The instrumental polarization of the Nasmyth focus polarimetric differential imager NAOS/CONICA (NACO) at the VLT - Implications for time-resolved polarimetric measurements of Sgr A*

Context. We report on the results of calibrating and simulating the instrumental polarization properties of the ESO VLT adaptive optics camera system NAOS/CONICA (NACO) in the Ks-band. Aims. Our goal is to understand the influence of systematic calibration effects on the time-resolved polarimetric observations of the infrared counterpart of the Galactic center super-massive black hole at the position of Sagittarius A* (Sgr A*). Methods. We use the Stokes/Mueller formalism for metallic reflections to describe the instrumental polarization. The model is compared to standard-star observations and time-resolved observations of bright sources in the Galactic center. The differences between calibration methods are simulated and tested for three polarimetric Ks-band light curves of Sgr A*. Results. We find the instrumental polarization to be highly dependent on the pointing position of the telescope and about 4% at maximum. Given the statistical uncertainties in the data acquisition, the systematic effects of the employed calibration method are negligible at high-time resolution, as it is necessary and achieved for in the case of Sgr A*. We report a polarization angle offset of 13.2. due to a position angle offset of the lambda/2-wave plate with respect to the header value that affects the calibration of NACO data taken before autumn 2009. Conclusions. With the new model of the instrumental polarization of NACO it is possible to measure the polarization with an accuracy of 1% in polarization degree. The uncertainty of the polarization angle is <= 5 degrees for polarization degrees >= 4%. For highly sampled polarimetric time series we find that the improved understanding of the polarization properties gives results that are fully consistent with the previously used method to derive the polarization. The small difference between the derived and the previously employed polarization calibration is well within the statistical uncertainties of the measurements, and for Sgr A* they do not affect the results from our relativistic modeling of the accretion process.