Coupled sea ice - ocean models used for Arctic- or Antarctic-wide simulations of sea ice kinematics and dynamics have usually been operated at spatial resolutions on the order of 100 km. Consequently, the validation of such models has been carried out using data from spaceborne passive microwave radiometers or scatterometers. Recently, a number of models work on finer regular grids or on grids with flexible spatial resolution which is high adjacent to the coast or in areas with spatially highly variable ocean conditions. The Finite Element Sea Ice Ocean Model (FESOM) is based on the latter approach. Areas of specific interest are e. g. coastal polynias since the continuous production of new ice and the resulting release of brine as well as the increased heat flux affect both ocean and atmosphere on local or even regional scales. In a number of studies the evolution of polynias in dependence of environmental conditions was investigated using local flux balance models which describe the balance between the advection of sea ice from the coast and the production rate of new ice. These models neglect interactions with the ocean and internal forces in the ice. In our study we focussed on the simulation of the evolution of coastal polynias using FESOM with three different atmospheric forcing data sets that differ in spatial and temporal resolution. We compared the variation of polynia extent and drift patterns obtained from the model simulations with two time series of Envisat ASAR wide swath images that were acquired in austral summer 2008 (February) and late fall 2008 (June). Our test site was the Ronne Polynia in the Weddell Sea, Antarctica. For the evolution of the polynia extent we found that simulations of the employed sea ice ocean model agree with the observations. The comparison of drift velocities showed that the modeled ice velocities were too small in many cases, and turning angles were slightly too large relative to the data obtained from the radar imagery. This was attributed to the forcing data, in which the wind velocities may have been too low. In summary our results demonstrate that the evolution of polynias can be realistically simulated with coupled sea ice models such as FESOM, provided that the modelling grid is dense (1-3 km) and the atmospheric forcing data are provided at high spatial resolution (less than 50 km). Furthermore the study demonstrates the usefulness of high-resolution satellite radar imagery for gathering validation data on regional and local scales.