Data assimilation for initialisation and parameter estimation of an ice sheet evolution model

• Main Author: Bonan, Bertrand
• Other Authors: Laboratoire Jean Kuntzmann (LJK), Université Pierre Mendès France - Grenoble 2 (UPMF)-Université Joseph Fourier - Grenoble 1 (UJF)-Institut polytechnique de Grenoble - Grenoble Institute of Technology (Grenoble INP )-Centre National de la Recherche Scientifique (CNRS), Université de Grenoble, Maëlle Nodet, Catherine Ritz
• Format: Doctoral or Postdoctoral Thesis
• Language: French
• Published: HAL CCSD 2013
• Subjects: Data assimilation; Inverse methods; Parameter estimation; Ensemble Kalman Filter; Glaciology; Polar ice sheets; Assimilation de données; Méthodes inverses; Estimation de paramètres; Filtre de Kalman d\'ensemble; Glaciologie; Calottes polaires; [PHYS.PHYS.PHYS-GEO-PH]Physics [physics]/Physics [physics]/Geophysics [physics.geo-ph]; Ice Sheet
• Online Access: https://theses.hal.science/tel-00930097; https://theses.hal.science/tel-00930097v2/document; https://theses.hal.science/tel-00930097v2/file/BONAN_2013_archivage.pdf

Ice sheet evolution is both driven by a complex flow dynamics and by physical mechanisms such as basal sliding, ice temperature or surface mass balance. In addition to those, many feedback loops are observed between the different implicated phenomena. That explains how complex is to model this evolution. However several models have been developed in that purpose. These models depend on influential parameters, which often are unfortunately poorly known. So they need to be correctly specified. Data assimilation can give a better estimation of these parameters thanks to observations which are quite rare in glaciology. In this thesis, we work on the setup of efficient data assimilation systems for two inverses problems involving ice sheet evolution. We work with a simplified ice sheet evolution model called Winnie in order to focus on the setup. Nevertheless Winnie takes into account the major complex processes of ice dynamics and can be used for studies with different time scales. The first part of the thesis focuses on developing a 4D-Var approach in order to retrieve the evolution of a climatic parameter for a typical time scale of 20 000 years. This approach require the implementation the adjoint code of the evolution model. In a second part, we focus on the spin-up problem. This calibration problem for short term (maximum 100 years) simulations involve retrieving jointly the initial state, the bedrock topography and basal sliding parameters. In order to solve this problem we develop an Ensemble Kalman Filter approach. L'évolution des calottes polaires est régie à la fois par une dynamique d'écoulement complexe et par des mécanismes tel le glissement à la base, la température de la glace ou le bilan de masse en surface. De plus, de nombreuses boucles de rétroactions sont constatées entre les différents phénomènes impliquées. Tout ceci rend la modélisation de cette évolution complexe. Malgré tout, un certain nombre de modèles ont été développés dans cette optique. Ceux-ci font tous intervenir des paramètres ...