A spectral nudging technique for dynamical downscaling purposes

The "spectral nudging" method imposes time-variable large-scale atmospheric states on a regional atmospheric model. It is based on the idea that regio nal-scale climate statistics are conditioned by the interplay between conti nental-scale atmospheric conditions and such regional features as marginal seas and mountain ranges. Following this "downscaling" idea, the regional m odel is forced to satisfy not only boundary conditions, possibly in a bound ary sponge region, but also large-scale flow conditions inside the integrat ion area. In the present paper the performance of spectral nudging in an extended cli mate simulation is examined. Its success in keeping the simulated state clo se to the driving state at larger scales, while generating smaller-scale fe atures is demonstrated, and it is also shown that the standard boundary for cing technique in current use allows the regional model to develop internal states conflicting with the large-scale state. It is concluded that spectr al nudging may be seen as a suboptimal and indirect data assimilation techn ique.