Quick links: calibrations | QC parameters | QC report | scoring & certification

Quality Control and documentation

The reprocessed UVES data set has a monitored and controlled quality. Data quality has been assured by

• using certified calibration products
• measurement of quality control parameters
• certification based on quality control parameters (scoring)

The data quality has been monitored by quality plots which also serve as spectral previews.

All calibration products in the archive have been certified earlier. They have been processed by earlier versions of the pipeline, analyzed and certified by the QC group, and stored in the archive, as part of the quality control process routinely implemented for all VLT and VLTI instruments.

The optimal association of calibrations and science data is another aspect of QC. It is guaranteed that always the closest-in-time certified calibration products are selected for processing. Thereby the instrument (and atmosphere) conditions during science exposures are generally matched best with those at the time when the calibrations were taken.

Find more about: UVES calibrations | calibration cascade

The pipeline measures properties of the product files. These numbers are called QC parameters. They form the backbone of the QC process for the UVES science data. The QC parameters are written into the product headers and into the QC1 database where they are available for statistical studies, plotting etc. The QC parameters of the UVES science data are available here: plotter | browser. The most important QC parameters are:

• qc_s2n (pipeline-derived S/N, averaged across the spectrum)
• mean_reduced (mean signal of extracted spectrum)
• qc_resolmed (resolving power, derived from the arclamp spectrum used for calibration)

Graphical reports are available to serve as previews and to indicate the quality of the products. There is one such report per CCD (i.e. one for a blue exposure, two for a red exposure, three in total for a dichroic twin exposure). The reports are stored in the archive and are part of the data set delivered in the downloads. They have names like r.UVES.2005-10-18T00:18:06.165_0000.fits_1.png (which belongs to the data set with ID UVES.2005-10-18T00:18:06.165).

Example of a QC report and preview for a UVES product file This product spectrum is always the extracted spectrum, not the flux-calibrated spectrum. The reason for this is that this type of spectrum always exists in a data set.
	Top panel: date, setup information, run ID. "product": this is the internal name (header key "PIPEFILE") of the extracted product spectrum. This name can be easily used to match all poducts derived from the same raw file (same data set). "sciname": this is the final name of the product, as it is archived; it includes information about the type, OB, time, and setup "OB, target": OB ID and target name "position": sketch of the slit (length and width to scale), the average object position, and the seeing
	Box 1: scans through the raw file; left: middle CCD row; right: middle CCD column (ADU vs. X pixels)
	Box 2: cross-dispersion profile in raw file, closeup view, on a logarithmic scale to show the full dynamic range (log ADU vs. X pixels)
	Box 3: Three spectral windows (6 A wide) of the product spectrum at full resolution. They are marked in plot 4 (below) by the red broken lines and labelled as a-c.Horizontal broken line indicates zero flux.
	Box 4: Preview of the product spectrum and its S/N. The full spectrum is shown here, with a 5 A rebinned version shown in red. The Y scale is in extracted ADU. Nominal positions of some prominent spectral lines are marked blue for orientation. The S/N plot is the ratio of the extracted spectrum and the errorbar file. It gives a crude indication of the spectral quality. It is not intended to replace precise measurements.
	Box 5: Raw data histograms. One histogram is shown around the background level (100...400 ADU; log frequency). The approximate BIAS level is marked by the red line. Any shift of the maximum away from the BIAS level indicates contributions from sky background. The second histogram focuses on the saturation limit and indicates any pixels in that range (the particular example here indeed shows saturation).
	Bottom panel: overview of exposure and quality control parameters. From left to right: Exposure time, airmass, seeing, slit width; wavelength start, end, step size (all in angstrom); number of orders, mean S/N, mean position, mean profile width.

top Scoring and certification

All UVES products have undergone a certification procedure. This was either automatic (assisted by scoring), or interactive. The scoring process has been tuned to apply for more than 90% of all UVES products created (after all, this is mass production).

[If you do not want to know more about the certification, you can skip this section.]

Scoring. The QC parameters have been used for automatic scoring of the product quality. This is a novel QC concept to assess the quality of pipeline products in an automatic way. The set of QC parameters, as measured by the pipeline, is tested against configured thresholds. For each parameter, compliance results in a score 0, while failure gives a score 1. In the process of UVES reprocessing, data products with a non-zero total score are flagged for further inspection, while products with score 0 are automatically certified.

The score results have been archived and are available in the requested data sets under the names <dateset_id>.html as HTML file, e.g. UVES.2005-07-03T04:36:52.519.html. There is also a more technical score file, <dataset_id>.tlog.

The scored parameters are:

• wlen_start (start of wavelength scale) and qc_nord (number of extracted orders): outliers may indicate a failure of the dispersion solution
• qc_fwhm: if this number is found to be below 0.5 arcs or above 3 arcs, this is most likely indicating a problem which needs inspection
• qc_resolmed: if this number deviates from average values expected for the slit width used, this may indicate a problem with the grating
• num_sat: the number of pixels in the raw file above 60,000 ADU
• diff_temp: temperature difference between science exposure and arclamp calibration; should be below 2 degrees, otherwise thermal grating drifts may result in a shift of the wavelength scale
• mean_airm: should be below 2, otherwise differential atmospheric dispersion may corrupt the extraction
• qc_pos: should be sufficiently off the slit edges, otherwise the extraction might be corrupted

Example of a UVES product score

Parameter score report UVES.2005-03-07T08:09:46.739.ab RAW_TYPE: SCI_POINT_ECH_RED setup: 600.0_RED_1_1_0.30 time range: 2005-01-21 ... 2005-04-21 Scores are collected per QC1 parameter.   wlen_start qc_nord qc_fwhm qc_resolmed num_sat diff_temp mean_airm qc_pos OK score result: 2/16 best: 0/16

This example contains scores for 8 QC1 parameters and their evaluation for the 2 red CCDs and product files. The little squares can be activated per mouse click to connect directly to the QC1 database and display a plot of that parameter and all other similar parameters in a time window spanning 3 months. The scoring scheme of UVES reprocessing is set up to display: green OK if total score = 0 yellow OK if score 1 or 2 red NOK is larger than 2

Certification. All products with non-zero scores have been reviewed. As a result of this review, there might have been comments about unusual features (e.g. "spectrum close to saturation but still looks ok"). These comments are meant to provide help or further information, but in no case imply any guarantee. In particular, there is also no guarantee for completeness. There might also always be products scored OK but having "hidden" problems.

The optional comment files always refer to a data set and are identfied by the name <dataset_id>.cmt, e.g. UVES.2005-03-07T08:09:46.739.cmt.