Towards Direct Variational Assimilation Of Scatiterometer Backscatter Measurements Into Numerical Weather Prediction Models

As a first step towards variational assimilation of scatterometer backscatter measurements into NWF' models, 8 variational surface wind analysis scheme has been developped, whereby a field of unique surface wind vectors is directly retrieved from sigma naught triplets, through the minimization of a cost function involving observation error penalty and smoothness constraint terms. Reallistic ERSl simulations lead to promising wind retrieval performances as regards accuracy, implicit ambiguity removal, preservation of actual wind structures of dynamical significance, and cost efficiency. No sensitivity is found to the initial wind field, and smoothness constraints are shown to reduce the RMS error induced by instrumental noise on wind divergence and key ocean forcing parameters.