The SIDE7 session, held on 10 March during the CMIP 2026 Workshop, focused on developing projections of ice sheet mass loss for global sea level rise within the CMIP7 framework. A central part of the discussion was the Ice Sheet Model Intercomparison Project for CMIP7 (ISMIP7), which aims to use climate output from CMIP simulations to generate boundary conditions for ice sheet models. In practice, ISMIP predominantly uses standard standalone ice sheet models forced by CMIP climate output, whereas fully coupled Earth system models with interactive ice sheets are under active development.
Ice sheets are a major component of the climate system and a key contributor to future sea level rise, yet they remain inadequately represented in current climate models. Projections of ice-sheet contributions to sea level rise remain highly uncertain. The session placed particular emphasis on the sources of this uncertainty. The choice of climate forcing strongly influences ice sheet projections and is a major source of uncertainty, particularly beyond 2100. Projections of Arctic and Antarctic contributions to sea level rise vary widely depending on the climate model and forcing used. Much of this uncertainty arises from ice sheet dynamics and increases rapidly on longer timescales, highlighting the need to extend projections beyond 2100, for example to 2300, while also improving the representation of ice sheet–climate interactions.
The session also highlighted the limitations in how ice sheet processes have historically been represented in climate models. Ice sheet freshwater fluxes were often either absent or treated in highly simplified ways. More recent studies have begun to explicitly include ice sheet–derived freshwater fluxes, improving the representation of their impacts on ocean circulation and the energy budget. In this context, ISMIP6 marked the first coordinated effort to bring together CMIP and ice sheet modelling, providing the foundation for ISMIP7.
Complementing this effort, the session also addressed the potential for ice sheets to undergo tipping transitions, characterised by hysteresis, irreversibility, abrupt changes, and self-amplifying feedbacks. These processes are investigated within TIPMIP, which aims to provide a systematic multi-model assessment of ice sheet tipping dynamics, including the likelihood of crossing critical thresholds, the resulting committed changes, their reversibility, and associated uncertainties. Together, ISMIP and TIPMIP aim to improve our understanding of the future evolution of the Greenland and Antarctic ice sheets under continued warming.
Highlights from the Q&A session
A recurring point in the Q&A discussion concerned scenario extensions, particularly whether simulations should be extended to 2500 and which scenarios should be prioritised. Participants expressed a preference for constructing “plausible worst-case” scenarios. Rather than focusing on very long timescales (e.g. 2500), a more practical priority is to obtain extensions beyond 2150 as soon as possible to support studies of long-term processes such as sea level rise and ice sheet evolution. The importance of extending projections beyond 2100, particularly towards 2300, was also repeatedly emphasised to capture the long-term evolution of the Greenland and Antarctic ice sheets. These multi-centennial timescales are essential for understanding future sea level rise, although they are associated with substantially increased uncertainty. Long-term projections are inherently uncertain and often rely on idealised assumptions, but they remain essential for exploring the range of possible future outcomes. At longer timescales, it becomes clearer which types of uncertainty dominate, highlighting the value of long-term projections and tipping dynamics studies.
A further question concerned whether experimental designs for very long-term simulations (e.g. up to 10,000 years) would be formalised and whether additional factors such as orbital changes would be included. The response noted that current experiments are standalone ice sheet simulations that do not include orbital forcing or full Earth system feedbacks, instead assuming a constant climate after branching to assess committed changes.
Besides the question of timescales, the importance of resolving key ice sheet and ice shelf processes was also repeatedly highlighted during the discussion. This is reflected in the limitations of current Earth system models (ESMs), many of which do not resolve the Antarctic continental shelf and exhibit large temperature biases near ice shelves, despite ocean processes playing a central role in Antarctic ice sheet melting. For Antarctica, further limitations are closely linked to uncertainties in ocean forcing, including ocean temperature, marine ice sheet instability, and Southern Ocean processes. More broadly, key physical processes driving ice sheet sensitivity remain inadequately captured and are strongly region-dependent. For Greenland, limitations include the representation of seasonal temperature, particularly summer warming, and cloud–climate interactions, which are regionally important. In addition, very few models include interactive dynamic ice sheets, meaning that ice sheet–climate feedbacks remain poorly represented and not fully understood. Beyond improving model parameterisations through observations, constraints from palaeorecord are also critical, as present-day ice sheets may not be in equilibrium but instead retain long-term memory of past climates, such as the Last Glacial Maximum and the Holocene. Initialization is therefore fundamentally a palaeoclimate-constrained problem and represents a key challenge for future CMIP developments.
I also noted that a question concerned whether freshwater fluxes from ISMIP could be used to force climate models and whether such outputs could be provided. The response confirmed that hybrid approaches are possible, in which freshwater fluxes are prescribed from model outputs or parameterisations. ISMIP7 is expected to provide such freshwater flux outputs. Ensuring that freshwater fluxes derived from ice sheet models are consistent with real-world observations is important not only for TIPMIP experiments, such as ocean simulations in which freshwater fluxes are prescribed using magnitudes derived from previous multi-model ice sheet experiments, but also for broader research on the impacts of ice sheet melting on ocean circulation and global climate.
Future evolution of ice sheets
Overall, this discussion focused on the current state of ice sheet modelling within CMIP frameworks, existing challenges, and key scientific questions that urgently need to be addressed. It also highlighted palaeoclimate perspectives in understanding ice sheet processes and the initialization state on longer timescales. The session underscored both the importance and the difficulty of representing long-term ice sheet–climate interactions within CMIP and pointed to the need for improved coupling strategies and longer-term experimental design in CMIP7. Both scientific understanding and technical limitations still require further investigation.
I have seen a growing collaborative community, with ongoing progress and increasing efforts. Looking ahead, continued exploration may lead to a more comprehensive understanding and more realistic projections. Given the profound impact of ice sheet melting on global sea level and the direct consequences for coastal populations, this is precisely where the significance of our scientific research lies.
My suggestions for the future are:
- Strengthen communication and collaboration between CMIP and ISMIP. Reliable ice sheet projections require Earth system models in CMIP7 to run relevant experiments, particularly on longer timescales. This calls for further dialogue and coordination between the communities.
- Ice sheet model uncertainty remains a critical issue and requires further systematic investigation and evaluation.
- Encourage greater inclusivity of different modelling approaches and solutions. Ice sheet models and alternative approaches such as freshwater flux products should be used in a complementary way to cross-validate and strengthen results.
