This file describes the data model for the Bias image. This is written to disk as a multi-extension FITS file prefixed by Bias in the Calibrations/ folder.

This is generally a simple combination of all input bias frames.


The first extension is the stacked image. You can view it with any standard image viewer, e.g.:

ginga Calibrations/Bias_A_1_01.fits

It should resemble any one of your input bias frames aside from:

  • Having been overscan subtracted

  • Being only a trimmed portion of the image

  • Re-oriented so that vertical is the spectral dimension with blue at the bottom

Here is an example for the LRIS RED spectrograph.


Pretty boring, as expected. Note that this is only 1 of the 2 detectors for this spectrograph.

Trouble Shooting

If your image appears to be in err, here are the things to consider:

  • Is one or more of your input bias frames junk?

  • Is one or more of your input bias frames mis-labeled?

Current BiasImage Data Model

Internally, the image is held in BiasImage, which subclasses from PypeItImage and DataContainer. The datamodel written to disk is:

Version 1.3.0

HDU Name

HDU Type

Data Type



Empty data HDU. Contains basic header information.



Primary image data



Inverse variance image


The detector (see DetectorContainer) or mosaic (see Mosaic) parameters