Mesoscale winds over the oceans

With the succesfull launch of MetOp-A carrying the Advanced Scatterometer (ASCAT) in October 2006, wind scatterometry entered its operational stage. Development of scatterometer processing software and products in Europe is organised through the EUMETSAT Satellite Application Facilities for Numerical Weather Prediction (NWP SAF) and for Ocean and Sea Ice (OSI SAF). Product developments in the NWP SAF are focussed on using the data for Numerical Weather Prediction and short-range weather forecasting. The former is well suited by the SeaWinds products as currently produced at the Royal Netherlands Meteorological Institute (KNMI) at 100-km resolution in preparation of the operational ASCAT production suite. For short-range forecasting or in semi-enclosed sea areas such as the Mediteranean, however, higher resolution is desirable, and therefore demonstration products at 25-km resolution are available from SeaWinds, ASCAT, and the scatterometer on board the Earth Remote Sensing satellite (ERS2). All products are currently provided in the BUFR format to facilitate the development of user interfaces. In the 2DVAR ambiguity removal method KNMI attempts to improve the spatial filtering properties of the wind retrieval by using prior information on the expected meteorological balance, e.g., favouring rotational structures in high-latitude regions. Moreover, solutions in all wind directions, weighted by their inherent probability, can be retained in the processing, the so-called Multiple Solution Scheme (MSS). The 2DVAR method has the advanced filtering properties needed for maintaining small-scale meteorological information in SeaWinds, while reducing noise. This is shown by comparing the autocorrelation of the scatterometer wind fields with that of model by the European Centre for Medium Range Weather Forecasting (ECMWF). It is shown that SeaWinds scatterometer wind fields at high resolution contain noise, but that this noise is effectively removed in the MSS. From an ASCAT example it is also shown that there is still room for improvement in the error model underlying the 2DVAR method. These findings will be helpful in the development of a 12.5-km ASCAT scatterometer wind product in the coastal zone.