3.11: WCS issues and FITS keywords

The DEIMOS specifications for user interaction with the images require that the images be produced with WCS information. WCS mechanisms built into the FITS headers of the DEIMOS give tangible benefits. When viewing a DEIMOS image frame they will permit quick look viewers to provide an instantaneous readout of the celestial coordinates of objects as the mouse tracks across the screen. When viewing a DEIMOS spectroscopy frame the WCS issues are more complex and would be most easily solved by looking at the table of slitlets in the mask. Either of these cases requires knowledge of the relative orientation of the detectors in the mosaic and of the distortions in the DEIMOS camera.

The draft describes 25 different WCS projections of the celestial sphere onto pixel coordinates. The authors also provide code libraries in both Fortran and C which perform the calculations. Unfortunately for DEIMOS none of the projections is able to handle an optical system with non-coaxial distortions. The given WCS projections should be able to map globally to about 10 pixels, or one arcsec. This is quite adequate for tracking the mouse over an image. More precise formulations will be required for the purposes of precision astrometry and slitmask design.

3.11.1: Enumeration of the coordinate systems

3.11.1.1: The Sky (Operational)

The responsibility for accuracy of input coordinates ultimately lies with the observer. Still, DEIMOS software should ensure that its handling of celestial coordinates is good enough not to be the cause of suboptimal slit positioning. The portions of DEIMOS which handle celestial coordinates (most notably the slitmask design) should make use of all astrometric terms which are relevant for extra-solar objects. (A description of the catalog tables is in section 7.5.3.4) If the astronomer cannot supply all astrometric quantities then the program should give a warning and supply reasonable defaults.

3.11.1.2: The Keck II axis (Operational)

3.11.1.3: The Nasmyth Focal Surface (Ideal)

In the DEIMOS design there is no object located on the Nasmyth Focal surface. There is no direct need to know the mapping between it and the sky. Conceptually, however, this mapping will be important for the design of the slitmasks.

3.11.1.4: The Slitmasks (Operational)

In practice it is likely that the tension on the slitmasks may globally deform them into a more nearly conical shape. In this case the mapping from the spherical Nasmyth Focal Surface to the Slitmask might resemble Lambertian projections common in cartography. The tension on the slitmasks will also tend to produce local deformations in regions between the support ribs.

For the purposes of slitmask manufacture the mapping from celestial coordinates to mask coordinates must be known to an accuracy of better than 0.1 arcsecond over the entire slitmask. The slitmask database should store the positions of the slitlets in mask coordinates, not in celestial coordinates. The CNC mill which produces the slitlets will require their coordinates to be specified in a Cartesian coordinate system aligned with the mask.

For the purpose of quick look reductions the mapping from CCD coordinates to slitmask coordinates will be required.

3.11.1.5: The DEIMOS Camera Focal Surface (Ideal)

There are radial distortions inherent in both the Keck II field of view and the DEIMOS camera field of view. The coordinate system in the camera focal surface is the product of these two non-coaxial distortions.

Because there is no precision imaging device which spans the entire surface there is no direct need to know the properties of the mapping between sky and camera focal plane. It will be conceptually useful for the CCD WCS and for slitmask design.

3.11.1.6: The DEIMOS CCDs (Operational)

For the purposes of quick look interaction with direct images the mapping between each CCD and the sky must be known. The same WCS mapping also serves archival purposes.

For the purposes of quick look interaction with spectral images the mapping between each CCD and the slitmask must be known. The coordinate on the slitmask will then be used to determine the sky position and/or observatory wavelength of any pixel.

3.11.1.7: The DEIMOS Guide Camera (Operational)

If the guide camera is used in a slit-viewing mode its field of view will overlap the slitmask. If the guide camera is used in a direct mode its field of view will be disjoint from the slitmask. A WCS for each scenario should be available prior to the observation at the time of slitmask design. The slitmask design program should use these draw a cartoon of the guider FOV within or next to the slitmask. This will assist the slitmask designer in choosing a field center and position angle that admit suitable guide stars.

3.11.2: Determination of the necessary mappings

3.11.2.1: Mappings between CCDs and the sky

It remains to be determined which of the WCS coordinate conventions will most nearly describe the DEIMOS focal plane. It may be necessary to define an extension to the FITS WCS mechanisms to handle the distortions to the desired accuracy. The predicted distortions of Keck II and DEIMOS camera should be analyzed in advance of first light.

3.11.2.2: Mapping between Keck II axis (DCS) and DEIMOS CCD images

If possible this mapping should be determined simultaneously with the CCD WCS mentioned just above. If the tertiary can be switched rapidly to feed DEIMOS alternately with an imaging device at the other Nasmyth or at the bent Cassegrain focus then the coordinate offset between the DCS and the DEIMOS CCD images should be measurable via classical astrometry. Alternatively it may be possible to place a small camera in the Keck II focal plane between the two slitmasks.

3.11.2.3: Mappings between the CCDs and the slitmask

The WCS for spectrally dispersed images will require another mask (not yet designed) which can be called the ``Gridslit Mask''. This mask should consist of a large number of slitlets which fill the width of the spatial direction of the mask. The slitlets should be offset in the dispersion direction of slitmask such that they sample most of the height of the mask. DEIMOS should provide a fully-automated procedure which can determine the slitlet transformations from a Gridslit Mask exposure. These transformations should be determined using various gratings at various angles.

In addition to the holes for metrology, both of these slitmasks should have small but unique patterns which fall onto each of the CCDs. These unique patterns are helpful for verifying that the readout orientation of each CCD has been correctly determined. These holes might be in the form of words or asymmetric letters of the alphabet.

The illumination of the slitmask in the laboratory may not match the illumination through the Keck II pupil. Furthermore, the Keck II pupil rotates w.r.t. the slitmask. For these reasons the slitmask WCS transformations should be verified after first light.

3.11.2.4: Determination of WCS for the other four CCDs

The mappings along the purely spatial dimension (rows of the CCDs) can be determined from such a calibration spectrum. The mappings along the direction of spectral dispersion are more complex; they involve both spatial and spectral information. As of the time of this PDR we do not have a WCS for describing spectral data. Its complexity will be sufficient that we expect to store it as a FITS table rather than as FITS keywords.

3.11.2.5: Mapping between sky and slitmask

At the time of observation the mapping between sky and slitmask may be better known than it was at the time of mask design and manufacture. The database should keep a record of the transformation in each case, and WCS information from both should be written into the FITS tables that accompany each CCD image. At the time of observation (or later) it should be possible for the user to see coordinates based on either the design-time or observation-time WCS.

3.11.2.6: Mapping between Guide Camera and anything else

3.11.3: Mapping from pixel coordinates to sky coordinates

3.11.3.1: Keywords derived from DCS

• RA_PNT
• DEC_PNT
• HA_PNT
• RADECPNT
• EQUINPNT
• MJD_PNT

3.11.3.2: WCS Keywords applying to all CCDs

• DEISKYPA
• DEIMOSPA
• RADECSYS
• DATE-OBS
• MJD-OBS
• MJD_WCS
• EQUINOX

3.11.3.3: WCS Keywords applying to individual CCDs

if (abs(tan(SkyPA)) < 1) {
	# slitlets are more nearly N-S
	SmTYPE1 = 'DEC-....'
	SmTYPE2 = 'RA--....'
} else {
	# slitlets are more nearly E-W
	SmTYPE1 = 'RA--....'
	SmTYPE2 = 'DEC-....'
}