Visit and Difference Image Mask Planes#
The following are the pixel mask bit planes defined in visit and difference images in Data Preview 1 (DP1). Each plane represents a specific per-pixel condition flagged during image processing. Multiple flags may be set on the same pixel simultaneously.
BADPixel marked as bad – e.g. known defective pixel or column, or part of a bad amplifier region. These pixels are identified via detector defect maps or instrument signature removal and flagged as BAD. They are typically interpolated over in processing.
CLIPPEDUsed on deep coadds only; see the deep and template coadds mask planes page. However, this may be back-propagated on visit and difference images in future data releases.
CRPixel hit by a cosmic ray. Identified by the cosmic-ray detection algorithm during single-frame processing; such pixels are flagged
CRand interpolated over. It may also be an indication that neighboring unmasked pixels were affected by aCR, because the algorithm doesn’t always catch the whole region affected by a cosmic ray.CROSSTALKPixel affected by electronic crosstalk from a bright source in another amplifier. Flagged during instrument signature removal (ISR) when crosstalk correction is applied. After subtracting the crosstalk ghost, the affected pixel is labeled with
CROSSTALK.DETECTEDPixel that is part of a detected source in this exposure. All pixels above the detection threshold belonging to a source footprint are flagged as
DETECTED.DETECTED_NEGATIVEPixel that is part of a negative source detection. This is used in image difference contexts (for detecting disappearances or negative flux transients). In DP1 static visit images, this plane is generally not used (no negative detections are run), but it is defined for compatibility with difference imaging.
EDGEIndicates regions near the boundaries of an image that could not be fully processed due to the application of a convolution kernel (typically the PSF-like detection kernel). These pixels are masked because the kernel footprint extends beyond the image boundary, making reliable measurements impossible.
INEXACT_PSFUsed on deep coadds only; see the deep and template coadds mask planes page.
INJECTEDPixel containing an injected synthetic source in the science exposure. This plane is used when artificial sources are added to images for testing or calibration. Any pixel whose value was modified by inserting a simulated source gets the
INJECTEDbit. DP1’s official processing did not inject extra sources into visit images, so this will be unset for most DP1 data.INJECTED_TEMPLATEPixel containing an injected synthetic source in the template image. Used in difference imaging: if a source was artificially added to the template coadd, those template-contributed pixels in the science image’s difference would get this flag.
INTRPInterpolated pixel – this pixel’s value was replaced via interpolation (usually because it was flagged
BAD,SAT, orCR). After interpolation, the pipeline sets theINTRPbit to indicate the value is not original data. For example, saturated cores and cosmic ray hits that have been patched will have both their original flag (SATorCR) andINTRPset.ITL_DIPPixel in a region affected by the “ITL dip” artifact. This is a vendor-specific detector effect seen in ITL CCDs (like those in LSSTComCam) where very bright stars cause a vertical dark trail (a drop in measured flux extending up/down along the column). The pipeline identifies these trails in ISR and masks them. Pixels along such a trail are flagged with
ITL_DIP.NOT_DEBLENDEDPixel in a source footprint that was not deblended. If a detected object was too large, too close to an image edge, or had too high a fraction of masked pixels, the deblender may skip it. In that case the entire footprint is flagged
NOT_DEBLENDED. For example, very bright stars or crowded cores that the deblender could not separate will have this mask.NO_DATAPixel with no valid data in this exposure. In single-visit images this can occur if a pixel falls outside the illuminated area or within a sensor artifact so severe that no data value is present. In general,
NO_DATAindicates that the pixel should be ignored in analysis (not observed).REJECTEDUsed on deep coadds only; see the deep and template coadds mask planes page.
SATSaturated pixel. The pixel’s value exceeded the Photon Transfer Curve (PTC) turnoff point — the threshold at which the detector begins to deviate from linearity and blooming starts. Pixels above this threshold are flagged as
SAT. In DP1 visit image processing, theSATmask is dilated slightly to ensure bleed trails from saturated stars are fully masked, covering adjacent pixels that may be affected by charge blooming. For comparison, theSUSPECTbit is also set above the PTC turnoff but not dilated — it flags pixels likely affected by saturation without meeting the criteria for full saturation.SAT_TEMPLATEPixel that corresponds to a saturated pixel in the template image. This is used in difference imaging: if the static sky template had a saturation at this location, the difference image flags it as
SAT_TEMPLATE(to distinguish from saturation in the new science exposure). This helps avoid false detections or mis-estimation in difference images. Not used in standalone visit images; relevant in DP1 difference image products.SENSOR_EDGEUsed on deep coadds only; see the deep and template coadds mask planes page.
STREAKPixel in a linear streak region — typically from satellites, aircraft, or occasionally diffraction spikes. The
STREAKmask is applied during difference imaging. Pixels are flaggedSTREAKwhen linear features are detected insideDETECTEDregions of the difference image, usually via Hough transform. Once a streak is identified, the masked region is extended across the full detector column or row to cover the artifact completely.SUSPECTPixel that is suspicious — likely affected by blooming, non-linearity, or readout effects — but not fully saturated. The
SUSPECTbit is set for pixels above the PTC turnoff (i.e., where the detector begins to deviate from linear response), just likeSAT. However, unlikeSAT, theSUSPECTmask is not dilated. It flags only those pixels directly above the threshold, without extending to surrounding regions. This meansSUSPECTpixels are typically on the shoulders or flanks of saturated regions, where flux is high but blooming is not yet strong enough to trigger theSATdilation.UNMASKEDNANPixel value is a NaN (Not-a-Number) that was not originally masked. This flags any pixels that turned into NaNs during processing. If a pixel ends up with an undefined value (NaN) and no other mask bit set, the pipelines will set the
UNMASKEDNANplane for that pixel. This alerts the user that the pixel has invalid data. Such cases are rare and typically indicate a processing error or division-by-zero in calibration.VIGNETTEDPixel in a vignetted region of the sensor. This means the pixel is significantly darkened by the optical vignetting (for example, at the very edge of the field of view where the camera’s optics or filter holder obscures light). Such pixels receive the
VIGNETTEDflag. Effectively, these areas have much lower exposure and are often excluded from analysis. By default, extremely vignetted sources are not deblended.