Postglacial sea-level change: novel insights from physical and statistical modelling

Abstract

Developing accurate projections of future sea-level change is a key challenge for the entire science community under the current warming climate. Due to the fact that modern instrumental sea-level observations are only available since the 19-20th century, sea-level projections based on them can only capture short-term effects, leaving physical processes that dominate over longer timescales underestimated. Therefore, an essential step towards accurate and robust long-term sea-level projections is to investigate the physical processes that impact the spatio-temporal evolution of sea-level change over centennial to millennial timescales. Due to sometimes scarce and often noisy palaeo sea-level observations, mechanisms of sea-level change over geological timescales are still not well-understood, with many outstanding questions to be resolved. This thesis develops novel physical and statistical models to better understand the mechanisms behind postglacial sea-level change. Specifically, this thesis focuses on three outstanding problems that are not only important in postglacial sea-level change but also in understanding past ice sheet dynamics and palaeoclimate change.

Firstly, a statistical framework is developed to invert the sources of meltwater pulse 1A, the largest and most rapid global sea-level rise event of the last deglaciation, with sophisticated treatment of uncertainties associated with sea-level reconstructions and geophysical modelling. The results suggest there were contributions from North America, 12.0 m (5.6-15.4 m; 95% probability), Scandinavia, 4.6 m (3.2-6.4 m), and Antarctica, 1.3 m (0-5.9 m), giving a total global mean sea-level rise of 17.9 m (15.7-20.2 m) in 500 years.

Secondly, the missing ice problem (distinctive imbalance between observed global mean sea-level rise and the reconstructed amount of ice-sheet melt) is revisited by including an extra physical process (sediment isostatic adjustment, SIA) which has not been considered in this problem before. In particular, this thesis investigates the impact of SIA on local RSL variation across the Great Barrier Reef (GBR), the world's largest mixed carbonate-siliciclastic sediment system. Based on a Bayesian calibration method, SIA can contribute up to 1.1 m relative sea-level rise in the outer shelf of the southern central GBR from 28 ka to present. Because the SIA-induced RSL rise is unrelated to ice mass loss, failing to correct for this signal will lead to systematic overestimation of grounded ice volume. Therefore, incorporating the SIA process will reduce the global grounded ice volume estimate for the Last Glacial Maximum (LGM), which can help to mitigate the missing ice problem.

Lastly, robust global barystatic sea-level maps with minimum dependency on the detailed geometry of past ice sheet change are reconstructed. Estimating such maps requires physical simulation of relative sea-level corresponding to thousands of different ice histories, which is computationally prohibitive. To improve this situation, this thesis develops a statistical emulator which can mimic the behaviour of a physics-based model and is computationally much cheaper to evaluate. The results highlight the Seychelles as an exceptionally good place to map barystatic sea level throughout the last deglaciation because RSL at this location only slightly departs from global barystatic sea level, with minor dependency on the assumed ice history.

Together, these physical and statistical models present powerful tools to yield novel insights into postglacial sea-level change mechanisms and hence they have the potential to yield more robust, accurate and trust-worthy sea-level change projections.