A new approach to projecting 21st century sea-level changes and extremes

Abstract

Future increases in flooding potential around the world’s coastlines from extreme sea level events is heavily dependent on projections of future global mean sea level (GMSL) rise. Yet, the two main approaches for projecting 21st century GMSL rise—i.e., process-based versus semi-empirical—give inconsistent results. Here, a novel hybrid approach to GMSL projection, containing a process-based thermosteric contribution and a semi-empirical ice-melt contribution, is embedded within a conceptual Earth system model (ESM). The ESM is run 10 million times with random perturbations to multiple parameters, and future projections are made only from the simulations that are historically consistent. The projections from our hybrid approach are found to be consistent with the dominant process-based GMSL projections from the Climate Model Intercomparison Project phase 5 (CMIP5) ensemble, in that our future ensemble-mean projections lie within ±2 cm of CMIP5 for the end of the 21st century when CMIP5-simulated histories are used to constrain our approach. However, when observations are used to provide the historic constraints for our hybrid approach, we find higher ice-melt sensitivity and additional ensemble-mean GMSL rise of around 13–16 cm by the end of the century. We assess the impact of this additional GMSL rise, projected from observation-consistency, on the increase in frequency of extreme sea level events for 220 coastal tide-gauge sites. Accounting for regional effects, we infer a 1.5–8 times increase in the frequency of extreme sea-level events for our higher GMSL projections relative to CMIP5.