Difference image analysis (DIA)

DIA uses the decorrelated Alard and Lupton (1998) image differencing algorithm, as described in “Implementation of Image Difference Decorrelation” (dmtn-021.lsst.io) and in the LSST Science Pipelines documentation for the Alard-Lupton implementation.

Key terminology:

• DiaSource: a detection, positive or negative, in a single difference image

• DiaObject: the astrophysical transient or variable object at a static sky coordinate

Light curves: the forced photometry fluxes should be used for light curves.

All measurements are stored in the catalog Data products.

Difference images

Template coadd images are warped, scaled, and PSF-matched to a visit image, and then subtracted from the visit image to create a difference image.

In the context of DIA, the visit image is sometimes referred to as the “direct” or “science” image.

For future data releases, to save space, the difference images will not be stored. Instead, a tool for on-the-fly difference image recreation will be provided to users.

For DP1, the difference images are served and can be retrieved and displayed by users instantaneously, but the inputs and tools for their re-creation are not supplied.

DIA source detection

DIA source detection refers to the process of finding regions with above-threshold absolute flux in difference images, where the threshold is a signal-to-noise ratio of 5.

Difference images contain sources of residual flux from the subtraction, both positive and negative. Real astrophyisical transients, variables, and moving objects appear as sources with a flux that is equal to the difference between their flux in the template and visit images.

Spurious sources are also present due to unavoidably imperfect template matching.

Deblending is not run on difference image sources.

The results of DIA source detection are stored in the DiaSource catalog.

Reliability score

Both real and spurious DIA sources can be detected with a signal-to-noise ratio of 5. An estimate of how likely a source is to be real is provided to enable filtering of DIA sources by users.

DP1 uses a relatively simple machine learning reliability model, trained on simulated data and injected point sources labelled by Rubin staff, and applied to DIA sources. The reliability value is stored in the reliability column of the DiaSource catalog.

Bright stars and correlated noise are the main sources of spurious detections.

Dipoles

Sources with footprints containing both positive and negative peaks were fit with a dipole centroid code. Dipoles represent sources that have moved only slightly (e.g., less than the PSF) between template and visit image.

Pixel flags

If the footprint contains one or more flagged pixels, e.g., for cosmic rays, detector edge, bad pixels, known artifacts, non-astrophysical trails, the source is also flagged.

DIA object association

DIA sources are associated into DIA objects using a radius of 1 arcsecond.

The DIA objects are stored in the DiaObject catalog.

Variability characterization

Light curve characterization parameters are calculated for all DIA objects, such as the minimum, mean, maximum, standard deviation, and skew in the difference image fluxes, the light curve’s slope, percentiles, and StetsonJ parameter.

DIA forced photometry

In general, “forced” photometry means a measurement made at a fixed coordinate in an image, regardless of whether an above-threshold region was detected in that particular image.

Forced PSF photometry measurements are made on all visit and difference images at the locations of all DIA objects. The results are stored in the ForcedSourceOnDiaObject catalog.