Uncertainty quantification and validation of data assimilation methods for an ice sheet model : application to a Greenland marine-terminated glacier, Upernavik Isstrøm

• Main Author: Jager, Eliot
• Other Authors: Institut des Géosciences de l’Environnement (IGE), Institut de Recherche pour le Développement (IRD)-Institut national des sciences de l'Univers (INSU - CNRS)-Centre National de la Recherche Scientifique (CNRS)-Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement (INRAE)-Université Grenoble Alpes (UGA)-Institut polytechnique de Grenoble - Grenoble Institute of Technology (Grenoble INP ), Université Grenoble Alpes (UGA), Université Grenoble Alpes 2020-., Fabien Gillet-Chaulet, Nicolas Champollion, Jérémie Mouginot
• Format: Doctoral or Postdoctoral Thesis
• Language: French
• Published: HAL CCSD 2023
• Subjects: Ensemble method; Ice sheet dynamics; Modeling; Calibration; Uncertainty quantification; Validation; Dynamique glaciaire; Modélisation; Méthode d'ensemble; Quantification d'incertitudes; [PHYS.PHYS.PHYS-GEO-PH]Physics [physics]/Physics [physics]/Geophysics [physics.geo-ph]; glacier; Greenland; groenlandais; Ice Sheet; Tidewater; Upernavik
• Online Access: https://theses.hal.science/tel-04658964; https://theses.hal.science/tel-04658964/document; https://theses.hal.science/tel-04658964/file/JAGER_2023_archivage.pdf

This thesis addresses the significant challenge of estimating future sea-level rise driven by ongoing polar ice sheet mass loss due to current climate change. The uncertainties surrounding this projection are substantial, with potential contributions ranging from 0.5 to over 1.5 meters by 2100. These uncertainties primarily stem from limited observations regarding the current state of polar ice sheet and their interactions with the atmosphere and ocean, known as forcings. However, the growing availability of satellite data presents new opportunities. This thesis pursues two primary objectives regarding their utilization: firstly, evaluating the ability of polar ice sheet models to replicate this data over several decades, and secondly, exploring methods to reduce uncertainties related to future sea-level rise.To address these challenges and refine our methods, we focused on Upernavik Isstrøm, a tidewater glacier, employing an ice sheet model (Elmer/Ice). We developed an initialisation method that accounts for model uncertainties (parameters, initial geometry, forcings) through an ensemble approach. This method enables a model calibrating a friction parameter using inverse methods to reproduce past observations of velocity and elevations changes. To achieve this, the friction parameterisation of the model must consider the subglacial hydrology effect. When the glacier front retreats, it brings more subglacial water upstream, lubricating the contact with the bedrock. Failing to account for this effect prevents the model from replicating observed acceleration. In the next step, we propagated the ensemble of members obtained at the end of the simulation into the future. This involved incorporating future forcing uncertainties, such as greenhouse gas emission scenarios, climate model uncertainties, and oceanic forcing uncertainties. We also developed a calibration approach to give more weight to model members that best replicated past observations while ensuring the model did not overfit, to see if this refines ...