Author: Warren Hack, Nadezhda Dencheva
Date: 12 Oct 2010
Abstract
The ‘headerlet’ serves as a mechanism for encapsulating WCS information which can be used to update the WCS solution of an image. This object needs to be as compact as possible while providing an unambigious and self-consistent WCS solution for an image while requiring a minimum level of software necessary to apply the headerlet to an image. These basic requirements come from the desire to create a mechanism for passing improved astrometric solutions for HST data and provide those solutions in a manner that would not require getting entirely new images from an archive when only the WCS information has been updated. This report describes the format and contents of a headerlet along with the procedures which would be used to update images with the updated WCS information from the headerlet.
The HST Archive provides access to all HST data, most of which can not be readily combined together due to errors in guide star astrometry imposing offsets between images taken using different pairs of guide stars. A lot of effort has gone into computing those offsets so that all the images taken at a particular pointing can be combined successfully with astrometry matched to external astrometric catalogs. Unfortunately, there is no current mechanism for passing those updated solutions along to the community without providing entirely new copies of all the data.
The concept of a ‘headerlet’ seeks to provide a solution where only the WCS solution for an image that has been aligned to an astrometric catalog can be archived and retrieved for use in updating copies of that image’s WCS information without getting the image data again. Multiple ‘headerlets’ could even be provided with each representing the alignment of an image to a different astrometric solution, giving the end user the option to get the solution that would allow them to best align their images with external data of interest to them. These benefits can only be realized with the proper definition of a ‘headerlet’ and the procedures used to define them and apply them to data.
All users get their copy of an image from the HST Archive (OTFR) after being processed with the latest image calibrations, including applying the latest available distortion models. The calibrated image gets passed to the user with the _flt.fits suffix and is referred to as the FLT image. These FLT images serve as the inputs to MultiDrizzle in order to apply the distortion models and combine the images into a single DRZ product.
The WCS information in the FLT images has been updated to include the full distortion model, including the full polynomial solution from the IDCTAB and all the corrections formerly combined into the DGEOFILE. The FITSConventions report by Dencheva (currently available online) contains the full description of the conventions used to describe all these components in a FITS file. The header now contains the following set of keywords and extensions to fully describe the WCS with distortion:
- Linear WCS keywords: specifically, CRPIX, CRVAL, CTYPE, CD matrix keywords
- SIP coefficients: A_*_* and B_*_*, A_ORDER, B_ORDER, OCX10, OCX11, OCY10, and OCY11 keywords
- NPOL file: if an NPOLFILE has been specified for the image, CPDIS and DP keywords to point to WCSDVARR extensions (Paper IV convention)
- Column correction file: if a D2IMFILE has been specified for use with the image, the D2IMEXT and D2IMERR keywords signify the use of a D2IM file extension
- WCSDVARR extensions: 2 extensions for each chip with lookup tables containing the NPOLFILE corrections, with each extension corresponding to an axis of the image (X correction or Y correction) (if an NPOLFILE has been applied to the image)
- D2IMVARR extension: an extension with a lookup table containing the column-correction for that chip from the D2IMFILE.
Each science header will have its own set of these keywords and extensions that will be kept together as part of the headerlet definition. This avoids any ambiguity as to what solution was used for any given WCS. A summary of all the WCS solutions for all the chips can be written out to the WCSCORR table, a binary FITS table extension described online.
An ACS/WFC exposure would end up with the following set of extensions:
EXT# FITSNAME FILENAME EXTVE DIMENS BITPI OBJECT
0 j8hw27c4q_flt j8hw27c4q_flt.fits 16
1 IMAGE SCI 1 4096x2048 -32
2 IMAGE ERR 1 4096x2048 -32
3 IMAGE DQ 1 4096x2048 16
4 IMAGE SCI 2 4096x2048 -32
5 IMAGE ERR 2 4096x2048 -32
6 IMAGE DQ 2 4096x2048 16
7 IMAGE D2IMARR 1 4096 -32
8 IMAGE WCSDVARR 1 64x32 -32
9 IMAGE WCSDVARR 2 64x32 -32
10 IMAGE WCSDVARR 3 64x32 -32
11 IMAGE WCSDVARR 4 64x32 -32
12 BINTABLE WCSCORR 18Fx10R
There may be a lot of extensions appended to this FITS file, but the sum total of all these new extensions comes to approximately 100kB for ACS/WFC images (our sample only requires 86400 bytes), making them a space efficient means of managing all the distortion and WCS information.
The headerlet needs to be a self-consistent, fully described definition of a WCS and its distortion. The WCS and SIP coefficients get derived from the SCI header directly, along with the keywords which refer to the extensions with the NPOLFILE and D2IMFILE corrections. The science observation will be stripped from the original WCS which will be saved to a headerlet so that the user can always revert back to the original WCS solution at any time. There will be an option to permanently delete the original WCS and not save it to a headerlet.
A new extension, named SIPWCS, containing all the WCS-related keywords from the SCI header, including all keywords referring to NPOL and D2IM extensions as well as all sets of alternate WCS keywords, will be created to serve as the record of the original WCS. Keywords (TBD) recording the alignment information are recorded in this header as well. All the sets of linear WCS keywords stored using FITS Paper I Multiple WCS Standard would be defined using the same set of distortion coefficients written to the SIP keywords and NPOL files. This insures that all the information in the header remains consistent. The keywords in this extension can be used to overwrite the keywords in the corresponding SCI header to update the WCS solution for each chip without any further modification or computation. The new extension then serves not only as a record of all the WCS solutions derived for the image, but also the source of values for restoring the SCI header WCS when desired.
This new extension along with the NPOLFILE and the D2IMFILE extensions fully describe the WCS of each chip and can serve without further modification as the definition of the headerlet. The listing of the FITS extensions for a headerlet for the sample ACS/WFC exposure after writing it out to a file would then be:
EXT# FITSNAME FILENAME EXTVE DIMENS BITPI OBJECT
0 j8hw27c4q j8hw27c4q_hdr.fits 16
1 IMAGE D2IMARR 1 4096 -32
2 IMAGE WCSDVARR 1 64x32 -32
3 IMAGE WCSDVARR 2 64x32 -32
4 IMAGE WCSDVARR 3 64x32 -32
5 IMAGE WCSDVARR 4 64x32 -32
6 IMAGE SIPWCS 1 8
7 IMAGE SIPWCS 2 8
This file now fully describes the WCS solution for this image, complete with all the distortion information used to originally define the solution. No further reference files or computations would be needed when this headerlet gets used to update an image.
The primary header must have 4 required keywords:
HDRNAME - a unique name for the headerlet
DESTIM - target image filename (the ROOTNAME keyword of the original archive filename)
STWCSVER - version of STWCS used to create the WCS of the original image
PYWCSVER - version of PyWCS used to create the WCS of the original image
The headerlet defined above serves as the default headerlet for any image provided by the HST Archive. However, should the user perform their own calibrations which they feel improve on the standard calibrations provided by the pipeline, a custom headerlet can be provided. Any headerlet should simply include:
- Required: A primary header with specific keywords which specify a unique headerlet name and a targeted image.
- Required: An SIPWCS extension for each chip which contains the linear WCS as well as any distortion model supported by FITS (for example, updated SIP coefficients)
- Optional: Any additional look up tables with refinements to the polynomial solutions in the SIPWCS extension. Any such extensions should be linked to the SIPWCS extension using the same Paper IV conventions used for the NPOLFILE tables.
- Optional: Detector to image correction array as a separate extension if needed.
This custom headerlet should be capable of being used to overwrite the existing SCI header WCS keywords to provide a FITS-supported WCS.
Updating an image retrieved from the HST Archive with a headerlet only requires a few very simple steps:
- Create a headerlet from the original WCS solution in the science image (this step can be turned off).
- Delete all WCS information from the science image
- Copy the WCS solution from the headerlet to the science observation
- Update the WCSCORR table with the linear WCS keyword values and name of the SIP solution (based on the name of the reference files) from each SIPWCS extension from the headerlet, along with the keyword values from the PRIMARY header of the headerlet
This process assumes that when an image gets updated with a headerlet, the new solution from the headerlet should become the prime WCS. Further implementations of the software to work with headerlets can expand on this functionality if necessary. Initially, the headerlet simply needs to be used to update the image’s FITS file so that the WCS information can be used at all.
Implementing support for the headerlet and its use in updating HST FITS files will require a few new software tasks; namely,
The operation of updating a science file with a headerlet only requires the use of basic FITS operations:
These operations do not require any computations and can be done using any FITS library. This allows a headerlet to be usable by the community even if they do not use the software we develop based on PyFITS and STWCS, both for creating and applying these files.
This section describes the current draft API for working with headerlets as implemented in the stwcs.wcsutil.headerlet module. First, there’s a potentially confusing point that should be cleared up: A headerlet, as implemented, is simply a FITS file containing multiple extensions that contain all the parameters necessary to reproduce the WCS solution in the science image it was created from. When a headerlet is applied to an image, a copy of the original headerlet file is appended to the image’s HDU list as a special extension HDU called a Headerlet HDU. A Headerlet HDU consists of a simple header describing the headerlet, and has as its data the headerlet file itself, (which may be compressed). A Headerlet HDU has an ‘XTENSION’ value of ‘HDRLET’. Though PyFits can handle such a non-standard extension type sensibly, this hasn’t been tested with other common FITS readers yet. If it becomes necessary, Headerlet HDUs could be implemented using a standard extension type like ‘IMAGE’.
To create a headerlet from an image, a createHeaderlet() function is provided:
>>> from stwcs.wcsutil import headerlet
>>> hdrlet = headerlet.createHeaderlet('j94f05bgq_flt.fits', 'VERSION1')
>>> type(hdrlet)
<class 'stwcs.wcsutil.headerlet.Headerlet'>
>>> hdrlet.info()
Filename: (No file associated with this HDUList)
No. Name Type Cards Dimensions Format
0 PRIMARY PrimaryHDU 12 ()
1 SIPWCS ImageHDU 111 ()
2 SIPWCS ImageHDU 110 ()
3 WCSDVARR ImageHDU 15 (65, 33) float32
4 WCSDVARR ImageHDU 15 (65, 33) float32
5 WCSDVARR ImageHDU 15 (65, 33) float32
6 WCSDVARR ImageHDU 15 (65, 33) float32
7 D2IMARR ImageHDU 12 (4096,) float32
As you can see, the Headerlet object is similar to a normal pyfits HDUList object. createHeaderlet() can be given either the path to a file, or an already open HDUList as its first argument.
What do you do with a Headerlet object? Its main purpose is to apply its WCS solution to another file. This can be done using the Headerlet.apply() method:
>>> hdrlet.apply('some_other_image.fits')
Or you can use the applyHeaderlet() convenience function. It takes an existing headerlet file path or object as its first argument; the rest of its arguments are the same as Headerlet.apply(). As with createHeaderlet() both of these can take a file path or opened HDUList objects as arguments.
When a headerlet is applied to an image, an additional headerlet containing that image’s original WCS solution is automatically created, and is appended to the file’s HDU list as a Headerlet HDU. However, this behavior can be disabled by setting the createheaderlet keyword argument to False in either Headerlet.apply() or applyHeaderlet().
When opening a file that contains Headerlet HDU extensions, it will normally look like this in PyFits:
>>> import pyfits
>>> hdul = pyfits.open('94f05bgq_flt_with_hlet.fits')
>>> hdul.info()
Filename: j94f05bgq_flt_with_hlet.fits
No. Name Type Cards Dimensions Format
0 PRIMARY PrimaryHDU 248 () int16
1 SCI ImageHDU 286 (4096, 2048) float32
2 ERR ImageHDU 76 (4096, 2048) float32
3 DQ ImageHDU 66 (4096, 2048) int16
4 SCI ImageHDU 282 (4096, 2048) float32
5 ERR ImageHDU 74 (4096, 2048) float32
6 DQ ImageHDU 66 (4096, 2048) int16
7 WCSCORR BinTableHDU 56 10R x 23C [40A, I, 1A, D, D, D, D, D, D, D, D, 24A, 24A, D, D, D, D, D, D, D, D, J, 40A]
8 WCSDVARR ImageHDU 15 (65, 33) float32
9 WCSDVARR ImageHDU 15 (65, 33) float32
10 WCSDVARR ImageHDU 15 (65, 33) float32
11 WCSDVARR ImageHDU 15 (65, 33) float32
12 D2IMARR ImageHDU 12 (4096,) float32
13 HDRLET NonstandardExtHDU 13
14 HDRLET NonstandardExtHDU 13
The names of the headerlet extensions are both HDRLET, but its type shows up as NonstandardExtHDU. Their headers can be read, and while their data can be read you’d have to know what to do with it (the data is actually either a tar file or a gzipped tar file containing the headerlet file). However, if you have stwcs.wcsutil.headerlet imported, PyFits will recognize these extensions as Headerlet HDUs:
>>> import stwcs.wcsutil.headerlet
>>> # Note that it's necessary to reopen the file
>>> hdul = pyfits.open('j94f05bgq_flt_with_hlet.fits')
>>> hdul.info()
Filename: j94f05bgq_flt_with_hlet.fits
No. Name Type Cards Dimensions Format
0 PRIMARY PrimaryHDU 248 () int16
1 SCI ImageHDU 286 (4096, 2048) float32
2 ERR ImageHDU 76 (4096, 2048) float32
3 DQ ImageHDU 66 (4096, 2048) int16
4 SCI ImageHDU 282 (4096, 2048) float32
5 ERR ImageHDU 74 (4096, 2048) float32
6 DQ ImageHDU 66 (4096, 2048) int16
7 WCSCORR BinTableHDU 56 10R x 23C [40A, I, 1A, D, D, D, D, D, D, D, D, 24A, 24A, D, D, D, D, D, D, D, D, J, 40A]
8 WCSDVARR ImageHDU 15 (65, 33) float32
9 WCSDVARR ImageHDU 15 (65, 33) float32
10 WCSDVARR ImageHDU 15 (65, 33) float32
11 WCSDVARR ImageHDU 15 (65, 33) float32
12 D2IMARR ImageHDU 12 (4096,) float32
13 HDRLET HeaderletHDU 13
14 HDRLET HeaderletHDU 13
>>> print hdul['HDRLET', 1].header.ascard
XTENSION= 'HDRLET ' / Headerlet extension
BITPIX = 8 / array data type
NAXIS = 1 / number of array dimensions
NAXIS1 = 102400 / Axis length
PCOUNT = 0 / number of parameters
GCOUNT = 1 / number of groups
EXTNAME = 'HDRLET ' / name of the headerlet extension
HDRNAME = 'j94f05bgq_orig' / Headerlet name
DATE = '2011-04-13T12:14:42' / Date FITS file was generated
SIPNAME = 'IDC_qbu1641sj' / SIP distortion model name
NPOLFILE= '/grp/hst/acs/lucas/new-npl/qbu16424j_npl.fits' / Non-polynomial corre
D2IMFILE= '/grp/hst/acs/lucas/new-npl/wfc_ref68col_d2i.fits' / Column correction
COMPRESS= F / Uses gzip compression
HeaderletHDU objects are similar to other HDU objects in PyFits. However, they have a special headerlet attribute that returns the actual headerlet contained in the HDU data as a Headerlet object:
>>> hdrlet = hdul['HDERLET', 1].headerlet
>>> hdrlet.info()
Filename: (No file associated with this HDUList)
No. Name Type Cards Dimensions Format
0 PRIMARY PrimaryHDU 12 () uint8
1 SIPWCS ImageHDU 111 () uint8
2 SIPWCS ImageHDU 110 () uint8
3 WCSDVARR ImageHDU 15 (65, 33) float32
4 WCSDVARR ImageHDU 15 (65, 33) float32
5 WCSDVARR ImageHDU 15 (65, 33) float32
6 WCSDVARR ImageHDU 15 (65, 33) float32
7 D2IMARR ImageHDU 12 (4096,) float32
This is useful if you want to view the contents of the headerlets attached to a file.