Calibration Butler dataset types#
Important
Calibration images are not part of Early Data Preview 2 (EDP2); they will be delivered with the full DP2 release. For more information, see “Rubin Plans for an Early Science Program” (RTN-011).
The calibration products are accessed via the Butler as a number of dataset types. The combined bias, dark, and flat frames described on the main calibrations page are the most commonly used, but the full set of calibration dataset types and their descriptions is given below.
These calibrations are produced and consumed by the Instrument Signature Removal (ISR) pipeline. For a detailed description of the calibration products and the ISR process, see Plazas Malagón et al. 2025 and RTN-117.
For each dataset type below, the required dimensions of the Butler dataId are shown in boldface.
Raw images#
raw(‘raw’, {band, instrument, day_obs, detector, group, physical_filter, exposure}, Exposure)
The unprocessed images from the camera, which are the input to the ISR pipeline; see Raw exposure for a full description.
Detector characterization#
defects(‘defects’, {instrument, detector}, Defects, isCalibration=True)
A mask identifying bad pixels, columns, and other detector defects (hot, dead, or otherwise unreliable pixels) that should be flagged or interpolated over; includes defects identified in bias, dark, and flat images, in addition to a set of manually marked pixels (e.g. vampires).
crosstalk(‘crosstalk’, {instrument, detector}, CrosstalkCalib, isCalibration=True)
The crosstalk matrix containing coefficients of the coupling between all combination pairs of amplifiers in a detector, used to correct copycat images induced by bright sources on neighboring channels.
linearizer(‘linearizer’, {instrument, detector}, Linearizer, isCalibration=True)
A correction that maps measured signal to a linear response, removing the detector’s nonlinear behavior as a function of flux level.
cti(‘cti’, {instrument, detector}, IsrCalib, isCalibration=True)
The deferred charge calibration. Though it is called CTI, it contains a set of model parameters for classical global charge transfer inefficiency (CTI) as well as serial traps and local electron offset drifts, characterizing and removing trailing/smearing of charge that occurs during detector readout.
ptc(‘ptc’, {instrument, detector}, PhotonTransferCurveDataset, isCalibration=True)
The photon transfer curve, relating signal variance to mean signal to derive per-amplifier gain and read noise for variance plane estimation, and also contains a set of model fit parameters used to characterize detector response (incl. brighter-fatter).
Combined calibration frames#
bias(‘bias’, {instrument, detector}, ExposureF, isCalibration=True)
The final combined bias frame removing static electronic structure present in zero-exposure reads; built by combining many bias exposures with defect masking and preliminary corrections applied.
dark(‘dark’, {instrument, detector}, ExposureF, isCalibration=True)
The final combined dark frame characterizing and removing dark current accumulated as a function of exposure time (measured from 30 s dark exposures).
flat(‘flat’, {band, instrument, detector, physical_filter}, ExposureF, isCalibration=True)
The final normalized, combined flat fields per filter correcting QE variations in the detector’s response to uniform illumination, and taking into account the (approximately flat) in-band SED of the illumination source from the in-dome calibration system.
Brighter-fatter correction#
electroBfDistortionMatrix(‘electroBfDistortionMatrix’, {instrument, detector}, IsrCalib, isCalibration=True)
The brighter-fatter distortion matrix (from Astier & Regnault 2023) describing the pixel area changes in response to a single charge. It contains a set of model parameters to solve for the electric field in a pixel due to a single charge, as well as a set of pixel distortion matrices that describe how pixel area changes in response to a charge in a pixel due to shifts in the north, south, east, and west boundaries.
We are no longer using the Coulton correction (Coulton et al. 2018), so there are no brighter-fatter kernels; instead there is a single correction that combines both Astier & Regnault 2023 and Broughton et al. 2024. This brighter-fatter correction has been meeting the Y10 science requirements.