"Snowball" masking with snowblind

[gbrammer]

Implement DQ masking with the snowblind module, which has better mask shapes that vary in size depending on the snowball intensity.

As of 5 Dec 2023, the functionality to run the masking on rate products has not yet been merged into the main snowblind repository, so it has to be installed from the fork:

pip install git+https://github.com/gbrammer/snowblind.git

[gbrammer]

With this PR, this algorithm is preferred in the initial data preprocessing in grizli.prep.mask_snowballs.

To refine the masks in existing exposures when making mosaics call grizli.aws.visit_processor.cutout_mosaic like

from grizli.aws import visit_processor

snowblind_kws = dict(new_jump_flag=1024, min_radius=4, growth_factor=1.5, unset_first=True)

mos = visit_processor.cutout_mosaic('gds-snow', ra=53.0917130, dec=-27.7425336,
                              size=120,
                              filters=['F210M-CLEAR'],
                              clean_flt=False, s3output=None,
                              ir_scale=0.04,
                              half_optical=False,
                              weight_type='jwst',
                              kernel='square', pixfrac=0.8,
                              make_exptime_map=False,
                              skip_existing=False,
                              snowblind_kwargs=snowblind_kws, # set to something other than `None`
                             )

[gbrammer]

h/t @jdavies-st

[pascaloesch]

Note that snowblind requires scikit-image>=0.20.0, where 'skimage.morphology.isotropic_dilation' was introduced.

[gbrammer]

Note that snowblind requires scikit-image>=0.20.0, where 'skimage.morphology.isotropic_dilation' was introduced.

Added in 93deb61. Thanks!

[jdavies-st]

Note that snowblind requires scikit-image>=0.20.0, where 'skimage.morphology.isotropic_dilation' was introduced.

Thanks for noting this @pascaloesch. I'll update the requirements for snowblind and release.

[jdavies-st]

snowblind 0.2.1 is now released with the updated functionality for working with _rate and _rateints files, and also with the updated scikit-image version requirement. Thanks! 👍

[gbrammer]

snowblind 0.2.1 is now released with the updated functionality for working with _rate and _rateints files, and also with the updated scikit-image version requirement. Thanks! 👍

Awesome, thanks!