We study in this paper a new data assimilation algorithm, called the Back and Forth Nudging (BFN). This scheme has been very recently introduced for simplicity reasons, as it does not require any linearization, or adjoint equation, or minimization process in comparison with variational schemes, but nevertheless it provides a new estimation of the initial condition at each iteration. We study its convergence properties as well as efficiency on a 2D shallow water model. All along the numerical experiments, comparisons with the standard variational algorithm (called 4D-VAR) are performed. Finally, a hybrid method is introduced, by considering a few iterations of the BFN algorithm as a preprocessing tool for the 4D-VAR algorithm. We show that the BFN algorithm is extremely powerful in the very first iterations, and also that the hybrid method can both improve notably the quality of the identified initial condition by the 4D-VAR scheme and reduce the number of iterations needed to achieve convergence.