Data Reduction Frequently Asked Questions

The response curve derived from my standard star looks very bad and the raw response shows huge residuals. Why?

The pipeline supports seven standard stars, for which high-resolution reference data based on model spectra exists: EG 21, EG 274, Feige 110, GD 71, GD 153, LTT 3218, and LTT 7987. For other standard stars observed with UVES the reference data were mostly derived from low-resolution observations and are sampled at coarse intervals (typically 50 Angstrom).

The figure below shows an example for the standard star HR 9087 observed with the 390 setup.
The left plot shows the ratio of the observed spectrum and the interpolated reference data (black) and the points used to fit the response (red). These fit points were optimized for the new high-resolution reference data, which yield raw response curves with small residuals at line cores only. Obviously they are not suitable for this observation.
The right plot illustrates this problem by showing in black the reference data, in red the interpolated reference spectrum and in blue the observed spectrum, scaled to fit into the plot. The observed lines are obviously much better resolved than the reference data. So even if the fit points were revised the response curve will not be good. If you absolutely need to use such a standard star you have to refine the fit points by editing the corresponding entry in the static calibration file RESP_FIT_POINTS_CATALOG found at:
ftp://ftp.eso.org/pub/dfs/pipelines/instruments/uves/uves_response_old_calib.tar.gz.