There are three ways of determining which pixels in a CCD image may need to be masked (this is in addition to whatever mask or bit fields the observatory at which you are taking images may provide).
Two of them are the same for all of the science images:
- Hot pixels unlikely to be properly calibrated by subtracting dark current, discussed in Identifying hot pixels.
- Bad pixels identified by
ccdproc.ccdmaskfrom flat field images, discussed in Creating a mask withccdmask.
The third, identifying cosmic rays, discussed in Cosmic ray removal, will by its nature be different for each science image.
The first two masks could be added to science images at the time the science images are calibrated, if desired. They are added to the science images here, as a separate step, because in many situations it is fine to omit masking entirely and there is no particular advantage to introducing it earlier.
We begin, as usual, with a couple of imports.
from pathlib import Path
from astropy import units as u
from astropy.nddata import CCDData
import ccdproc as ccdp
Read masks that are the same for all of the science images
Click here to comment on this section on GitHub (opens in new tab).
In previous notebooks we constructed a mask based on the dark current and a mask created by ccdmask from a flat image. Displaying the summary of the the information about the reduced images is a handy way to determine which files are the masks.
ex2_path = Path('example2-reduced')
ifc = ccdp.ImageFileCollection(ex2_path)
ifc.summary['file', 'imagetyp']
We read each of those in below, converting the mask to boolean after we read it.
mask_ccdmask = CCDData.read(ex2_path / 'mask_from_ccdmask.fits', unit=u.dimensionless_unscaled)
mask_ccdmask.data = mask_ccdmask.data.astype('bool')
mask_hot_pix = CCDData.read(ex2_path / 'mask_from_dark_current.fits', unit=u.dimensionless_unscaled)
mask_hot_pix.data = mask_hot_pix.data.astype('bool')
Combining the masks
Click here to comment on this section on GitHub (opens in new tab).
We combine the masks using a logical "or" since we want to mask out pixels that unit=u.dimensionless_unscaled bad for any reason.
combined_mask = mask_ccdmask.data | mask_hot_pix.data
It turns out we are masking roughly 0.056% of the pixels so far.
combined_mask.sum()
Detect cosmic rays
Click here to comment on this section on GitHub (opens in new tab).
Cosmic ray detection was discussed in detail in an earlier section. Here we loop over all of the calibrated science images and:
- detect cosmic rays in them,
- combine the cosmic ray mask with the mask that applies to all images,
- set the mask of the image to the overall mask, and
- save the image, overwriting the calibrated science image without the mask.
Since the cosmic ray detection takes a while, a status message is displayed before each image is processed.
ifc.files_filtered()
for ccd, file_name in ifc.ccds(imagetyp='light', return_fname=True):
print('Working on file {}'.format(file_name))
new_ccd = ccdp.cosmicray_lacosmic(ccd, readnoise=10, sigclip=8, verbose=True)
overall_mask = new_ccd.mask | combined_mask
# If there was already a mask, keep it.
if ccd.mask is not None:
ccd.mask = ccd.mask | overall_mask
else:
ccd.mask = overall_mask
# Files can be overwritten only with an explicit option
ccd.write(ifc.location / file_name, overwrite=True)