Objective analysis of temperature and salinity historical data set over the Mediterranean basin

The hydrological data set of the MEDATLAS project, is used to compute temperature (T) and salinity (S) climatological fields over the Mediterranean sea. The method used to perform objective analysis of the historical data set is optimal interpolation by kriging. Its implementation to compute basinscale climatology led the authors to address specific issues on data preparation, neighborhood search procedure and optimization of estimation points location. Consequently, estimation is achieved suboptimaly, due to large amount of data, and with a local computation of the spatial variogram to account for the changes in spatial scales over the domain. No temporal variogram is adopted but estimation is performed with observations each acquired by a different year to avoid bias by synoptic data subset, to retrieve long term component and to reach requested temporal representativeness. Results show that the MEDATLAS project succeeded in building up a comprehensive, quite homogeneous and very clean data set (0.65% rejection rate during the cross validation). With this data set a monthly, 40-km resolution climatology with 10-year representativeness could be computed. A first validation, by comparing with other recent climatology, shows that a precision within 0.2/spl deg/C rms is achieved.