Modeling the global and regional sea level variability in the last decades
Luciana Fenoglio-Marc(1) and Matthias Becker(1)
Technical University Darmstadt,
The study analyses the sea level variability from tide gauge and multi-mission satellite altimetry and its reconstruction in the last decades using the dominant patterns and the climatic indices.
The method involves performing a decomposition of the variability of each field and of their coupled variability by Principal Component Analysis and by Canonical Correlation Analysis and constructing the sea level using spatial patterns from altimetry and temporal patterns from tide gauge stations. The first dominant modes are chosen to be correlated with the climatic indices, the others are determined to best fit the observations.
We expect the corresponding spatial patterns to be persistent and constant in time, as linked to a physical phenomena. The stability and persistency of the spatial temporal patterns are investigated using sea surface temperature and thermo-steric sea level heights that, differently from the sea level data, are available globally over several decades. In this case the correlation of the dominant modes with the climatic indices is lower than when sea level heights are used.
Globally, in 1993-2005 the empirical model reproduce the interannual variability observed by altimetry in 1993-2005 with a root mean square error of about 2 centimeters and corresponds to about 50% of the interannual variability.
For the European region a regional empirical model is estimated using multi-mission altimeter data. Comparing steric interannual variability and observations in the Mediterranean Sea over the last 15-years a better agreement is found with altimetry than with some of the tide gauges.
The selection of the tide-gauge stations used is of relevance for the model construction.