Marta Marcos (IMEDEA, Spain); Chris Piecuch (Atmospheric and Environmental Research, USA); Gabriel Jorda (IMEDEA, Spain); Jürgen Jensen (University of Siegen, Germany)

Here we examine the non-linear evolution of regional mean sea level (RMSL) over the 20th century using tide gauge records corrected for the effects of vertical land motion, atmospheric pressure and barotropic wind forcing. To do so, we apply a Singular Spectrum Analysis (SSA), where the uncertainties are assessed in a Monte-Carlo framework under the consideration of the temporal correlation structure of the tide gauge records. One of the main issues is to find an “optimal” smoothing window, which determines a nonlinear trend unrelated to natural processes. We hypothesize that this optimal smoothing window is reached, if the noise structure (determined by the Hurst coefficient or the lag1 autocorrelation) of the residuals (after the removal of the non-linear trend) does not significantly diverge from the noise structure in the initial records. Our analysis demonstrates that this condition is fulfilled for all considered records, if the smoothing window has a length of at least ~25 years. Therefore, we apply the SSA with a smoothing window of 25 years to all tide gauge records and map changes in rates over the entire 20th centuries. This provides the unique possibility of evaluating the spatio-temporal features of changes in the rates of MSL rise. Possible sources of periods of acceleration and deceleration will be discussed with respect of the relative role of atmospheric forcing, the oceans dynamic contribution and ice melt fingerprints.