Level sets of implicit functions are used for segmentation and tracking of deformable structures encountered in remote sensed imagery. Iso-contours are proposed as a basic tool to model motion of natural structures undergoing viscoelastic deformation and topological change. In this study we present an energy minimization process for the segmentation of deformable structures. The energy functional carries information about position, tangency and curvature. By using implicit functions, it is possible to satisfy in a simple way complex geometrical requirements about motion which would be rather difficult to ensure using a parametrized interpolating surface. The method is applied over a temporal sequence of satellite images displaying a vortex motion.