Astron. Astrophys. 334, 969-975 (1998)

3. Results

Data reduction and analysis were done by the help of software developed at the MPIA Heidelberg (see Köhler 1997). Data reduction consisted of excluding specklegrams with large noise, subtraction of the mean sky from each remaining specklegram, flat field correction, bad pixel correction and baseline subtraction. For the resulting frames a 2D FFT was applied. Finally, point source calibrated Fourier modulus (visibility) and phase (from bispectrum analysis) were averaged for each data cube. K band visibilities of the Z CMa image at 4 different analyzer orientations (averaged over all cubes) are shown in Fig. 1. Although close to the cut-off frequency, the plots of the visibility for the four polarizer settings show that the fringe pattern typical for a binary star was clearly detected. This allowed the application of a fitting algorithm to reproduce visibility and phase images by adjusting a binary model in the Fourier plane. Thus, the polarization derived from this fit has to be considered as the net (i.e. spatially unresolved) polarization of each component. Visibility and phase images were reproduced by adjusting a binary model in the Fourier plane. By this, separation and position angle of the binary and the contribution of each component to the total luminosity of the system were extracted. Afterwards we were able to derive the degree and position angle of the linear polarization of the binary components A and B on the basis of the intensities , measured at the four polarizer settings and the intensity ratios of the components (deduced before):

The total intensities were calculated using the requirement that and .

Fig. 1. Point source calibrated Fourier modulus (visibility) of the K band image of Z CMa at 4 different analyzer orientations. The box sizes correspond to about 28 arcsec-1 28 arcsec-1.

Since the data cubes for each polarizer setting were obtained sequentially, we cannot rule out variations of the seeing conditions. In order to minimize this effect, we used a novel statistical approach to derive the polarization properties and to assess their formal errors. The polarization of the Z CMa binary was calculated for all possible combinations of the measurements at 4 analyzer orientations. For 7 sets of data we got combinations. Polarization was deduced from the resulting distributions (see Fig. 2). Noise biasing of the polarization degree was corrected using the analytical formula (Wardle & Kronberg 1974)

where is the standard deviation of the observed polarization .

Fig. 2. Polarization degree and orientation of the Z CMa system and its components. The results correspond to the median values of the distributions of all possible (2401) combinations of measurements. Errors are given by standard deviations.

For the Z CMa binary a separation of and a position angle of were measured for the observational epoch (average over the measurements at all analyzer orientations). These results are in a fairly good agreement with those from speckle observations mentioned before. The trend of an increasing position angle with time due to the binary orbit should be checked by future observations.

© European Southern Observatory (ESO) 1998

Online publication: June 2, 1998

helpdesk.link@springer.de