Previous: Flatfielding
Up: Calibration exposures
Next: Flux calibration
Previous Page: Flatfielding
Next Page: Flux calibration

Wavelength calibration

The relation between wavelength and position on the detector is not linear and it is necessary to determine this relationship through observations of wavelength calibration sources. The presence of optical distortion means that there will be small differences in this relationship between different spectra in the field so it is important to derive this relationship separately for each slit in the mask.

In addition, in common with all spectrographs, the instrument flexes slightly as the telescope tracks across the sky. The main manifestation of this is a slight shift in the position of spectral features during very long exposures. In general, it is good idea to break long exposures into several shorter sections of 1000-2000s to allow cosmic ray events to be removed, and this also allows the gradual shift in the wavelength-position relation to be monitored by taking exposures of wavelength-calibration light sources.

The worst shift in the position of a spectral line on the detector that was measured during flexure tests at RGO was found to be 6m for a change of attitude of 15. Measurements at the telescope confirm the test results. This amounts to about 0.25 pixel/hour in the worst case.

Wavelength calibration is quite difficult because of problems in finding a suitable arc lamp which provides enough unblended lines over the full spectral range of LDSS-2. After much experimentation, it was found that a CuAr lamp was the best even though the lines in the blue are very faint. As for the dispersed flat-fields, it may be worth using a colour filter to attenuate the red lines, or perhaps use a long exposure for the blue lines and a short exposure for the red. In order to aid the observer, an atlas of emission line identifications has been prepared (see Appendix A). Note that some of these lines have wavelengths which have been determined empirically.