PLE2.1 – AMOC and Earth system tipping points in CMIP: Progress Challenges and TIPMIP insights
The CMIP2026 workshop, held in Kyoto in March 2026, brought together the climate modelling community to assess progress and chart directions for the next phase of the Coupled Model Intercomparison Project. I was grateful to attend the workshop with support from the CMIP travel award, and I would like to acknowledge the CMIP-IPO and the workshop organising committee for enabling early-career researchers like myself to participate in these discussions.
Among a diverse set of sessions over the course of 5 days, one that stood out was PLE2.1 – AMOC and Earth system tipping points in CMIP: progress, challenges, and TIPMIP insights. This session brought together a set of talks that collectively tackled a central question in climate science: how close are we to large-scale, potentially irreversible changes in the Earth system, and how well can current models capture them?
Framing tipping points: from theory to coordinated experiments
Ricarda Winkelmann introduced the Tipping Points Model Intercomparison Project (TIPMIP), which aims to systematically assess tipping risks across Earth system components. In recent years, tipping points have become prominent enough to warrant a dedicated chapter in the upcoming IPCC assessment cycle. As a result, there is growing urgency to move beyond conceptual understanding toward coordinated model experiments.
A key focus is on ice sheets as tipping elements. Their evolution is governed by competing feedbacks, but crucially includes self-reinforcing processes that can lead to abrupt transitions and long-term commitment—even if forcing is later reduced. These dynamics are often hysteretic, meaning that reversing the forcing does not necessarily reverse the change.
An important takeaway here is that tipping behaviour is not uniform: it emerges across nested spatial and temporal scales, with some regions exhibiting gradual responses and others abrupt shifts. TIPMIP provides a structured framework to diagnose and compare these behaviours across models, even though—at least for now—most experiments rely on coarse-resolution Earth system models.
Overshoot pathways and reversibility: a more nuanced picture
Using idealised overshoot scenarios with the EC-Earth3-ESM, Chuncheng Guo explored how the climate system responds to warming levels up to 10°C followed by stabilisation and cooling.
Several results challenge simple expectations:
- Polar amplification persists even after temperatures return to pre-industrial levels, with residual warming at high latitudes.
- The Arctic sea ice recovers, while Antarctic sea ice exhibits an overshoot, temporarily exceeding its initial extent during cooling.
- Most notably, the Atlantic Meridional Overturning Circulation (AMOC) recovers under emissions reductions and can even overshoot its initial strength, showing little to no hysteresis.
That last point is particularly striking. Given expectations of tipping-like behaviour in the AMOC, the apparent reversibility—despite ongoing Greenland ice sheet melt—raises important questions about model limitations and missing processes.
AMOC thresholds and global climate impacts
Saloua Peatier presented results from UKESM TIPMIP experiments, highlighting threshold behaviour in AMOC decline under net-zero scenarios.
Here, the AMOC response appears state-dependent:
- Below a certain warming level, the AMOC can stabilise or recover.
- Beyond that threshold, decline continues even without further emissions.
This behaviour is closely tied to North Atlantic SST patterns and shifts in the Intertropical Convergence Zone (ITCZ). The downstream impacts are substantial:
- Weakening AMOC states are associated with drying over West Africa.
- Similar signals emerge in the Indian monsoon system.
The implication is clear: even under net-zero emissions, regional climate risks remain strongly conditioned by AMOC evolution.
Missing processes: ice sheet melt and ocean eddies
A recurring concern in CMIP-class models is the representation of key physical processes. Oliver Mehling addressed this by incorporating Greenland ice sheet meltwater and comparing standard models with eddy-resolving ocean simulations.
The results suggest that:
- Meltwater leads to additional AMOC weakening, but the magnitude remains modest in the near term.
- Even in higher-resolution simulations, abrupt transitions are not prominent, and AMOC changes scale approximately linearly with cumulative emissions.
- Recovery under emissions reductions still occurs, again with minimal hysteresis.
These findings reinforce a tension: while theory and paleoclimate evidence suggest tipping risks, current models may underestimate non-linear behaviour, possibly due to missing or simplified processes.
Coupling matters: ice sheets, oceans, and the Southern Hemisphere
Kyung-Sook Yun and Axel Timmermann highlighted limitations in how ice sheets are represented in Earth system models. Many approaches rely on offline or weakly coupled frameworks, missing key feedbacks between ice, ocean, and atmosphere.
Their results suggest that the East Antarctic Ice Sheet may be more stable than previously thought, partly due to atmospheric circulation shifts and precipitation responses. However, they also emphasised the importance of bi-hemispheric coupling, including the “bipolar seesaw” behaviour linking the two hemispheres.
Complementing this, Sam Sherriff-Tadano examined Southern Ocean feedbacks on millennial timescales. A weakened AMOC redistributes heat into deeper ocean layers and the Southern Ocean, potentially reinforcing variability through feedbacks involving sea ice and deep-water formation.
Key takeaways
Across these talks, a few themes emerged:
- Tipping points are complex: they involve multi-scale, interacting processes that are not yet fully captured in current models.
- AMOC behaviour is more reversible in models than expected, with limited evidence of strong hysteresis—even under large perturbations.
- Thresholds still matter even without abrupt collapse, state-dependent changes in AMOC can have major regional impacts.
- Model limitations remain critical: missing processes such as dynamic ice sheet coupling and ocean eddies may bias results toward smoother, more reversible behaviour.
- TIPMIP provides a crucial step forward, offering a coordinated framework to rigorously test these dynamics.
Looking ahead
This session underscored both progress and uncertainty. While CMIP-class models are increasingly capable of exploring complex Earth system behaviour, the question of tipping points remains open—not because it lacks importance, but because it sits at the edge of what current models can robustly resolve.
Initiatives like TIPMIP are therefore essential. By systematically probing thresholds, feedbacks, and reversibility, they help bridge the gap between theoretical understanding and model-based projections—ultimately improving how we assess and communicate climate risks.
