Description
- Class:
- Alias:
tweakreg
Overview
This step creates image catalogs of point-like sources whose centroids are then used to compute corrections to the WCS of the input images such that sky catalogs obtained from the image catalogs using the corrected WCS will align on the sky.
Source Detection
If the meta.tweakreg_catalog
attribute of input data models is a non-empty
string and use_custom_catalogs
is True
, then it will be interpreted
as a file name of a user-provided source catalog. The catalog must be in a
format automatically recognized by read()
.
When the meta.tweakreg_catalog
attribute of input data models is None
or
an empty string, then the tweakreg
step will attempt to detect sources in the
input images. Stars are detected in the image with one of the following source
detection algorithms: photutils.detection.DAOStarFinder
(default),
photutils.detection.IRAFStarFinder
, or photutils.segmentation.SourceFinder
in conjunction with photutils.segmentation.SourceCatalog
.
DAOStarFinder is an implementation of the DAOFIND algorithm (Stetson 1987, PASP 99, 191). It searches images for local density maxima that have a peak amplitude greater than a specified threshold (the threshold is applied to a convolved image) and have a size and shape similar to a defined 2D Gaussian kernel. DAOFind also provides an estimate of the object’s roundness and sharpness, whose lower and upper bounds can be specified.
IRAFStarFinder is a Python implementation of the IRAF star finding algorithm, which also calculates the objects’ centroids, roundness, and sharpness. However, IRAFStarFinder uses image moments instead of 1-D Gaussian fits to projected light distributions like DAOStarFinder.
SourceFinder implements a segmentation algorithm that identifies sources in an image based on a number of connected pixels above a specified threshold value. The sources are deblended using a combination of multi-thresholding and watershed segmentation. SourceCatalog finds the centroids of these sources, which are used as the retrieved star positions.
Warning
It has been shown (STScI Technical Report JWST-STScI-008116, SM-12)
that for undersampled PSFs, e.g. for short-wavelength NIRISS
imaging data, DAOStarFinder
gives bad results no matter the input parameters
due to its use of 1-D Gaussian fits.
IRAFStarFinder
or SourceFinder
should be used instead.
Note
SourceFinder
is likely to detect non-stellar sources
such as galaxies because sources are not assumed to be
point-source-like. This may lead to mismatches between the
derived source catalog and the reference catalog during the
alignment step.
Custom Source Catalogs
Source detection built into the tweakreg
step can be disabled by
providing a file name to a custom source catalog in the
meta.tweakreg_catalog
attribute of input data models.
The catalog must be in a format automatically recognized by
read()
. The catalog must contain
either 'x'
and 'y'
or 'xcentroid'
and 'ycentroid'
columns which
indicate source image coordinates (in pixels). Pixel coordinates are
0-indexed. An optional column in the catalog is the 'weight'
column,
which when present, will be used in fitting.
For the tweakreg
step to use user-provided input source catalogs,
use_custom_catalogs
parameter of the tweakreg
step must be set to
True
.
In addition to setting the meta.tweakreg_catalog
attribute of input data
models to the custom catalog file name, the tweakreg_step
also supports two
other ways of supplying custom source catalogs to the step:
Adding
tweakreg_catalog
attribute to themembers
of the input ASN table - seeModelContainer
for more details. Catalog file names are relative to ASN file path.Providing a simple two-column text file, specified via step’s parameter
catfile
, that contains input data models’ file names in the first column and the file names of the corresponding catalogs in the second column. Catalog file names are relative tocatfile
file path.
Specifying custom source catalogs via either the input ASN file or
catfile
will update input data models’ meta.tweakreg_catalog
attributes to the catalog file names provided in either in the ASN file or
catfile
.
Note
When custom source catalogs are provided via both catfile
and
ASN file members’ attributes, the catfile
takes precedence and
catalogs specified via ASN file are ignored altogether.
Note
Providing a data model file name in the
catfile
and leaving the corresponding source catalog file name empty – same as setting'tweakreg_catalog'
in the ASN file to an empty string""
– would set the corresponding input data model’smeta.tweakreg_catalog
attribute toNone
. In this case,tweakreg_step
will automatically generate a source catalog for that data model.If an input data model is not listed in the
catfile
or does not have the'tweakreg_catalog'
attribute provided in the ASN file, then the catalog file name in that model’smeta.tweakreg_catalog
attribute will be used. Ifmodel.meta.tweakreg_catalog
isNone
,tweakreg_step
will automatically generate a source catalog for that data model.
Alignment
The source catalogs for each input image are compared to each other and linear (affine) coordinate transformations that align these catalogs are derived. This fit ensures that all the input images are aligned relative to each other. This step produces a combined source catalog for the entire set of input images as if they were combined into a single mosaic.
If the step parameter abs_refcat
is set to ‘GAIADR3’, ‘GAIADR2’, or ‘GAIADR1’,
an astrometric reference catalog then gets generated by querying
a GAIA-based astrometric catalog web service for all astrometrically
measured sources in the combined field-of-view of the set of input
images. This catalog is generated from the catalogs available
through the STScI MAST Catalogs and has the ability to account
for proper motion to a given epoch. The epoch is computed from the observation date and time
of the input data.
The combined source catalog derived in the first step then gets cross-matched and fit to this astrometric reference catalog. The pipeline initially supports fitting to the GAIADR3 catalog, with the option to select the GAIADR2 or GAIADR1 instead. The results of this one fit then gets back-propagated to all the input images to align them all to the astrometric reference frame while maintaining the relative alignment between the images.
For this part of alignment, instead of ‘GAIADR1’, ‘GAIADR2’, or ‘GAIADR3’, users can
supply an external reference catalog by providing a path to an existing
file. A user-supplied catalog must contain 'RA'
and 'DEC'
columns
indicating reference source world coordinates (in degrees). An optional column
in the catalog is the 'weight'
column, which when present, will be used
in fitting. The catalog must be in a format automatically recognized by
read()
.
Grouping
Images taken at the same time (e.g., NIRCam images from all short-wave
detectors) can be aligned together; that is, a single correction
can be computed and applied to all these images because any error in
telescope pointing will be identical in all these images and it is assumed
that the relative positions of (e.g., NIRCam) detectors do not change.
Identification of images that belong to the same “exposure” and therefore
can be grouped together is based on several attributes described in
ModelContainer
. This grouping is performed automatically
in the tweakreg
step using the
models_grouped
property, which assigns
a group ID to each input image model in meta.group_id
.
However, when detector calibrations are not accurate, alignment of groups
of images may fail (or result in poor alignment). In this case, it may be
desirable to align each image independently. This can be achieved either by
setting the image_model.meta.group_id
attribute to a unique string or integer
value for each image, or by adding the group_id
attribute to the members
of the input ASN
table - see ModelContainer
for more details.
Note
Group ID (group_id
) is used by both tweakreg
and skymatch
steps
and so modifying it for one step will affect the results in another step.
If it is desirable to apply different grouping strategies to the tweakreg
and skymatch
steps, one may need to run each step individually and
provide a different ASN as input to each step.
WCS Correction
The linear coordinate transformation computed in the previous step
is used to define tangent-plane corrections that need to be applied
to the GWCS pipeline in order to correct input image WCS.
This correction is implemented by inserting a v2v3corr
frame with
tangent plane corrections into the GWCS pipeline of the image’s WCS.
Step Arguments
The tweakreg
step has the following optional arguments:
Source finding parameters:
save_catalogs
: A boolean indicating whether or not the catalogs should be written out. This parameter is ignored for input data models whosemeta.tweakreg_catalog
is a non-empty string pointing to a user-supplied source catalog. (Default=False)use_custom_catalogs
: A boolean that indicates whether to ignore source catalog in the input data model’smeta.tweakreg_catalog
attribute. IfFalse
, new catalogs will be generated by thetweakreg
step. (Default=False)catalog_format
: Astr
indicating catalog output file format. (Default='ecsv'
)catfile
: Name of the file with a list of custom user-provided catalogs. (Default=''
)bkg_boxsize
: A positiveint
indicating the background mesh box size in pixels. (Default=400)starfinder
: Astr
indicating the source detection algorithm to use. Allowed values:'iraf'
,'dao'
,'segmentation'
. (Default='dao'
)snr_threshold
: Afloat
value indicating SNR threshold above the background. Required for all star finders. (Default=10.0)
Additional source finding parameters for DAO and IRAF:
kernel_fwhm
: Afloat
value indicating the Gaussian kernel FWHM in pixels. (Default=2.5)minsep_fwhm
: Afloat
value indicating the minimum separation between detected objects in units of number of FWHMs. (Default=0.0)sigma_radius
: Afloat
value indicating the truncation radius of the Gaussian kernel in units of number of FWHMs. (Default=2.5)sharplo
: Afloat
value indicating The lower bound on sharpness for object detection. (Default=0.2)sharphi
: Afloat
value indicating the upper bound on sharpness for object detection. (Default=1.0)roundlo
: Afloat
value indicating the lower bound on roundness for object detection. (Default=-1.0)roundhi
: Afloat
value indicating the upper bound on roundness for object detection. (Default=1.0)brightest
: A positiveint
value indicating the number of brightest objects to keep. If None, keep all objects above the threshold. (Default=200)peakmax
: Afloat
value used to filter out objects with pixel values >=peakmax
. (Default=None)
Additional source finding parameters for segmentation:
npixels
: Anint
value indicating the minimum number of connected pixels that comprises a segment (Default=10)connectivity
: Anint
value indicating the connectivity defining the neighborhood of a pixel. Options are4
, i.e., connected pixels touch along edges, or8
, i.e, connected pixels touch along edges or corners (Default=8)nlevels
: Anint
value indicating the number of multi-thresholding levels for deblending (Default=32)contrast
: Afloat
value indicating the fraction of total source flux an object must have to be deblended (Default=0.001)multithresh_mode
: Astr
indicating the multi-thresholding mode. Allowed values:'exponential'
,'linear'
,'sinh'
. (Default='exponential'
)localbkg_width
: Anint
value indicating the width of rectangular annulus used to compute local background around each source. If set to 0, then local background will not be subtracted. (Default=0)apermask_method
: Astr
indicating the method used to handle neighboring sources when performing aperture photometry. Allowed values:'correct'
,'mask'
,'none'
. (Default='correct'
)kron_params
: A tuple offloat
values indicating the parameters defining Kron aperture. If None, the parameters(2.5, 1.4, 0.0)
are used. (Default=None)
Optimize alignment order:
enforce_user_order
: a boolean value indicating whether or not take the first image as a reference image and then align the rest of the images to that reference image in the order in which input images have been provided or to optimize order in which images are aligned. (Default=False)
Reference Catalog parameters:
expand_refcat
: A boolean indicating whether or not to expand reference catalog with new sources from other input images that have been already aligned to the reference image. (Default=False)
Object matching parameters:
minobj
: A positiveint
indicating minimum number of objects acceptable for matching. (Default=15)searchrad
: Afloat
indicating the search radius in arcsec for a match. (Default=2.0)use2dhist
: A boolean indicating whether to use 2D histogram to find initial offset. (Default=True)separation
: Minimum object separation in arcsec. (Default=1.0)tolerance
: Matching tolerance forxyxymatch
in arcsec. (Default=0.7)xoffset
: Initial guess for X offset in arcsec. (Default=0.0)yoffset
: Initial guess for Y offset in arcsec. (Default=0.0)
Catalog fitting parameters:
fitgeometry
: Astr
value indicating the type of affine transformation to be considered when fitting catalogs. Allowed values:'shift'
: x/y shifts only'rshift'
: rotation and shifts'rscale'
: rotation and scale'general'
: shift, rotation, and scale
The default value is “rshift”.
Note
Mathematically, alignment of images observed in different tangent planes requires
fitgeometry='general'
in order to fit source catalogs in the different images even if mis-alignment is caused only by a shift or rotation in the tangent plane of one of the images.However, under certain circumstances, such as small alignment errors or minimal dithering during observations that keep tangent planes of the images to be aligned almost parallel, then it may be more robust to use a
fitgeometry
setting with fewer degrees of freedom such as'rshift'
, especially for “ill-conditioned” source catalogs such as catalogs with very few sources, or large errors in source positions, or sources placed along a line or bunched in a corner of the image (not spread across/covering the entire image).nclip
: A non-negative integer number of clipping iterations to use in the fit. (Default=3)sigma
: A positivefloat
indicating the clipping limit, in sigma units, used when performing fit. (Default=3.0)
Absolute Astrometric fitting parameters:
Parameters used for absolute astrometry to a reference catalog.
abs_refcat
: String indicating what astrometric catalog should be used. Currently supported options: ‘GAIADR1’, ‘GAIADR2’, ‘GAIADR3’, a path to an existing reference catalog,None
, or''
. Seejwst.tweakreg.tweakreg_step.SINGLE_GROUP_REFCAT
for an up-to-date list of supported built-in reference catalogs.When
abs_refcat
isNone
or an empty string, alignment to the absolute astrometry catalog will be turned off. (Default=''
)abs_minobj
: A positiveint
indicating minimum number of objects acceptable for matching. (Default=15)abs_searchrad
: Afloat
indicating the search radius in arcsec for a match. It is recommended that a value larger thansearchrad
be used for this parameter (e.g. 3 times larger) (Default=6.0)abs_use2dhist
: A boolean indicating whether to use 2D histogram to find initial offset. It is strongly recommended setting this parameter toTrue
. Otherwise the initial guess for the offsets will be set to zero (Default=True)abs_separation
: Minimum object separation in arcsec. It is recommended that a value smaller thanseparation
be used for this parameter (e.g. 10 times smaller) (Default=0.1)abs_tolerance
: Matching tolerance forxyxymatch
in arcsec. (Default=0.7)abs_fitgeometry
: Astr
value indicating the type of affine transformation to be considered when fitting catalogs. Allowed values:'shift'
: x/y shifts only'rshift'
: rotation and shifts'rscale'
: rotation and scale'general'
: shift, rotation, and scale
The default value is “rshift”. Note that the same conditions/restrictions that apply to
fitgeometry
also apply toabs_fitgeometry
.abs_nclip
: A non-negative integer number of clipping iterations to use in the fit. (Default = 3)abs_sigma
: A positivefloat
indicating the clipping limit, in sigma units, used when performing fit. (Default=3.0)save_abs_catalog
: A boolean specifying whether or not to write out the astrometric catalog used for the fit as a separate product. (Default=False)
Further Documentation
The underlying algorithms as well as formats of source catalogs are described in more detail at
https://tweakwcs.readthedocs.io/en/latest/
Further description of the input parameters and algorithms for star finding can be found at the following links:
Reference Files
The tweakreg
step uses the PARS-TWEAKREGSTEP parameter reference file.
PARS-TWEAKREGSTEP Parameter Reference File
- REFTYPE:
PARS-TWEAKREGSTEP
- Data model:
N/A
Reference Selection Keywords
CRDS selects appropriate pars-tweakregstep references based on the following keywords.
Instrument |
Keywords |
---|---|
FGS |
EXP_TYPE |
MIRI |
EXP_TYPE, FILTER |
NIRCAM |
EXP_TYPE, FILTER, PUPIL |
NIRISS |
EXP_TYPE, FILTER, PUPIL |
Standard Keywords
The following table lists the keywords that are required to be present in all reference files. The first column gives the FITS keyword names. The second column gives the jwst data model name for each keyword, which is useful when using data models in creating and populating a new reference file. The third column gives the equivalent meta tag in ASDF reference file headers, which is the same as the name within the data model meta tree (second column).
FITS Keyword |
Data Model Name |
ASDF meta tag |
---|---|---|
AUTHOR |
model.meta.author |
author |
DATAMODL |
model.meta.model_type |
model_type |
DATE |
model.meta.date |
date |
DESCRIP |
model.meta.description |
description |
FILENAME |
model.meta.filename |
N/A |
INSTRUME |
model.meta.instrument.name |
instrument: {name} |
PEDIGREE |
model.meta.pedigree |
pedigree |
REFTYPE |
model.meta.reftype |
reftype |
TELESCOP |
model.meta.telescope |
telescope |
USEAFTER |
model.meta.useafter |
useafter |
NOTE: More information on standard required keywords can be found here: Standard Required Keywords