Subsections

• 7.1.1 The use of SCDs within AAC

7.1 AAC's General Approach

AAC is the stage of the pipeline in which most of the work is done to generate products in physical units. As expected with an automatic procedure with no intervention possible on behalf of the observer, the final results may not be as good as those possible with an interactive system like CIA, for example. Experience has shown, however, that the pipeline products archived in IDA are of consistently high enough quality to offer a good overall assessment of the scientific merit of the data and enable the observer to make an initial astronomical interpretation using the set of up to 8 different data products provided. The tasks performed by AAC may be summarised as follows, showing the data products ( $\Longrightarrow$) in which the results appear:

• AAC operates on a single prime or parallel observation;
• data are assembled into coherent units in memory;
• static pointing and ideal calibration references are added;
• cosmic-ray glitches are detected and flagged $\Longrightarrow$CGLL;
• the amplitude of spacecraft pointing jitter is assessed $\Longrightarrow$CJAM;
• Fouks-Schubert transient model correction is applied;
• detectors are checked for incomplete illumination;
• attention is drawn to any saturation problems;
• images are calculated in detector coordinates for all data $\Longrightarrow$CCIM;
• sky images are calculated in celestial coordinates $\Longrightarrow$CMAP;
• sky mosaics are calculated in celestial coordinates $\Longrightarrow$CMOS;
• point sources are detected $\Longrightarrow$CPSL;
• source spectra are calculated $\Longrightarrow$CSSP;
• archive summary data are generated;
• comments are made on each stage of the analysis $\Longrightarrow$CUFF.

AAC's raw ingredients are the revolution's EOHA and EOHC; and CISP SPD and IRPH & IIPH pointing files for prime data or CPSP SPD and CRPH & CIPH pointing files for parallel data; and the set of CAL-G files. The pipeline was designed to offer images and point source measurements in formats as close to standard FITS as possible. Individual calibrated images are reported, for example, in the CMAP file in which, as explained above, FITS PRIMARY image conventions are reproduced in the columns of TABLE[1] over 3 consecutive rows for FLUX, FLUX_ERROR and EXPOSURE. These occur often and are referred to below as CCIM or CMAP or CMOS 3-row images. If more than one part of the sky was observed, such as in a raster or in the beam-switch observing mode, individual CMAP images are combined in the CMOS. Similarly, if a detected point source was observed but not necessarily detected at more than one wavelength, individual CPSL fluxes are combined for the spectra in the CSSP.

AAC's work from beginning to end is reported in the CUFF, or CAM User-Friendly log File, which prints details of the procedures executed and summaries of the results obtained. One of its useful jobs is to show which calibration components were used in the analysis, especially when the exact instrumental configuration was unavailable. The following example CUFF extract shows that, while an optical flat-field of the same optical configuration was available, the nearest detector flat-field in selection parameter space had to be used.

     Extraction of components from CAL-G files
     LW Dflt       EWHL    FCVF    SWHL    PFOV    TINT
     Tried:         308     140      88     360      15
     Got:           308     125     220     192      36
     LW Oflt       PFOV    EWHL    FCVF    SWHL    TINT
     Tried:         360     308     140      88      15
     Got:           360     308     140      88      15

7.1.1 The use of SCDs within AAC

Data are divided into the longest coherent units during which the configurations of satellite and instrument were constant, incorporating pointing and calibration data. These units are known as Standard Calibrated Data, or `SCDs', and the user will find frequent references to them in the CUFF and other data products. The SCD boundaries are decided empirically, by a process that has become known as `slicing', on the basis of a change in any of the following 14 parameters:

OBST	operating mode
CNFG	configuration counter
DEID	LW or SW detector
EWHL	aperture entrance wheel setting
SWHL	mirror selection wheel setting
PFOV	pixel size lens wheel setting
FCVF	waveband filter wheel setting
GAIN	detector gain
PROC	on-board processing mode
ACSA	number of accumulated or sampled images
BSFG	beam-switch flag
ITIM	integration time
RPID	raster-point ID
IID	prime instrument aperture

all but the last two of which are from the SPD record. The RPID and IID are taken from the pointing files. As the instrument operated continuously, passing an uninterrupted stream of EOI and RESET frames into telemetry, SCDs are of types OBS, DRK, CAL, CLN or IDLE depending on the operating mode. A number of SCDs form the configurations, or `CNFs', which in turn form the AOT that encapsulates the entire observation. Each of AAC's tasks of calibration or data reduction takes place by manipulation of SCDs either individually or in the groups forming the CNFs or the AOT as appropriate. SCDs are designated 16-character names CSCDnnnnnnnnccii, where nnnnnnnn is the 8-digit observation number, cc the 2-digit configuration number and ii is a 2-character alphanumerical index. These SCD names are to be found, for example, in CCIM[1].DATAREF(*) and are regularly mentioned in the CUFF.

The assembly of SCD structures marks the end of the frame-by-frame character of ERD and SPD that reflects their operational function with the recognition of the astronomical context and scientific coherence of the data. For example, each SCD has a pointing direction - that remains undefined for DRK, CAL, CLN or IDLE SCDs; images that are stored in explicit 2-D $32\times 32$ structures; and a set of references showing which calibration components should ideally be used in analysis. Every pixel readout has associated an observed value, a mask and a model value, reflecting the non-destructive approach to data analysis adopted by AAC. The mask is used to signal various conditions that might be detected during the course of analysis, such as that the pixel was dead or had been affected by a cosmic-ray glitch, and serves as the basis for the inclusion of individual pixels in calculations or data products. The model value is used to store the reconstructed value from the application of Fouks-Schubert transient modelling.