Observational data from 61 moorings was used to decompose vertical velocity fields into its modal components using two different approaches, one assuming a flat bottom ocean and one assuming an ocean with rough topography. Also an analytical solution with idealized stratification was calculated. The modes were compared to the most dominant empirical orthogonal functions (EOFs). To be able to predict the vertical structure of the kinetic energy is of great relevance when interpreting surface velocity fields measured by altimeters. The results show that the most dominant EOF on average accounts for 75\% of the variance and resembles the shape of a first baroclinic steep bottom mode. We are able to predict the structure in the pycnocline layer fairly well for the mooring located in the mid-latitudes as long as they are not too close to the coast or in shallow areas. For the high latitudes the picture is more complicated. An analysis of the flat bottom mode is also carried out and gives results in agreement with previous studies finding the barotropic and first baroclinic modes to account for most of the column-averaged kinetic energy. The question remains if we should interpret the dominant EOF as a coupling between the barotropic and first baroclinic mode, or as the first baroclinic steep bottom mode. If the latter is true only one mode is needed for interpreting altimeter data and the dominant elements for the kinetic energy will be the first baroclinic mode, and a bottom trapped topographic wave which is independent and do not have to be correlated to the baroclinic modes.