematic biases in colors and morphology introduced by independent band-by-band photometry.

For mid-infrared photometry, forced photometry is performed on WISE W1–W4 images at (3.4, 4.6, 12, 22 μm) using the optical-model positions and profiles as priors. This enables reliable multi-wavelength photometry—often below the formal WISE detection thresholds—for all optical sources.

4. Photometric Depth and Quality

The Legacy Surveys achieve uniform, deep optical imaging:

• Typical 5σ AB magnitude depths (for an $r_{1/2} = 0.45″$ exponential galaxy): $g\sim24.0$, $r\sim23.4$, and $z\sim22.5$.
• Image quality (FWHM) ranges between $1.1″$ and $1.6″$ depending on band and survey.
• Coadded products, inverse-variance maps, and photometric calibrations ensure high-fidelity, spatially uniform color and morphological measurements.
• The inclusion of WISE data provides robust optical–mid-infrared SED construction, supporting efficient separation of stars from extragalactic QSOs and galaxies.

5. Data Products and Public Access

Each data release (~twice annually) includes:

• Calibrated, sky-subtracted single-exposure images.
• Coadded ("brick") images with associated inverse-variance and weight maps.
• Source catalogs, with Tractor-based photometry, morphological classifications, and cross-matched WISE photometry.
• Ancillary data (bad pixel masks, variance images).
• All data-processing software (legacypipe, tractor) is also made publicly available.
• Data delivery is supported via the NOAO Science Archive and a dedicated Legacy Surveys portal, which provides direct web access, interactive viewers, and documentation.

Successive numbered releases (DR1, DR2, ..., DR7+) have progressively increased sky coverage and incorporated improved calibrations; for example, DR7 delivers nearly 10,000 deg² of DECaLS imaging alongside BASS+MzLS in the north.

6. Scientific Applications and Astrophysical Impact

The DESI Legacy Imaging Surveys underpin the DESI spectroscopic experiment by supplying the photometric input for selection of millions of target galaxies and quasars. The uniformity, depth, and area of the dataset enable:

• Construction of spectroscopic samples to $z>1$ for cosmological clustering and expansion studies.
• Color–color and optical–IR selection of luminous red galaxies, emission-line galaxies, and broad-line QSOs.
• Highly complete and morphologically reliable galaxy catalogs for weak lensing, galaxy evolution, local group stellar populations, and low surface brightness features.
• Photometric redshift estimation and calibration, supported by forced optical to mid-IR photometry.
• Identification of rare phenomena, such as gravitationally lensed systems and ultra-faint dwarf candidates.
• Detailed mapping of dust extinction and stellar populations in the Milky Way away from the Galactic plane, including identification of new stellar streams and satellites inaccessible to Gaia.

Technical advances, particularly in forward-modeling–based photometry and the use of dynamic observations, establish a new standard for wide-area optical surveys in terms of uniformity, calibration accuracy, and cross-band morphological coherence.

7. Legacy and Future Prospects

The DESI Legacy Imaging Surveys provide a foundational public dataset for extragalactic and Galactic astrophysics, as well as for cosmology. The survey model of deep, dynamic, uniform, and multi-wavelength imaging has proven critical for:

• Enabling robust and reproducible spectroscopic targeting for DESI and similar experiments.
• Seeding ongoing and future analyses in lensing, galaxy groups/clusters, and low surface brightness science.
• Providing public software and pipeline resources for downstream custom analyses and reprocessing.
• Informing the design and calibration strategies of upcoming, very-large-area optical surveys, such as LSST, Roman, and Euclid.

The rich multi-band, multi-epoch dataset, in combination with complementary public surveys, will continue to support new scientific investigations far beyond the primary DESI mission, including studies of galaxy assembly and feedback, environmental effects, and rare transient and variable phenomena.