Deep coadd#

The combination of multiple processed, calibrated, and background-subtracted images, for a patch of sky, for each of the six LSST filters.

Citation: NSF-DOE Vera C. Rubin Observatory (2025); Legacy Survey of Space and Time Data Preview 1: deep_coadd dataset type DOI Logo https://doi.org/10.71929/rubin/2570313 [BibTeX]

Access#

The deep coadd images are accessible via the Butler, SIA, and TAP services.

Butler#

  • Dataset type: (‘deep_coadd’, {band, skymap, tract, patch}, ExposureF)

  • Format: FITS

  • Number of Butler datasets: 2,644

SIA and TAP#

Schema: ObsCore table

IVOA calibration level: 3

Dataproduct subtype: lsst.deep_coadd

Description#

A DP1 visit image must have a PSF FWHM < 1.7 arcseconds to be selected as an input image to the coadd. A mean stacking algorithm, weighted by inverse variance, combines selected exposures.

Each individual deep coadd image covers a single patch of the sky: a quadrilateral sub-region of the overall skymap that covers approximately 79 square arcminutes. Patches slightly overlap at their edges. Deep coadd images are for a single filter.

Processing#

The deep coadd images are the result of Image coaddition.

Pixel data#

The deep coadd images have three planes of pixel data.

Image: sky pixel data in flux units of nJy.

Variance: uncertainty (noise) in the flux in units of nJy^2.

Mask: an integer bitmask of representative flag values that indicate processing status or issues, similar to the SDSS bitmasks.

Metadata#

The metadata for deep coadd images retrieved from the Butler include a list of the input visit images, and the derived PSF, photometric calibration, and WCS.

WCS#

The World Coordinate System objects for coadd images are exactly representable as FITS (they are simple TAN projections), but each tract has its own WCS that is shared by all patches, with an integer offset that must be applied for each patch manually to index the pixel arrays.

See WCS FITS approximations and misleading interfaces for more information.

Tutorials#

See the 200-level notebook or 200-level portal tutorials demonstrating how to access the deep coadd images.