Résumé : Antarctic ice shelves control the ice flux from the continent to the ocean. As such, they play a major role in the stability of the ice sheet and its potential contribution to sea level rise, especially in the context of global change. Below some of these ice shelves, marine ice can be found which is a product of the Deep Thermohaline Circulation. Due to its specific genetic process, marine ice has intrinsic physical (grain size, ice fabric, bubble content, ....) and chemical (impurities, water stable isotopes) properties, that differ from those of 'meteoric ice' formed on the continent through snow metamorphism or 'sea ice' resulting from sea water freezing at the ocean-atmosphere surface. Until now however, the effect of these specific properties on marine ice rheology is still very poorly understood.

The principal objective being to include realistic mechanical parameters for marine ice in ice shelf flow models, uniaxial compression experiments have been performed on various types of marine ice samples. Technical developments are an important component of this thesis has they were necessary to equip the laboratory with the appropriate tools (pneumatic rig, automatic ice fabric data handling).

Results from experimental compression on isotropic marine ice show that it represents the higher boundary for meteoric ice viscosity throughout the whole temperature range, thereby validating Cuffey and Paterson's relationship with an enhancement factor equals to 1.

Marine ice is however often quite anisotropic, showing elongated crystals and wide single maximum fabric, that should impact its mechanical properties. Experiments on pre-oriented marine ice samples have therefore been carried out combining the study of epsilon_{oct} vs. tau_{oct} with a thorough analysis of microstructural data 'before' and 'after' the experiment.

Depending on the orientation of the sample in the applied stress field and on the intensity of the latter, anisotropic marine ice can be harder or softer than its isotropic counterpart, with n=4 often observed in Glen's flow law. Associating the experimental geometrical settings to potential natural equivalent, results suggest that anisotropic marine ice would strengthen ice shelf flow in most areas (for a same given temperature), apart from suturing areas between individual ice streams as they merge to form the ice shelf, where it could become weaker than meteoric ice in certain circumstances.

Finally, preliminary sensitivity studies, using a simple ice shelf model with our experimental parameters of Glen's flow law have allowed us to discuss the potential impact of rift location, rift size and thermal regime in the ice shelf behavior.