Abstract. In this study we focus on the sea level trend pattern observed by satellite altimetry in the tropical Pacific over the 1993–2009 time span (i.e. 17 yr). Our objective is to investigate whether this 17-yr-long trend pattern was different before the altimetry era, what was its spatio-temporal variability and what have been its main drivers. We try to discriminate the respective roles of the internal variability of the climate system and of external forcing factors, in particular anthropogenic emissions (greenhouse gases and aerosols). On the basis of a 2-D past sea level reconstruction over 1950–2009 (based on a combination of observations and ocean modelling) and multi-century control runs (i.e. with constant, preindustrial external forcing) from eight coupled climate models, we have investigated how the observed 17-yr sea level trend pattern evolved during the last decades and centuries, and try to estimate the characteristic time scales of its variability. For that purpose, we have computed sea level trend patterns over successive 17-yr windows (i.e. the length of the altimetry record), both for the 60-yr long reconstructed sea level and the model runs. We find that the 2-D sea level reconstruction shows spatial trend patterns similar to the one observed during the altimetry era. The pattern appears to have fluctuated with time with a characteristic time scale of the order of 25–30 yr. The same behaviour is found in multi-centennial control runs of the coupled climate models. A similar analysis is performed with 20th century coupled climate model runs with complete external forcing (i.e. solar plus volcanic variability and changes in anthropogenic forcing). Results suggest that in the tropical Pacific, sea level trend fluctuations are dominated by the internal variability of the ocean–atmosphere coupled system. While our analysis cannot rule out any influence of anthropogenic forcing, it concludes that the latter effect in that particular region is stillhardly detectable.
Validation and quantification of uncertainty in coupled climate models using network analysis