The Antarctic ice sheet is increasingly contributing to sea-level rise because of accelerated mass losses at its floating extensions -- its ice shelves. By floating while remaining attached to the grounded ice sheet, ice shelves buttress (i.e. restrain) the inland ice in such a way that ice-shelf losses lead to accelerated ice discharge in the ocean. This thesis investigates the stability of Antarctic ice shelves -- so crucial for the stability of the entire ice sheet -- using the Roi Baudouin Ice Shelf (RBIS), Dronning Maud Land, East Antarctica, as a case study. The RBIS has remained relatively stable over the last millennia and presents various kilometre-scale features (pinning point, ice-shelf channels and englacial lakes) with potential impact on its present and future stability.We first derive a horizontal velocity field, combining interferometry and speckle tracking with Synthetic Aperture Radar images from ERS 1/2 and ALOS-PALSAR, respectively. The resulting velocities and associated shear-strain rates represent the most detailed fields, currently available for the RBIS and clearly resolve small-scale features of the RBIS: significant slow-down and shearing are observed upstream of a small pinning point and ice converges at ice-shelf channels. We then combine our flow field with high-resolution elevations from TanDEM-X to infer the Basal Mass Balance (BMB) of the RBIS. This method relies on mass conservation in a Lagrangian frame and enables us to finely detect spatial variability in the BMB. We show that the BMB of the RBIS varies substantially on sub-kilometre scales. Our technique is promising and could easily be applied more widely.Additionally, the flow field is used to investigate how considering/ignoring small pinning points in observations (geometry and velocities) impacts data initialisation of poorly known parameters (e.g. basal friction, ice viscosity) and subsequent ice-sheet modelling with BISICLES. We find that overlooking the pinning point in the bathymetry leads to erroneous ice-shelf properties whereas accurately capturing the pinning point in velocities is of secondary importance. Finally, before concluding the thesis, we discuss the stability of the RBIS and its neighbouring ice shelves. Most studies agree that the ice shelf has remained stable over the last decades to millennia and would likely remain so in the absence of external forcing. We however point out to three potential triggers of instabilities: (i) large quantities of surface meltwater are formed in the grounding zone and subsequently stored on the ice shelf, thereby providing fuel for hydrofracturing; (ii) ice-shelf channels are found to significantly incise the ice inland and (iii) a bathymetric trough beneath the RBIS forms a potential gateway for warm water intrusions the ice-shelf cavity, which could destabilise the ice shelf from below. We close with a short essay on the importance of outreach, where we argue that public engagement as a scientist should be considered as being part of science and should be valued for its worth. This chapter gives us the opportunity to present outreach activities undertaken in the frame of this thesis. We conclude that, just like ice shelves control Antarctic ice losses, science communication determines transfers of scientific expertise to public knowledge. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished