At its basic level this step flat-fields an input science dataset by dividing by a flat-field reference image. In particular, the SCI array from the flat-field reference file is divided into the SCI array of the science dataset, the flat-field DQ array is combined with the science DQ array using a bitwise OR operation, and variance and error arrays in the science dataset are updated to include the flat-field uncertainty. Details for particular modes are given in the sections below.

Upon completion of the step, the step status keyword “S_FLAT” gets set to “COMPLETE” in the output science data.

Imaging and Non-NIRSpec Spectroscopic Data

Simple imaging data, usually in the form of an ImageModel, and some spectroscopic modes, use a straight-forward approach that involves applying a single flat-field reference file to the science image. The spectroscopic modes included in this category are NIRCam WFSS and Time-Series Grism, NIRISS WFSS and SOSS, and MIRI MRS and LRS. All of these modes are processed as follows:

  1. If the science data have been taken using a subarray and the FLAT reference file is a full-frame image, extract the corresponding subarray region from the flat-field data.

  2. Find pixels that have a value of NaN or zero in the FLAT reference file SCI array and set their DQ values to “NO_FLAT_FIELD” and “DO_NOT_USE.”

  3. Reset the values of pixels in the flat that have DQ=”NO_FLAT_FIELD” to 1.0, so that they have no effect when applied to the science data.

  4. Propagate the FLAT reference file DQ values into the science exposure DQ array using a bitwise OR operation.

  5. Apply the flat according to:

    \[SCI_{science} = SCI_{science} / SCI_{flat}\]
    \[VAR\_POISSON_{science} = VAR\_POISSON_{science} / SCI_{flat}^2\]
    \[VAR\_RNOISE_{science} = VAR\_RNOISE_{science} / SCI_{flat}^2\]
    \[VAR\_FLAT_{science} = ( SCI_{science}^{2} / SCI_{flat}^{2} ) * ERR_{flat}^{2}\]
    \[ERR_{science} = \sqrt{VAR\_POISSON + VAR\_RNOISE + VAR\_FLAT}\]

Multi-integration datasets (“_rateints.fits” products), which are common for modes like NIRCam Time-Series Grism, NIRISS SOSS, and MIRI LRS Slitless, are handled by applying the above equations to each integration.

For guider exposures, the flat is applied in the same manner as given in the equations above, except for several differences. First, the variances due to Poisson noise and read noise are not calculated. Second, the output ERR array is the combined input ERR plus the flatfield ERR, summed in quadrature.

NIRSpec Spectroscopic Data

Flat-fielding of NIRSpec spectrographic data differs from other modes in that the flat-field array that will be applied to the science data is not read directly from CRDS. This is because the flat-field varies with wavelength and the wavelength of light that falls on any given pixel depends on the mode and which slits are open. The flat-field array that is divided into the SCI and ERR arrays is constructed on-the-fly by extracting the relevant section from the reference files, and then – for each pixel – interpolating to the appropriate wavelength for that pixel. This interpolation requires knowledge of the dispersion direction, which is read from keyword “DISPAXIS.” See the Reference File section for further details.

For NIRSpec Fixed-Slit and MOS exposures, an on-the-fly flat-field is constructed to match each of the slits/slitlets contained in the science exposure. For NIRSpec IFU exposures, a single full-frame flat-field is constructed, which is applied to the entire science image.

NIRSpec NRS_BRIGHTOBJ data are processed just like NIRSpec Fixed-Slit data, except that NRS_BRIGHTOBJ data are stored in a CubeModel, rather than a MultiSlitModel. A 2-D flat-field image is constructed on-the-fly as usual, but this image is then divided into each plane of the 3-D science data arrays.

In all cases, there is a step option that allows for saving the on-the-fly flatfield to a file, if desired.

NIRSpec Fixed-Slit Primary Slit

The primary slit in a NIRSpec fixed-slit exposure receives special handling. If the primary slit, as given by the “FXD_SLIT” keyword value, contains a point source, as given by the “SRCTYPE” keyword, it is necessary to know the flatfield conversion factors for both a point source and a uniform source for use later in the master background step in Stage 3 processing. The point source version of the flatfield correction is applied to the slit data, but that correction is not appropriate for the background signal contained in the slit, and hence corrections must be applied later in the master background step.

So in this case the flatfield step will compute 2D arrays of conversion factors that are appropriate for a uniform source and for a point source, and store those correction factors in the “FLATFIELD_UN” and “FLATFIELD_PS” extensions, respectively, of the output data product. The point source correction array is also applied to the slit data.

Note that this special handling is only needed when the slit contains a point source, because in that case corrections to the wavelength grid are applied by the wavecorr step to account for any source mis-centering in the slit and the flatfield conversion factors are wavelength-dependent. A uniform source does not require wavelength corrections and hence the flatfield conversions will differ for point and uniform sources. Any secondary slits that may be included in a fixed-slit exposure do not have source centering information available, so the wavecorr step is not applied, and hence there’s no difference between the point source and uniform source flatfield conversions for those slits.