Effective visualization of large, multidimensional image data is a key instrument for data exploration in biological and medical applications. We propose a method to visualize motion patterns on a spherical surface, illustrated for the example of the zebrafish embryo. In particular, we analyze cellular motion by integrating prior knowledge about the underlying biological process and show how this improves visualization and connects to the conceptional description of the gastrulation process. We conclude that specialized visualizations are necessary to match data representation with conceptional ideas and to get a more intuitive understanding of the design principles of embryonic tissue formation.