Sky Subtraction

The pipeline performs automatic sky subtraction by identifying areas of the sky that do not contain object flux. If this fails, it is possible to modify the sky subtraction algorithm in two ways:

by using a different frame as the sky frame and
by specifying areas of the target or sky frame to exclude

The latter method is accomplished with a mask file (see Building a mask file) and will exclude the user-selected object-affected pixels from the calculation of the sky.

To specify the preferred method for sky subtraction, create a file in the data directory called kcwi.sky. For each KCWI frame for which you want to modify the sky subtraction algorithm, enter a row with this format:

raw_object_file.fits raw_sky_frame.fits <mask.fits>

For example:

kb230605_00098.fits kb230605_00100.fits

If you want to specify an external sky frame, only use the first two columns and do not specify a mask file.

If you want to use a mask on the original file, make sure that the first two columns contain the same file name and add the mask file.

For example:

kb230605_00098.fits kb230605_00098.fits kb230605_00098_smsk.fits

Building a mask file

To build a mask file based on ds9 regions, use the script kcwi_masksky_ds9.py contained in the scripts directory.

This script uses the _intf.fits files which are generated as part of the first execution of the pipeline. You should run the pipeline first and verify the quality of the sky subtraction. If you are not satisfied with the sky subtraction, use the procedure described here and run the pipeline again.

To start, display the _intf.fits frame in ds9 and create regions around areas of the frame that you would like to exclude. Save the regions to a file and run the kcwi_masksky_ds9 command indicating the file name (_intf.fits) and the region file. This will produce a mask file with a filename that is the same as the input _intf.fits file, but replacing the intf with smsk. That filename can be added to the kcwi.sky file as in the example above.

Automatic masking

To perform automatic masking, use a different form for the entry in kcwi.sky. For a bright continuum source that facilitates automatic object flux rejection add the entry:

raw_object_file.fits cont

where the word cont indicates a continuum source. For example:

kr230605_00120.fits cont

For a fainter continuum source where automatic masking is not possible, add the entry:

raw_object_file.fits cont object_position width_of_object

For example:

kr230605_00120.fits cont 14.3 3.5

In the first case, the DRP will find the location and width of the continuum source automatically. This case should be used for bright continuum sources.

In the second case, the third and fourth columns specify the location and width of a fainter continuum source. In both cases, the pixel positions across the slice where the continuum source resides are all masked. The sky is determined from a small region just outside the continuum source mask.

To determine the pixel position of a faint continuum source, you can load the _intf.fits frame in ds9 and then load the corresponding position map image _posmap.fits into an adjacent frame. If you blink between the two frames you can determine the object_position and width_of_object values to enter in the kcwi.sky file.

All standard star observations are automatically masked as bright continuum sources.

Re-run the reduction of this file

Rather than re-run the entire pipeline, it is possible to run it only on the file that needs an improved sky subtraction.

The instructions are listed in Running the pipeline. In summary, you will need to specify just one file, the one for which you want to improve the sky subtraction, and make sure that clobber is set to True in the config file and be sure to remove the entry for this file in the proc file (kcwib.proc or kcwir.proc). For example:

reduce_kcwi -b -f kb230605_00098.fits

or

reduce_kcwi -r -f kr230605_00120.fits

depending on which channel was used for the observation.

Disabling sky subtraction

The DRP makes a general attempt at modeling the sky, however, no sky subtraction algorithm will be perfect for all objects. To acknowledge that fact and allow the user to attempt their own sky subtraction, the DRP creates a new FITS extension called NOSKYSUB that contains the data frame without sky subtraction.

In addition, there are two methods to allow the user to skip sky subtraction.

If you enter the word skip on the line after the raw file in the kcwi.sky file, sky subtraction will be skipped for that file. For example: kr230605_00120.fits skip.
If you use the command line parameter -k, sky subtraction will be skipped for all images processed from that command line.