Automated sorting of Keck/MOSFIRE frames by instrument configuration

Version History






Debora Pelliccia

12 Jul 2021


To prepare for the data reduction, PypeIt automatically associates fits files to specific Frame Types (see Automated typing of Keck/MOSFIRE frames) and collects groups of frames in unique instrument configurations (see below). This is performed by the pypeit_setup script, which sorts the frames and write a PypeIt Reduction File for each unique configuration. See Setup.

MOSFIRE configuration identification

The MOSFIRE instrument configurations are determined by pypeit.metadata.PypeItMetaData.unique_configurations(), which finds unique combinations of the following keywords:

fitstbl key

Header Key




No Key




No Key


No Key

decker_secondary, slitlength, and slitwid are defined as follow:

  • decker_secondary is by default equal to the header key MASKNAME. However, when “LONGSLIT” is included in MASKNAME, decker_secondary is equal to “LONGSLIT”.

  • slitlength is the length of the slit, expressed in number of CSUs (Configurable Slit Units). This value is included in the header key MASKNAME for LONGSLIT mask. Therefore, slitlength is available only for LONGSLIT mask, and is None for multi-object and long2pos masks.

  • slitwid is the width of the slit, expressed in arcsec. This value is included in the header key MASKNAME for LONGSLIT mask. Therefore, slitwid is available only for LONGSLIT mask, and is None for multi-object and long2pos masks.

The unique configurations are determined by collating the relevant metadata from the headers of all frames found by a run of pypeit_setup. Each unique configuration is given a capital letter identifier (e.g., A,B,C,D…).

After that, pypeit.metadata.PypeItMetaData.set_configurations() associates each frame to the relevant unique configuration (“setup”), by assigning a setup identifier (e.g., A,B,C,D…) to every frames for which the values of the above keywords match the values of the specific unique configuration.

This process is slightly modified for LONGSLIT and long2pos masks. It is common observation practise with MOSFIRE to use LONGSLIT masks that are 46 CSUs long for the calibration of science frames obtained with shorter LONGSLIT masks of the same width. For this reason, when LONGSLIT masks are used, we don’t require that the value of slitlength for the calibration frames (arcs, flats) matches the value of the specific unique configuration. This allows to associate the same calibration frames (only the ones with LONGSLIT masks that are 46 CSUs long) to different unique configurations, i.e., where the science/standard frames are obtained with shorter LONGSLIT.

Similarly, when long2pos masks are reduced, calibration frames taken with masks that have MASKNAME equal to “long2pos” are used to calibrated science/standard frames that have MASKNAME equal to “long2pos_specphot”. In this case, the default behaviour would identify the calibration frames and the science/standard frames as part of different unique configurations. Instead, when long2pos masks are used, we modify in place the requirement for the calibration frames. Specifically, only when we are assigning the setup identifier to the calibration frames, we temporarily change the required value of decker_secondary in the specific unique configuration from “long2pos_specphot” to “long2pos”. This allows to identify “long2pos” calibration frames and “long2pos_specphot” science/standard frames as part of the same unique configuration.

MOSFIRE calibration groups

PypeIt uses the concept of a “calibration group” to define a complete set of calibration frames (e.g., arcs, flats) and the science frames to which these calibration frames should be applied.

By default, pypeit_setup uses the setup identifier (e.g., A,B,C,D…) to assign frames to a single calibration group. Frames that are in the same calibration group will have the same PypeIt keyword calib. No automated procedure exists to do anything except this. However, the user can edit the PypeIt Reduction File to, within a given configuration, assign specific calibration frames to specific science frames using the data in the calib column of the Data Block.

MOSFIRE combination and background groups

PypeIt is able to reduce data taken with a nodding pattern, by grouping the science frames into combination and background groups. Science frames that should be combined together are assigned the same combination ID (comb_id), while a background ID (bkg_id) identifies frames that are used as background images. Frames with the same value of bkg_id will be combined together. The values of comb_id and bkg_id are provided in the PypeIt Reduction File as two columns in the Data Block, so that users can modify them according to their preferred reduction. See more detail in A-B image differencing.

For MOSFIRE, PypeIt attempts to automatically assign comb_id and bkg_id to the science frames, by using the information on the nodding pattern available in the files headers. Specifically, the keywords used are:

fitstbl key

Header Key







which are also provided in the Data Block.

If the observations were taken with a “Slit Nod”/”Mask Nod” dithpat or using the long2pos slitmask, the Data Block will look like:

              filename | frametype | ... |  dithpat | dithpos | dithoff | frameno | calib | comb_id | bkg_id
MF.20141126.17372.fits |   science | ... | Mask Nod |       A |    10.0 |     182 |     4 |      21 |     22
MF.20141126.17526.fits |   science | ... | Mask Nod |       B |   -10.0 |     183 |     4 |      22 |     21
MF.20141126.17686.fits |   science | ... | Mask Nod |       A |    10.0 |     184 |     4 |      23 |     24
MF.20141126.17842.fits |   science | ... | Mask Nod |       B |   -10.0 |     185 |     4 |      24 |     23

where all the science frames have different comb_id (i.e., no frames will be combined), while the bkg_id for the frame at the “A” dithpos is equal to the comb_id of the frame at the “B” dithpos and vice versa. This combination of comb_id and bkg_id will create four reduced frames:

MF.20141126.17372.fits - MF.20141126.17526.fits (A-B)
MF.20141126.17526.fits - MF.20141126.17372.fits (B-A)
MF.20141126.17686.fits - MF.20141126.17842.fits (A-B)
MF.20141126.17842.fits - MF.20141126.17686.fits (B-A)

If the observations were taken with an “ABAB” or “ABBA” dithpat, the frames in the same dither sequence will be combined. Here is an example for “ABBA”:

              filename | frametype | ... |  dithpat | dithpos | dithoff | frameno | calib | comb_id | bkg_id
MF.20130903.35554.fits |  standard | ... |     ABBA |       A |     6.0 |     276 |     0 |       1 |      2
MF.20130903.35576.fits |  standard | ... |     ABBA |       B |    -6.0 |     277 |     0 |       2 |      1
MF.20130903.35593.fits |  standard | ... |     ABBA |       B |    -6.0 |     278 |     0 |       2 |      1
MF.20130903.35620.fits |  standard | ... |     ABBA |       A |     6.0 |     279 |     0 |       1 |      2
MF.20130903.35679.fits |  standard | ... |     ABBA |       A |     6.0 |     280 |     0 |       5 |      6
MF.20130903.35697.fits |  standard | ... |     ABBA |       B |    -6.0 |     281 |     0 |       6 |      5
MF.20130903.35710.fits |  standard | ... |     ABBA |       B |    -6.0 |     282 |     0 |       6 |      5
MF.20130903.35726.fits |  standard | ... |     ABBA |       A |     6.0 |     283 |     0 |       5 |      6

This combination of comb_id and bkg_id will create four reduced frames:

MF.20130903.35554.fits+MF.20130903.35620.fits - MF.20130903.35576.fits+MF.20130903.35593.fits (AA-BB)
MF.20130903.35576.fits+MF.20130903.35593.fits - MF.20130903.35554.fits+MF.20130903.35620.fits (BB-AA)
MF.20130903.35679.fits+MF.20130903.35726.fits - MF.20130903.35697.fits+MF.20130903.35710.fits (AA-BB)
MF.20130903.35697.fits+MF.20130903.35710.fits - MF.20130903.35679.fits+MF.20130903.35726.fits (BB-AA)

Lastly, if observations were taken with the long2pos_specphot slitmask, where two frames are taken at the “A” and “B” dithpos and one frame is taken at the center (dithoff = 0) of a wider slit (see ), one of the two “A” and “B” frames are used as background image for the frame taken at the center of the wider slit. Here is an example:

              filename | frametype | ... |  dithpat | dithpos | dithoff | frameno | calib | comb_id | bkg_id
MF.20181217.55882.fits |  standard | ... | long2pos |       A |     0.0 |     286 |     7 |      27 |     28
MF.20181217.55901.fits |  standard | ... | long2pos |       B |    -7.0 |     287 |     7 |      28 |     29
MF.20181217.55920.fits |  standard | ... | long2pos |       A |     7.0 |     288 |     7 |      29 |     28

This combination of comb_id and bkg_id will create three reduced frames:

MF.20181217.55882.fits - MF.20181217.55901.fits (center pos -B)
MF.20181217.55901.fits - MF.20181217.55920.fits (A-B)
MF.20181217.55920.fits - MF.20181217.55901.fits (B-A)


  • Requirement PM-5 states: “As a user, I want the pipeline to automatically correctly associate calibrations with science data.”

  • Requirement PM-34 states: “As a user, I expect the calibration association system to use calibrations from full long slits (46 bars long) for short “long slits” of the same width.”

  • Requirement PM-7 states: “Use longslit calibrations for longslit calibrations even if the length of the slit is different.”

  • Requirement PM-8 states: “Use narrow slit long2pos calibration for long2pos reduction.”

  • Requirement PM-13 states: “As a user, I expect my dithering pattern to automatically turn into the proper reduction sequence. If I used ABBA, I expect this to be taken into account”

PypeIt meets these requirements in the majority of use cases.

The test used to demonstrate that PM-4 is satisfied (Automated typing of Keck/MOSFIRE frames) is also relevant here since it shows that PypeIt correctly identifies MOSFIRE data frame types and associates them with a single configuration, all written to a single pypeit file.

To test that PypeIt can successfully identify multiple configurations among a set of files, and can assign comb_id and bkg_id to science frames following the information on the dither pattern, we have added the test_setup_keck_mosfire_multiconfig test to ${PYPEIT_DEV}/unit_tests/

To run this test:

cd ${PYPEIT_DEV}/unit_tests
pytest -W ignore

The test requires that you have downloaded the PypeIt PypeIt Development Suite and defined the PYPEIT_DEV environmental variable that points to the relevant directory.

The algorithm for this test is as follows:

  1. Collect the names of all files in the following two directories:

  2. Use PypeItSetup to automatically identify the configurations for these files.

  3. Check that the code found four configurations and wrote the pypeit files for each.

  4. For each configuration:

    1. Read the pypeit file

    2. Check that the name for the setup is correct (‘A’, ‘B’, ‘C’, or ‘D’)

    3. Check that the calibration group is the same for all frames (‘0’, ‘1’, ‘2’, ‘3’)

    4. Check that comb_id and bkg_id for the science frames are what expected. The dither sequences used here are: “ABBA”, “long2pos”, “ABA’B’”, “Mask Nod”.

Because this test is now included in the PypeIt Unit Tests (GitHub CI), these configuration checks are performed by the developers for every new version of the code.