Pan-STARRS1 variability of XMM-COSMOS AGN. I. Impact on photometric redshifts
[Abbreviated] Upcoming large area sky surveys like EUCLID and eROSITA crucially depend on accurate photometric redshifts (photo-z). The identification of variable sources, such as AGNs, and the achievable redshift accuracy for varying objects are important in view of the science goals of the EUCLID and eROSITA missions. We probe AGN optical variability for a large sample of X-ray-selected AGNs in the XMM-COSMOS field, using the light curves provided by the Pan-STARRS1 (PS1) 3pi and MDF04 surveys. Utilizing two different variability parameters, we defined a sample of varying AGNs for every PS1 band. We investigated the influence of variability on the calculation of photo-z by applying three different input photometry sets for our fitting procedure. For each of the five PS1 bands, we chose either the epochs minimizing the interval in observing time, the median magnitude values, or randomly drawn light curve points to compute the redshift. In addition, we derived photo-z using PS1 photometry extended by GALEX/IRAC bands. We find that the photometry produced by the 3pi survey is sufficient to reliably detect variable sources provided that the fractional variability amplitude is at least 3%. Considering the photo-z of variable AGNs, we observe that minimizing the time spacing of the chosen points yields superior photo-z in terms of the percentage of outliers (33%) and accuracy (0.07), outperforming the other two approaches. Drawing random points from the light curve gives rise to typically 57% of outliers and an accuracy of 0.4. Adding GALEX/IRAC bands for the redshift determination weakens the influence of variability. Although the redshift quality generally improves when adding these bands, we still obtain not less than 26% of outliers and an accuracy of 0.05 at best, therefore variable sources should receive a flag stating that their photo-z may be low quality.