Uncertainty in Sea Level Rise Projections Due to the Dependence Between Contributors

11 Using two process-based models to project sea level for the 21st century, it is shown that 12 taking into account the correlation between sea level contributors is important to bet13 ter quantify the uncertainty of future sea level. In these models the correlation primar14 ily arises from global mean surface temperature that simultaneously leads to more or less 15 ice melt and thermal expansion. Assuming that sea level contributors are independent 16 of each other underestimates the uncertainty in sea level projections. As a result, high17 end low probability events that are important for decision making are underestimated. 18 For a probabilistic model it is shown that the 95th percentile of the total sea level rise 19 distribution at the end of the 21st century is underestimated by 5 cm for the RCP4.5 20 scenario under the independent assumption. This underestimation is up to 16 cm for the 21 99.9th percentile of the RCP8.5 scenario. On the other hand, assuming perfect corre22 lation overestimates the uncertainty. The strength of the dependence between contrib23 utors is difficult to constrain from observations so its uncertainty is also explored. New 24 dependence relation between the uncertainty of dynamical processes and surface mass 25 balance in glaciers and ice caps and in the Antarctic and Greenland ice sheets are in26 troduced in our model. Total sea level uncertainty is found to be as sensitive to the de27 pendence between contributors as to uncertainty in individual contributors like thermal 28 expansion and Greenland ice sheet. 29