In search of climate attractors

Climate attractors are asymptotic steady states of the climate system, embedded in a high-dimensional phase space. They represent distinct climatic regimes, separated by unstable boundaries where small perturbations can cause the climate to transition from one attractor to another. Identifying climate attractors in simulations performed with state-of-the-art models is challenging [1] due to the high computational costs associated with running multi-millennial, multi-component simulations with a continuous spectrum of variability. Nevertheless, the number of attractors and their stability ranges can provide crucial information about the numerical representation of nonlinear interactions in a model, and reveal the  dynamical structure of the climate system.

In the search for climate attractors under the present-day continental configuration, we used a recently developed modelling framework called biogeodyn-MITgcmIS [2], in which the dynamical core of both the atmosphere and the ocean is provided by the MIT general circulation model, while offline coupling ensures the consistent evolution of vegetation and ice sheets. Using this coupled setup, we identified three distinct climatic states: a glacial state, an interglacial state, and a hot state with a strongly reduced Greenland ice sheet. These states coexist over a range of atmospheric CO₂ concentrations, thereby defining hysteresis paths between the attractors.

Here, we describe the methodology used to identify these attractors and highlight the crucial role of ice-sheet and vegetation evolution. We characterize the attractors in terms of their dominant feedback mechanisms. We find that, while the positive overturning cell mainly changes in intensity during the transition from the cold to the warm state, it collapses during the transition from the warm to the hot state. Crossing the warm-hot boundary involves substantial vegetation changes, the disappearance of the Greenland ice sheet, and a reduction of sea ice in the Antarctic region. Finally, we discuss the need to repeat similar investigations using different climate models to assess the robustness of the identified attractors and mechanisms.

[1] Brunetti and Ragon, Phys. Rev. E 107, 054214 (2023)

[2] Moinat et al., EGUsphere [preprint], https://doi.org/10.5194/egusphere-2025-2946 (2025)