Conde-Criado, Justo (Servicio de analisis e investigacion del clima. Instituto Nacional de Meteorologia, Spain)
A statistical technique to derive climatological fields on regional scale from the synoptic ones has been designed and developed. This technique has been used to relate winter mean sea level pressure anomalies over the Atlantic area to winter precipitation volume anomalies over the Mediterranean hydrographic basins. The main idea is to identify characteristic patterns of observed simultaneous variations of both variables.
To do this, a canonical correlation analysis (CCA) is carried out to the above mentioned sets of data. The CCA is performed to no original data. Prior correlation analysis a principal component analysis is made to both datasets to a) reduce the amount of variables, b) to avoid possible singularities in the covariance matrices and, finally, c) to distinguish large physical signals from those ones produced from a spatially and temporally uncorrelated random process.