The supersonic separated flowfield aft of a blunt-faced cylinder aligned with the freestream is highly complex. Therefore, new methods are needed to study this flowfield. Recently, empirical mode decomposition has emerged in the fluid dynamics community as a way to separate spatial and temporal scales in a signal. In this study, the multi-dimensional extension of ensemble empirical mode decomposition is used on the velocity fields obtained with particle image velocimetry of said flowfield. Multi-dimensional ensemble empirical mode decomposition was performed on both three-component planar and volumetric data sets. The analysis resulted in the scale separation of the different structures within the flowfield and revealed horseshoe-shaped and streamwise-oriented structures. This paper also presents the idea of eliminating small-scale features and noise from data sets by subtracting the corresponding small-scale components of multi-dimensional ensemble empirical mode decomposition from the original image.