Simulation models and image data are simultaneously available for numerous scientific domains, such as oceanography and meteorology. They are indeed two complementary descriptions of the same complex system. Data Assimilation is a well-known mathematical technique used, in environmental sciences, to improve forecasts obtained from the simulation models, thanks to the observation data. One class of data assimilation algorithms, named 4D-Var, globally adjusts the model output to the observations, that are available over a period of time. The question of how to derive accurate characteristic features from images, with an optimal use of the simulation model, is of major interest for the image processing community. In this article, we consider applying data assimilation methods for motion estimation on a sequence of satellite images acquired over the ocean. We describe various strategies that can be derived in the framework of variational data assimilation (4D-Var). They mostly depend on the choice of the state vector itself. According to this definition, the dynamics has to be described and observation operators specified in order to characterize the information displayed by the image sequence. We detail the mathematical setting of these strategies and analyze their properties. Results are provided on twin experiments to quantify the methods and on satellite acquisitions acquired over the Black Sea.