Reconstructing with numerical Ice Sheet Models the post-LGM decay of the Eurasian Ice Sheets: data-model comparison and focus on the Storfjorden (Svalbard) ice stream dynamics history

The challenge of reconstructing the growth and decay of palaeo-ice sheets represents a critical task to better understand mechanisms of global climate change and associated sea-level fluctuations in the present and future. The Eurasian palaeo-Ice Sheet (EIS) at its maximum volume and extent (between 21,000 and 20,000 years ago) comprised the Scandinavian (SIS), Svalbard/Barents/Kara seas (SBKSIS), and British-Irish (BIIS) palaeo-ice sheets. The SBKSIS was a so called marine based ice sheet, as it rested several hundreds of meters below the sea level in the Barents and Kara seas. From a bathymetric and topographic point of view, there is a strong similarity between the SBKSIS and the present-day West Antarctic Ice Sheet (WAIS). Consequently, to reconstruct the dynamic processes driving the last deglaciation of the SBKSIS may represent an important task to better understand the WAIS present and future evolution. In this study, the evolution of the EIS during the last deglaciation is reconstructed with two hybrid Shallow-Ice/Shallow-Shelf-Approximation numerical Ice Sheet Models (ISMs), namely PSU and GRISLI. These two ISMs differ mainly in the ice stream parametrization and in the complexity with which grounding line migration is treated. A particular focus in this study is given to the Storfjorden glacial system in the north-western Barents Sea. In fact, several palaeo-data from this area furnish insights on the Storfjorden ice stream dynamics history, providing a good testing ground for ISMs. The ISMs are forced with macro-regional indexes representative of the climate evolution over Siberia/Kara Sea, Svalbard/Barents Sea and Fennoscandia during the last deglaciation. The climate in- dexes are based on TraCE-21ka, a transient climate simulation of the last 21,000 years carried out with the Atmosphere-Ocean General Circulation Model CCSM3. Two different ocean basal melting parametrizations based on ice-ocean heat fluxes are tested. The ocean basal melting parametrizations are forced with time-varying ocean temperature and salinity from TraCE-21ka transient climate simulation. Ocean temperature and salinity representative of Arctic Ocean, Norwegian Sea, north-western and south- western Barents Sea are employed. In order to deal with ISMs poorly constrained model parameters, a statistical approach is adopted. A Latin Hypercube Sampling (LHS) of five GRISLI parameters is performed. Due to large computational costs it is not possible to perform such a statistical approach with PSU ISM. Therefore, a restricted number of simulations performed with PSU ISM employ model parameter values from GRISLI significant simulations. GRISLI-simulated ice streams dynamics has a strong control on the deglaciation of the EIS, in particular in the Southern SIS and in the Western SBKSIS. In terms of total ice volume there is a good agreement between GRISLI simulations, ICE-5G reconstruction and global Eustatic Sea Level data. In contrast, GRISLI-simulated ice sheet extent evolution presents discrepancies with geological observation in the Southern SIS and in the Eastern SBKSIS. The use of different basal melting parametrizations in GRISLI has a strong impact both on the deglaciation of the SBKSIS and on the retreat of Storfjorden ice stream. The Storfjorden ice stream dynamics history in GRISLI simulations is in good agreement with palaeo-data in terms of timing of the ice stream retreat, Grounding Zone Wedges formation and response to Meltwater Pulse 1A. The different treatment in PSU of ice streams and Grounding Line migration has a remarkable effect on the EIS deglaciation. In particular, in PSU the Grounding Line dynamics plays a primary role with respect to ocean basal melting, thus exerting a strong control on the decay of the marine-based SBKSIS. Finally, an inverse-type approach is adopted to match PSU simulations with ICE-5G reconstruction and the other palaeo-data.