2012A&A...540A.142M

In 2004, changes in the radio morphology of the Be/X-ray binary system LS I +61 303 suggested that it is a precessing microquasar. In 2006, a set of VLBA observations performed throughout the entire orbit of the system were not used to study its precession because the changes in radio morphology could tentatively be explained by the alternative pulsar model. However, a recent radio spectral index data analysis has confirmed the predictions of the two-peak microquasar model, which therefore does apply in LS I +61 303. We revisit the set of VLBA observations performed throughout the orbit to determine the precession period and improve our understanding of the physical mechanism behind the precession. By reanalyzing the VLBA data set, we improve the dynamic range of images by a factor of four, using self-calibration. Different fitting techniques are used and compared to determine the peak positions in phase-referenced maps. The improved dynamic range shows that in addition to the images with a one-sided structure, there are several images with a double-sided structure. The astrometry indicates that the peak in consecutive images for the whole set of observations describes a well-defined ellipse, 6-7 times larger than the orbit, with a period of about 28d. A double-sided structure is not expected to be formed from the expanding shocked wind predicted in the pulsar scenario. In contrast, a precessing microquasar model can explain the double- and one-sided structures in terms of variable Doppler boosting. The ellipse defined by the astrometry could be the cross-section of the precession cone, at the distance of the 8.4GHz-core of the steady jet, and 28d the precession period.