Improved Glacial Isostatic Adjustment Models for Eastern North America: Implications for Interpreting Vertical Land Motion and Projecting Sea-Level Change

Abstract

Much of the Atlantic coast of North America has been sinking for thousands of years, with maximum rates reaching ~20 cm per century, due to solid earth deformation in response to the deglaciation of the Laurentide Ice Sheet following the Last Glacial Maximum, between approximately 18,000 and 7,000 years ago. This process, known as glacial isostatic adjustment (GIA), is a dominant driver of vertical land motion (VLM) and sea-level change in the region. The research presented in this thesis aims to improve the quantification and understanding of the processes contributing to contemporary and future VLM and gravity change in eastern North America, in order to enhance projections of mean sea-level change and nuisance flooding frequency over the coming decades. A key aspect is to determine optimal GIA model parameters that are consistent with observations of past sea-level change and to better constrain their associated uncertainties, thereby improving the interpretation of long-term gravity change and VLM and refining the GIA component of sea-level projections. In Chapter 2, we constrain the GIA signal and its uncertainty for southeastern Canada and the northeastern United States using 1,013 paleo relative sea-level (RSL) data points from 38 sites. These data, comprising 544 sea-level index points, 232 marine limiting data points, and 237 terrestrial limiting data points, were compared against output from 14,960 simulations that combine a 1D, spherically symmetric earth model with 34 different North American ice sheet reconstructions. To account for laterally variable earth structure, the 1D GIA modeling results suggest that the region is best represented by two subregions that yield significantly improved fits to the RSL data: (1) HUL, near the center and thickest part of the former Laurentide Ice Sheet, encompassing Hudson Bay, the Ungava Peninsula, and Labrador; and (2) NSNM, located along the former ice margin, including Newfoundland, the St. Lawrence Corridor, New Brunswick and Nova Scotia, and Maine and Massachusetts. This subregional partitioning, inferred from the 1D analysis, was further tested using a 3D finite-volume earth model incorporating lateral viscosity variations based on two shear-wave tomographic models. A central component of this research is the estimation of GIA model uncertainty. Three complementary approaches, including heuristic, nominal Bayesian, and history matching, were used to quantify this uncertainty. Building on the results of Chapter 2, Chapter 3 delivers the second component of this thesis by isolating and interpreting the processes contributing to contemporary VLM across eastern Canada and the northeastern United States. This chapter evaluates VLM over two timescales, millennial and decadal, across six subregions in the study region. The analysis integrates GNSS-derived VLM rates from 88 stations, late-Holocene RSL reconstructions from 18 sites spanning the past 4,000 years (348 sea-level index points), outputs from regionally constrained GIA models, and GRACE-derived mass change estimates from April 2002 to November 2023. By comparing decadal-scale and millennial-scale VLM estimates, this methodology enables the identification of departures from long-term GIA-driven trends and the interpretation of these deviations in terms of non-GIA processes. The results underscore the dominant role of GIA while also revealing regional contributions from contemporary non-GIA signals, primarily driven by hydrological mass changes. This chapter presents an uncertainty analysis of the VLM signal and GRACE-derived total water storage variations. Extending the analyses from Chapters 2 and 3, Chapter 4 represents the third original component of this thesis, which aims to evaluate the influence of VLM, particularly its GIA-induced signal, on regional projections of mean sea-level change and nuisance flooding frequency across eastern Canada and the northeastern United States. This analysis focuses on 25 tide gauge stations across the region and provides projections for the years 2050, 2100, and 2150 CE. Projections of regional mean sea-level changes are developed under three Shared Socioeconomic Pathway (SSP) scenarios, including SSP1-1.9, SSP3-7.0, and SSP5-8.5, and evaluated across three distinct projection cases: (1) full Intergovernmental Panel on Climate Change (IPCC) projections including all components of sea-level change, (2) IPCC projections excluding the VLM signal, and (3) IPCC projections excluding VLM but incorporating regionally optimized GIA-induced VLM rates with associated model uncertainty bounds. Projections of nuisance flooding events are based on harmonic analysis that accounts for local nodal corrections and secular changes in major tidal constituents, while also incorporating projected regional mean sea-level changes under each SSP scenario and projection case. The results generally indicate that incorporating VLM (primarily its GIA signal) substantially modifies projections of both regional mean sea-level change and nuisance flooding frequency, underscoring the critical importance of accounting for the VLM signal in long-term coastal risk assessments in the study region.