The arc lamp exposures with the slit are not used for processing but only
for quality control monitoring. The pipeline identifies the lines using
the same catalog as for the distortion correction. Then the resolution
and the difference between the position of a given line in the 9-pinhole
observations and that in the slit observations is determined. The physical
model configuration parameters are not updated.
The number of found lines and the median difference is scored.
The threshold for the number of lines is 10% of the median value.
The thresholds for the difference values is based on statistical grounds.
VIS lamp provides about 30% more flux; more lines found
UVB lamp provides about 20% more flux; more lines found
pipeline version 1.1.6 has changed the values of several QC parameters, esp. line intensity, number of lines found and differences with respect to model predictions
pipeline problems with number of accepted lines in NIR
Ne lamp in NIR arm dies
Ne lamp in NIR arm replaced
Xe lamp in NIR arm replaced; Ne lamp in NIR arm aligned correctly
new slit mask in NIR: QC parameters may change slightly due to manufacturing tolerances; 0.4" pinhole and 1.5"x11" slit are replaced by 0.6"x11" and 0.9"x11" slits with K-band blocking (resulting in fewer orders and lines for these slits (0.6x11JH, 0.9x11JH) compared to their unblocked equivalents.
Flat field lamp replacements are marked in the flat spectrum health check plots.
The deviations of the 9-pinhole positions from the physical model fit, as reported in the QC report, have been marked by different colors by order.
The resolving power is an instrument key performance indicator and is scored.
For SLIT, the current thresholds are 5% of the median, for IFU, with a larger intrinsic scatter, the thresholds are 10% of the median.
During July 4-8, 2011, a new slit mask was installed in the NIR arm. The 0.4" pinhole and the 1.5"x11" slit were replaced by an 0.9"x11" and an 0.6"x11" slit with K-band blocking (0.9x11JH, 0.6x11JH). The change of slit mask resulted also in small changes of resolution for the other slits since the new and old slits are not exactly identical (manufacturing tolerances).
We modified the hitherto used default command line parameter of the xsh_wavecal recipe from --followarclines-search-window-half-size=6 to --followarclines-search-window-half-size=13, which allows for a larger window pixel size when searching for arc lines. For 1x2 binning arc frames, the parameter covers a double as large wavelength range than for 1x1 binning, (since binning two detector pixel into one frame pixel covers the same wavelength range). The effect is negligible for IFU frames, 1x2 binning setups and 1x1 narrow slit arc frames, and strongest for wide slit 1x1 binning slit arc frames.
The spectral resolution (= resolving power) reported by the xshooter pipeline was biased by about 19% (1x1 binning UVB 1.0x11) and 15% (1x1 binning VIS 0.9x11) arc frame calibrations from begin of operations on. The pipeline arc frame recipe underestimated the spectral resolution. The pipeline results for 1x2 binning data were closer to measurements retrieved directly from the product frames. The trending plot for 1x1 binning arc frames did not show the absolute resolving power, but still monitored the long-term trend.