Predictability of twentieth century sea-level rise from past data

Klaus Bitterman, Stefan Rahmstorf, Mahé Perrette, Martin Vermeer

Research output: Contribution to journalArticleScientificpeer-review

30 Citations (Scopus)
143 Downloads (Pure)


The prediction of global sea-level rise is one of the major challenges of climate science. While process-based models are still being improved to capture the complexity of the processes involved, semi-empirical models, exploiting the observed connection between global-mean sea level and global temperature and calibrated with data, have been developed as a complementary approach. Here we investigate whether twentieth century sea-level rise could have been predicted with such models given a knowledge of twentieth century global temperature increase. We find that either proxy or early tide gauge data do not hold enough information to constrain the model parameters well. However, in combination, the use of proxy and tide gauge sea-level data up to 1900 AD allows a good prediction of twentieth century sea-level rise, despite this rise being well outside the rates experienced in previous centuries during the calibration period of the model. The 90% confidence range for the linear twentieth century rise predicted by the semi-empirical model is 13–30 cm, whereas the observed interval (using two tide gauge data sets) is 14–26 cm.
Original languageEnglish
Article number014013
Pages (from-to)1-8
JournalEnvironmental Research Letters
Issue number1
Publication statusPublished - 2013
MoE publication typeA1 Journal article-refereed


  • climate change
  • model validation
  • projections
  • sea-level rise

Fingerprint Dive into the research topics of 'Predictability of twentieth century sea-level rise from past data'. Together they form a unique fingerprint.

Cite this