Is the Mediterranean Sea surface height variability predictable?

Suselj, Kay, Tsimplis, Michael N. and Bergant, Klemen (2008) Is the Mediterranean Sea surface height variability predictable? Physics and Chemistry of the Earth, Parts A, B and C, 33 (3-4), 225-238. (doi:10.1016/j.pce.2006.12.001).

Abstract

The relation of sea surface height (SSH) in the Mediterranean Sea with numerous atmospheric variables on time scales from month to few years were investigated in order to construct simple linear statistical models for predicting SSH. Monthly SSH data for the period from January 1993 to December 2001 from the Topex/Poseidon altimetry are used as the dependent variable in models. Numerous large scale atmospheric indices (LSI) as well as gridded fields of various meteorological variables above the Mediterranean region are used as predictors. The annual cycle derived from nine year SSH climatology explains 40% of SSH variance. The statistical models were built on the anomalies from the annual cycle, separated to the following seasons: winter (December–March), spring (April–May), summer (June–September) and autumn (October–November). The results were tested with cross-validation where the quality of the models were judged on the basis of the explained variance of SSH. Among all seasons, SSH was the best predicted in the winter. The most successful LSI, Mediterranean Oscillation Index, is able to explain 46% of winter variance of SSH anomaly from annual cycle (SSHA). The best model, which uses sea level pressure (SLP), northward and eastward components of wind speed at 10 meters above sea level, is able to explain on average 55% of SSHA variance with the values up to 80% in Ionian, Tyrrhenian and eastern part of Algero-Provençal Sea. SSH is the worse predictable in the summer, when the root mean square of SSH is not well above the threshold of its measurements. The SLP and temperature at 2 meters in the atmosphere (T2M) as combined predictors are able to explain 20% of SSHA variance, but only in the Adriatic and Tyrrhenian Seas noticeable part of SSHA could be explained. In spring explained variance of SSHA reaches 38% (when SLP and T2M are used as predictors), while in autumn 28% of SSHA can be explained with SLP as a predictor.

This record has no associated files available for download.

Contributors

Download statistics