Eliot Jager

Basal friction is one of the most uncertain processes in ice‐sheet models used to project future sea level rise, yet friction parameters remain poorly constrained by observations. Here, we present the Greenland Ice Sheet Ensemble Friction Inversion (GrISEFI), a data set quantifying basal shear stress and its uncertainty at the Greenland scale using an ensemble approach to inverse methods. This ensemble accounts for uncertainties in ice rheology, friction exponents, regularization parameters, and velocity observations, generating 40 plausible friction states rather than a single deterministic solution. Our spatially resolved basal shear stress fields reveal systematic patterns controlled by bed geometry and surface mass balance: low values occur within ablation area for bed below sea level, while elevated values characterize regions upstream of the melt limit. Analysis of flow‐lines and cross‐sections of different Greenland Ice sheet glaciers (tidewater, and ice‐shelf glaciers) indicates that spatial variations in basal shear stress are primarily controlled by effective pressure changes, highlighting the importance of subglacial hydrology in friction parameterization. We show that effective pressure parameterizations based on distance from the terminus and melt onset better capture tidewater glacier behavior than commonly used height‐above‐flotation approaches. Constraining friction parameters through our ensemble, we find that basal shear stress exhibits substantially weaker velocity dependence than commonly assumed in ice‐sheet models, challenging widespread modeling assumptions. These findings have direct implications for how models represent the coupling between basal processes and ice dynamics in future sea level projections.

Exceeding the 1.5 °C Paris Agreement limit is now unavoidable, raising a critical question: how does the overshoot duration affect Antarctica's contribution to sea-level rise? Here we use an ice-sheet model to compare scenarios in which high-emissions forcing (SSP3-7.0) persists for 20, 50, 75 or 100 years before returning towards a low-emissions pathway (SSP1-2.6). When atmospheric and ocean forcing recover together, the additional contribution relative to sustained low forcing remains below 3 mm sea-level equivalent (SLE) by 2300 after 20-50 years of high forcing, but increases nonlinearly to 8.6 and 21.5 mm SLE after 75 and 100 years, respectively. When atmospheric and ocean forcing both recover after 50 years of high forcing, the additional contribution is only 2.3 mm SLE. However, this contribution increases to 44.1 mm SLE when ocean forcing is kept high for a further 50 years, in line with expectations of delayed ocean recovery, even while atmospheric forcing declines. Thus, shortening the overshoot duration limits Antarctica's additional contribution to sea-level rise, but the benefit depends strongly on how quickly ocean temperature declines.

Cette thèse aborde le défi majeur de l'estimation de l'élévation future du niveau moyen des mers à travers la quantification de la perte de masse des calottes polaires due au changement climatique en cours. Les incertitudes associées à ces estimations sont significatives, variant de 0.5 à plus de 1.5 mètre en 2100. Ces incertitudes résultent principalement du manque d'observations sur l'état actuel des calottes polaires et leurs interactions avec l'atmosphère et l'océan, appelées forçages. Cependant, la croissance du nombre de données satellites offre de nouvelles perspectives, et cette thèse vise deux objectifs clés pour leur utilisation : évaluer la capacité des modèles de calottes polaires à reproduire ces données sur plusieurs décennies, puis explorer des moyens d'utiliser ces données pour réduire les incertitudes liées à la future élévation du niveau de la mer.Pour répondre à nos objectifs, nous nous sommes intéressés à Upernavik Isstrøm, un glacier émissaire côtier, et simulé avec le modèle de calotte Elmer/Ice. Nous avons développé une méthode d'initialisation, qui permet de prendre en compte les incertitudes du modèle (paramètres, géométrie initiale, forçages) au travers de simulations d'ensembles et qui permet au modèle de calotte utilisé de reproduire les observations passées de changement de vitesse et d'élévation grâce à la calibration par méthode inverse de la friction (frottement). Pour cela, la paramétrisation de la friction du modèle doit prendre en compte l'effet de l'hydrologie sous-glaciaire qui, lorsque le front recule, amène plus d'eau sous-glaciaire en amont du glacier et lubrifie le contact de celui-ci au niveau du socle. Sans la prise en compte de cet effet, le modèle est incapable de reproduire l'augmentation de vitesse observée. Dans un deuxième temps, nous avons continué dans le futur les simulations de l'ensemble des membres, en ajoutant aux incertitudes du modèle celles liées au forçage futur : scénario d'émission de gaz à effet de serre, incertitudes des modèles de climat et incertitudes du forçage océanique. Nous avons également développé dans cette partie une méthode de calibration, pour donner plus de poids aux membres du modèle qui reproduisaient le mieux les observations passées, en vérifiant que cette calibration n'entraînait pas de sur-apprentissage ( extit{overfitting}), afin d'affiner les prédictions futures. Nous avons alors trouvé que la source d'incertitude la plus importante est le scénario d'émission, ce qui limite fortement la réduction d'incertitude permise par la calibration.Au final, nos résultats démontrent qu'Upernavik Isstrøm contribuera à l'élévation du niveau des mers à hauteur d'au moins 0.6 mm à l'avenir. Toutefois, cette contribution dépendra des émissions futures de gaz à effet de serre et pourrait atteindre jusqu'à 7.2 mm. De plus, la paramétrisation développée fait grimper ce chiffre, qui pourrait atteindre près de 9.1~mm, montrant son importance et la nécessité de la développer à l'échelle du Groenland, en parallèle de nos méthodes. Les fortes incertitudes liées au scénario d'émission montre le besoin de s'intéresser en profondeur aux futurs possibles des émissions de gaz à effet de serre pour mieux quantifier cette incertitude et la réduire.

Antarctica’s contribution to sea level rise depends critically on basal sliding—the friction between ice and bedrock—yet the physical processes governing this resistance remain poorly constrained. Building on the Greenland Ice Sheet Ensemble Friction Inversion (GrISEFI), we present the Antarctic Ice Sheet Ensemble Friction Inversion (AISEFI), quantifying basal shear stress and its uncertainty across Antarctica using 60 ensemble members that sample uncertainties in ice rheology, basal properties, and observations. Antarctic basal shear stress patterns require two distinct regimes controlled by effective pressure (the difference between ice overburden and subglacial water pressure) and basal velocity: a velocity-dependent Weertman regime and an effective pressure-limited Coulomb regime. Comparison with subglacial hydrology model output confirms that effective pressure variations control spatial patterns in Coulomb regime areas, although simplified parameterizations based on ice geometry can capture first-order patterns for most regions. Statistical analysis reveals that basal sliding exhibits weaker velocity dependence than commonly assumed in ice sheet models, with potential velocity-weakening behavior in the Coulomb regime, where basal shear stress decreases as velocity increases. Remarkably, these behaviors are consistent with those of Greenland despite vastly different characteristics, suggesting that the same two-regime physics reflects fundamental ice-bed processes rather than region-specific conditions. We synthesize the findings of both ice sheets to provide unified recommendations for basal sliding treatment in ice sheet models. The AISEFI and GrISEFI datasets together provide a framework for connecting model parameters to observable bed properties, advancing integration of glacial geology and ice sheet modeling for improved sea level projections.

Basal friction is one of the most uncertain processes in ice-sheet models used to project future sea level rise, yet friction parameters remain poorly constrained by observations. Here, we present the Greenland Ice Sheet Ensemble Friction Inversion (GrISEFI), a dataset quantifying basal shear stress and its uncertainty at the Greenland scale using an ensemble approach to inverse methods. This ensemble accounts for uncertainties in ice rheology, friction exponents, regularization parameters, and velocity observations, generating 40 plausible friction states rather than a single deterministic solution. Our spatially-resolved basal shear stress fields reveal systematic patterns controlled by bed geometry and surface mass balance: low values occur within ablation area for bed below sea level, while elevated values characterize regions upstream of the melt limit. Analysis of flow-lines and cross-sections of different Greenland Ice sheet glaciers (land-terminating, tidewater, and ice-shelf glaciers) demonstrates that spatial variations in basal shear stress are primarily controlled by effective pressure changes, validating the importance of subglacial hydrology in friction parameterization. We show that effective pressure parameterizations based on distance from the terminus and melt onset better capture tidewater glacier behavior than commonly-used height-above-flotation approaches. Constraining friction parameters through our ensemble, we find that basal shear stress exhibits substantially weaker velocity dependence than commonly assumed in ice-sheet models, challenging widespread modeling assumptions. These findings have direct implications for how models represent the coupling between basal processes and ice dynamics in future sea level projections.

Antarctica’s contribution to global sea-level rise is deeply uncertain, with subglacial water suspected to play a critical role, yet its impact remains unclear. We demonstrate that water at the base of ice sheets influences sliding behaviour and that its exclusion from models can underestimate sea-level rise projections and delay the predicted onset of tipping points. Here we use an Antarctic Ice Sheet model (Elmer/Ice) to explore how different assumptions about water pressure at the ice base affect sea-level rise projections from 2015 to 2300. Our results indicate that incorporating subglacial water can amplify ice discharge across the Antarctic Ice Sheet by up to threefold above the standard approach, potentially contributing an additional 2.2 metres to sea-level rise by 2300. Notably, a smoothly decreasing basal drag near the grounding line more than doubles grounding line flux by 2300 relative to scenarios where effective pressure is simplified into a spatially constant coefficient. Basin-specific responses vary significantly, with some scenarios advancing tipping points by up to 40 years. These findings underscore the critical need to integrate evolving subglacial hydrology into ice sheet models.

The Antarctic and Greenland Ice Sheets (AIS and GrIS) play a critical role in shaping future sea-level rise (SLR), but their contributions and human greenhouse gas emissions remain the largest sources of uncertainty in SLR projections (Edwards et al., 2021). This uncertainty, along with the risk of potential tipping points leading to rapid ice loss, arises from a limited understanding of key processes that govern ice-sheet behaviour (Fox-Kemper et al., 2021). In the Ice Sheet Model Intercomparison for CMIP6 (ISMIP6), 55% of AIS and 15 to 35% of GrIS future mass loss uncertainty is due to ice-sheet models’ uncertainty (Seroussi et al., 2023; Jager et al., 2024; Goelzer et al., 2020). While modeling the AIS’s and GrIS’s complex interactions with the climate is difficult (Seroussi et al., 2023), one other uncertain process is basal sliding over bedrock. Because this process is not directly observed due to the large thickness, its representation in current models remains rudimentary. The primary goals of Combining Coupled Modelling and Machine Learning to Constrain Antarctica’s Uncertain Future (ICEMAP) project include (i) better quantification of uncertainties related to basal sliding, (ii) comparison of these uncertainties to other sources of uncertainty, and (iii) exploration of how satellite data can help reduce these uncertainties. To achieve these goals, we employ the Shallow Shelf Approximation (SSA) implemented in the Elmer/Ice model, an open source finite element software for ice sheets, glaciers and ice flow modelling, which was one of the participants in ISMIP6 (Gagliardini et al., 2013; Seroussi et al., 2023). It uses inverse methods to calibrate the many unknown parameters related to rheology and friction. Here, we take into account various physical and numerical uncertainties to perform multiple calibrations using remote-sensing velocity data to compute basal sliding and basal shear stress. Subsequently, these values and their spatial variations can be compared with the diverse existing theories that have been developed from small-scale physical and numerical experiments (Gagliardini et al., 2007; Zoet and Iverson, 2020).Our analysis demonstrates that the principal characteristics of these parameterisations, derived from small-scale experiments, are observable at a large scale. However, the values may deviate from expected norms, particularly with regard to the exponent of the Weertman friction law. This investigation enables the quantification of both the parameter values and their associated uncertainties within the friction parameterisations governing the AIS and the GrIS. Furthermore, it highlights the critical influence of basal water presence, which appears to play a pivotal role in the variability of basal sliding. Incorporating this factor into models, whether through varying levels of model complexity or the use of proxies, is deemed essential for accurately capturing the temporal variations in basal sliding.

This study investigates the uncertain future contribution to sea-level rise in response to global warming of Upernavik Isstrøm, a tidewater glacier in Greenland. We analyse multiple sources of uncertainty, including Shared Socioeconomic Pathways (SSPs), climate models (global and regional), ice–ocean interactions, and ice sheet model (ISM) parameters. We use weighting methods based on spatio-temporal velocity and elevation data to reduce ice flow model uncertainty and evaluate their ability to prevent overconfidence. Our developed initialization method demonstrates the capability of Elmer/Ice to accurately replicate the hindcast mass loss of Upernavik Isstrøm. Future mass loss predictions in 2100 range from a contribution to sea-level rise from 1.5 to 7.2 mm, with an already committed sea-level contribution projection from 0.6 to 1.3 mm. At the end of the century, SSP-related uncertainty constitutes the predominant component of total uncertainty, accounting for 40 %, while uncertainty linked to the ISM represents 15 % of the overall uncertainty. We find that calibration does not reduce uncertainty in the future mass loss between today and 2100 (+2 %) but significantly reduces uncertainty in the hindcast mass loss between 1985 and 2015 (−32 % to −61 % depending on the weighting method). Combining calibration of the ice sheet model with SSP weighting yields uncertainty reductions in future mass loss in 2050 (−1.5 %) and in 2100 (−32 %).

This study investigates the uncertain future contributions to sea-level rise in response to global warming of Upernavik Isstrøm, a tidewater glacier in Greenland. We analyze multiple sources of uncertainty, including shared socio-economic pathways (SSPs), climate models (global and regional), ice-ocean interactions, and ice sheet model parameters (ISM). We use weighting methods based on spatio-temporal velocity and elevation data to reduce ice flow model uncertainty, and evaluate their ability to prevent overconfidence. Our developed initialization method demonstrates the capability of Elmer/Ice to accurately replicate the historical mass loss of Upernavik Isstrøm. This provides confidence in the model's ability to project the future evolution of this region. Future mass loss predictions range from a contribution to sea level rise from 1.5 to 7.2 mm, with an already committed sea-level contribution projection from 0.6 to 1.3 mm. While all sources of uncertainty contribute at least 15 % to uncertainty until the end of the century, SSP-related uncertainty dominates at 40 %. We find that calibration does not reduce uncertainty of the future mass loss between today and 2100 of Upernavik Isstrøm (+2 %) but significantly reduces uncertainty in the historical mass loss of Upernavik Isstrøm between 1985 and 2015 (-32 to -61 % depending on the weighting method). Combining calibration of the ice sheet model with SSP weighting yields uncertainty reductions of future mass loss in 2050 (-1.5 %) and in 2100 (-32 %).

This study investigates the uncertain future contributions to sea-level rise in response to global warming of Upernavik Isstrøm, a tidewater glacier in Greenland. We analyze multiple sources of uncertainty, including shared socio-economic pathways (SSPs), climate models (global and regional), ice-ocean interactions, and ice sheet model parameters (ISM). We use weighting methods based on spatio-temporal velocity and elevation data to reduce ice flow model uncertainty, and evaluate their ability to prevent overconfidence. Our developed initialization method demonstrates the capability of Elmer/Ice to accurately replicate the historical mass loss of Upernavik Isstrøm. This provides confidence in the model's ability to project the future evolution of this region. Future mass loss predictions range from a contribution to sea level rise from 1.5 to 7.2 mm, with an already committed sea-level contribution projection from 0.6 to 1.3 mm. While all sources of uncertainty contribute at least 15 % to uncertainty until the end of the century, SSP-related uncertainty dominates at 40 %. We find that calibration does not reduce uncertainty of the future mass loss between today and 2100 of Upernavik Isstrøm (+2 %) but significantly reduces uncertainty in the historical mass loss of Upernavik Isstrøm between 1985 and 2015 (-32 to -61 % depending on the weighting method). Combining calibration of the ice sheet model with SSP weighting yields uncertainty reductions of future mass loss in 2050 (-1.5 %) and in 2100 (-32 %).