The aim of this work is to compute quantitative measures of turbulence intensity and structure during the observed boundary-layer separation and rotor formation event. Given the complex topography and the limited period of time of the observations, measuring turbulence from airborne in situ and radar data proves to be a challenging task. Some assumptions traditionally taken in turbulence studies (e.g., spatial homogeneity of turbulence or separation of scales) are clearly pushed to their limits, for several reasons. For example, the existence of a spectral gap in the spectral energy density of vertical velocity is hard to demonstrate. This is due to both the limited data available and the physical character of the phenomenon. The occurrence of BLS involves in fact a broad range of relevant scales, from mesoscale to microscale (mean wind - waves - wind pulsations - turbulence). Sophisticated methods to assess the location of the spectral gap, however, exist (e.g., multi-resolution flux decomposition) and are used to recover the correct turbulence averaging interval. From the available high-frequency radar data it is possible to measure turbulence intensity at multiple levels within the boundary layer. Conditions in the relatively unperturbed upstream boundary layer are contrasted with those in the very turbulent rotor region downstream of the mountain.