An EOF and SVD analysis of interannual variability of GPS coordinates, environmental parameters and space gravity data

Eleven years, 1999–2010, of GPS positions of stations located in Europe and the Mediterranean were analyzed together with time series of atmospheric pressure, terrestrial water storage, and surface mass anomalies provided by the GRACE gravity mission. The Empirical Orthogonal Functions (EOF) technique was applied to single variables data series detrended and de-seasoned to identify the main modes of interannual variability in both time and space. For the GPS coordinates, the analysis has revealed the presence of significant interannual fluctuations which were also identified in the time series of the environmental parameters and of the space gravity data. A prominent feature common to the main modes of the time components of all variables is a slope inversion occurring around 2003–2004. During 2003 most of the European continent was affected by a severe drought. A second type of analysis was performed by applying the Singular Value Decomposition (SVD) approach to a few series of variable pairs to study the coupled space/time variability. This mathematical tool is used to examine possible cause-effect physical relationships between variables pairs. The analysis shows, for example, for the atmospheric pressure and the up coordinate that the first two SVDs account for about 97% of the total covariance. The clear anticorrelation observed between the two SVD1 spatial patterns represents the vertical crustal displacements associated with variations in atmospheric pressure. East–West (EW) and North–South (NS) gradients of the environmental parameters and of the GRACE surface mass anomalies were computed and compared to the EOFs of the east and north GPS coordinates respectively. Significant correlations were found, in particular, between the EW atmospheric pressure gradient and the east EOFs and between the GRACE EW and NS gradients and the east and north EOFs respectively. Significant correlations were also identified between the Multivariate ENSO Index (MEI) and the EOFs of GPS coordinates.