The stability analysis of in-operation wind turbines is a very important topic, that has received considerable attention in the last years. Many identification algorithms have been developed to estimate frequencies and damping ratios, but very few papers have been dedicated to the mode shapes. The knowledge of high-resolution mode shapes could be exploited for several applications including model validation, accurate description of the vibratory content of a machine and spatially-accurate damage detection. In this work, we will present a procedure to compute the high-resolution periodic mode shapes of a wind turbine, and apply it to a high-fidelity wind turbine model. The results show that this methodology is able to identify the first low-damped modes of the system with good accuracy.