7 Conclusions

The process of data acquisition and reduction for polarization observations taken with the ESO ADONIS adaptive optics system has been described. Whilst certain precautions both in the observing method and in data reduction are required for imaging polarimetry and adaptive optics seperately, several other problems are presented arising from the combination of the two methods.

• Since the PSF varies in time the wavefront sensor reference star should be as similar as possible to the target object in terms of brightness (since the achieved Strehl ratio strongly depends on the reference star magnitude) and spectral characteristic, to ensure similar AO correction. Since the PSF varies across the field of view (depending on the anisoplanatic angle), it is also preferable to select the WFS reference star as close as possible to the target object. In practice this is rarely achieved when the target itself can not be used as WFS reference star. However a good estimate of the PSF provided by the reference star allows accurate deconvolution of the target without the introduction of artifacts arising from the differing PSF's.
• Imaging polarimetry requires good photometric conditions. By oversampling the $\cos(2\theta$) polarization curve at more than three position angles of the polarization analyser, an averaging over the photometric conditions is achieved. However depending on the time period of the photometric variations the averaging can result in zero polarization even from a substantially polarized source. In principle the use of an AO system should not compromise the photometric quality of the observations.
• The two polarization calibrations that are required impose the observation of an unpolarized source, to determine the instrumental polarization, and of a target with known polarization, to calibrate the angle of polarization. In the IR there is a very distinct lack of unpolarized and polarized standards. Stars with known very low optical polarization are suitable as IR unpolarized standards since the Serkowski interstellar polarization law shows that the polarization is much less in the IR than the optical. However polarized standards typically have a circumstellar origin to their high polarization and the value in the IR cannot be predicted. Many of the reflection nebulae around Young Stellar Objects have variable polarization and are therefore not ideal polarized standards.

Several strategies have been described for flat fielding and sky subtraction and it was shown how deviations from the expected cos(2$\theta$) curve can give an indication of the photometric conditions at the time of observation and allow any discrepant polarizer angles to be discarded as was found for the ADONIS polarizer at PA 157.5$^\circ$. The instrumental polarization for ADONIS was determined at 1.7% over the J, H and K range. Polarization maps have been succesfully produced for the reflection nebula around $\eta$ Carinae (the Homunculus). By using the PSF's determined from blind deconvolution at the same polarizer angles as the data, it has been shown that polarization structure can be revealed as close as two times the diffraction limit to a point source. The interpretation of the ADONIS AO polarization results on $\eta$ Carinae will be presented in a forthcoming paper (Walsh & Ageorges 1999).

Acknowledgements

We would like to thank the ESO ADONIS team for their advice and help during the development of the data reduction strategy. S.M. Scarrott is also acknowledged for useful comments on imaging polarimetry.