A Note on Naming Conventions

The Elixir, QSO, and DADS teams refer to several conceptually distinct periods of time as a 'run'. There is the `Camera run', the period of time when the instrument (CFH12K) is mounted on the telescope. This is the usual quantum of interest for Elixir processing. There is the `QSO (or queue) run', a contiguous period of time during which the QSO system is being used as the main mode of operation of the telescope. A single QSO run will always be contained within a single Camera run, but it is possible for a single Camera run to be divided into multiple QSO runs, or not to include any QSO runs at all. A `PI run' identifies a collection of images which have been obtained, either classically or in QSO mode for a specific PI. A `PI run' is not necessarily a contiguous period of time, but PI runs are only valid within a specific semester. We use 6 character codes to identify any of these runs. Camera runs are given names of the form 01Ak01, where the first two digits refer to the year, and the following letter to the semester in that year, the 'k' identifies this as the 'CFH12K' camera (as opposed to m for Megacam, w for Wircam, etc), and the last two digits identify the run within the semester. Similarly, QSO runs are given names of the form 01BQ01. PI runs have the form 01AH24, where the H identifies UH, C for Canada, F for France, D for discretionary time, E for engineering time, Q for QSO specific projects (ie, 01AQ97 = standard stars, 01AQ99 = detrend images). There will be some minor ambiguity between camera runs and Korean agency PI runs, but this should not introduce confusion.

Elixir End-of-Run Processing

At the end of each camera run, the Elixir system is used for a variety of tasks needed for complete calibration of the science images. The tasks include: the creation of the master detrend images, the astrometric calibration for each science and standard star image, the photometric analysis of the standard stars, and an assessment of the results from the photometric standards. The analysis steps are performed in sequence, as the later steps depend on the results of the earlier steps.

The detrend creation system involves the creation of master dark, bias, and flat images. Master fringe frames are created in a second step. Both steps are described in more detail elsewhere (REFERENCE), so we provide a summary here. All of the relevant images obtained during the run may potentially be used to generate a given master frame. The master frames which are generated by the Elixir system are associated with the science images by the period of time for which the master was created. For dark and bias images, only a single set of masters are needed for a typcial CFH12K run, as these data are quite stable with time. (Can we use the same dark \& bias masters run-to-run??). Flat-field images are frequently stable enough that a single master may be used for an entire run, but the option is available to sub-divide the run in case the flat-field structure changes. This may happen if dust is introduced by shutter or filter changes.

The creation of the master detrend frames involves an iterative process. A first-pass master is created by use all appropriate images. This master is used to generate residual frames, which are examined to identify signifcant outliers which should be excluded. This process is done by hand, with the aid of both convenient, scaled-down images and statistics on the individual images. The images which remain after outliers are excluded are then combined to create a second-pass master frame. This process is repeated until the residuals are sufficiently small.

Once master detrend frames are produced, they are inserted in a database which records the relevant information needed to identify a specific science image with the various types of detrend data. At this point, some of the flat-field images which have been created are corrected for a known error which is thought to be caused by scattered light. This correction is currently applied only to the BVRI flat-field images since only these have sufficient calibration information to ensure that the correction is an improvement. The newly corrected flats are also added to the detrend database, which maintain information needed both to distinguish the corrected and uncorrected versions and to supply the appropriate version as needed.

Master fringe frame creation can only occur once the previous set of master detrend frames are created. The fringe frames are created by combining a set of input images which have been already corrected for bias, dark, and flat. The details of the fringe creation can be found elsewhere. The master fringe creation is generally similar to the creation of other detrend frames. All of the science images which have enough sky flux to be useful are selected and prepared by being detrended. These images are then used to generate the master fringe frame in an iterative process. One significant difference from the other types of detrend frames is the method used to determine the strength of the fringe pattern. The fringe pattern is an additive term which sits on top of the base level sky emission. Since these two terms are only weakly correlated, it is necessary to distinguish the sky from the fringe strength. The analysis system measures the fringe strength for each input CCD image, and it is necessary to maintain the resulting fringe strength in the master frame so it is clear how to apply it to any science image.

After the various detrend frames have been created, all images of the sky are passed to the photometric and astrometric analysis

 Processing performed for each distributed image:
 
        - detrending & defringing
        - creation of jpeg images for each detrended images
        - insertion of keywords in headers
 
 Data products distributed:
 
        - detrended & defringed science images
        - detrended standard stars (short images not defringed)
        - all relevant master detrend images
        - html manifest (dads)
 
 Elixir databases used to relate images and reduced products 
 
        - detrend database
        - image zero point database
        - nightly zero point statistics database
 
 Outstanding issues:
 
        - low-frequency 'defringing'?