The Panchromatic High-Resolution Spectroscopic Survey of Local Group Star Clusters - I. General Data Reduction Procedures for the VLT/X-shooter UVB and VIS arm
Our dataset contains spectroscopic observations of 29 globular clusters in the Magellanic Clouds and the Milky Way performed with VLT/X-shooter. Here we present detailed data reduction procedures for the VLT/X-shooter UVB and VIS arm. These are not restricted to our particular dataset, but are generally applicable to different kinds of X-shooter data without major limitation on the astronomical object of interest. ESO's X-shooter pipeline (v1.5.0) performs well and reliably for the wavelength calibration and the associated rectification procedure, yet we find several weaknesses in the reduction cascade that are addressed with additional calibration steps, such as bad pixel interpolation, flat fielding, and slit illumination corrections. Furthermore, the instrumental PSF is analytically modeled and used to reconstruct flux losses at slit transit and for optimally extracting point sources. Regular observations of spectrophotometric standard stars allow us to detect instrumental variability, which needs to be understood if a reliable absolute flux calibration is desired. A cascade of additional custom calibration steps is presented that allows for an absolute flux calibration uncertainty of less than ten percent under virtually every observational setup provided that the signal-to-noise ratio is sufficiently high. The optimal extraction increases the signal-to-noise ratio typically by a factor of 1.5, while simultaneously correcting for resulting flux losses. The wavelength calibration is found to be accurate to an uncertainty level of approximately 0.02 Angstrom. We find that most of the X-shooter systematics can be reliably modeled and corrected for. This offers the possibility of comparing observations on different nights and with different telescope pointings and instrumental setups, thereby facilitating a robust statistical analysis of large datasets.