A Low Order Theory of Arctic Sea Ice Stability

We analyze the stability of a low-order coupled sea ice and climate model and extract the essential physics governing the time scales of response as a function of greenhouse gas forcing. Under present climate conditions the stability is controlled by longwave radiation driven heat conduction. However, as greenhouse gas forcing increases and the ice cover decays, the destabilizing influence of ice-albedo feedback acts on equal footing with longwave stabilization. Both are seasonally out of phase and as the system warms towards a seasonal ice state these effects, which underlie the bifurcations between climate states, combine exhibiting a “slowing down” to extend the intrinsic relaxation time scale from ~ 2 yr to 5 yr.

💡 Research Summary

This paper conducts a linear stability analysis of the low‑order coupled sea‑ice–climate model originally introduced by Eisenman and Wettlaufer (2009). The model describes the energy stored in the system by a single variable E (units W m⁻² yr), which is negative when sea ice is present and positive when the ocean mixed layer is exposed. Radiative forcing is split into short‑wave input F_S(t), long‑wave outgoing F_0(t) and an imposed greenhouse‑gas perturbation ΔF₀. The surface albedo α(E) depends on ice thickness through a hyperbolic‑tangent transition, capturing the rapid drop in albedo as ice thins.

A periodic steady state E_S(t) represents the annual cycle of ice growth and melt. Small perturbations ξ(t) about this state obey ξ̇ = a(t) ξ, where the instantaneous response rate a(t) is decomposed into four physically interpretable terms:

1. Ice‑albedo feedback (a_IA) – Positive; a thinning ice sheet reduces albedo, increasing absorbed short‑wave radiation and amplifying the perturbation.
2. Albedo response (a_AR) – Represents the temperature‑mediated response of long‑wave emission to changes in absorbed short‑wave radiation; it is non‑zero mainly in spring and autumn when the surface temperature departs from the freezing point.
3. Long‑wave stabilization (a_LW) – Negative; during the polar night, long‑wave radiative loss drives conductive heat loss through the ice, promoting rapid growth of thin ice and damping perturbations.
4. Ice export (a_EX = −v₀) – Constant negative term accounting for the loss of ice through advection.

The time‑averaged growth rate γ = (1/T)∫₀ᵀ a(s) ds determines the exponential fate of perturbations: γ < 0 indicates a stable periodic solution with relaxation time τ = |1/γ|, while γ > 0 signals instability and a transition to a different climate regime.

The authors systematically vary the greenhouse forcing ΔF₀ from 10 to 22 W m⁻² and compute the seasonal evolution of a(t) and the resulting γ. At low forcing (ΔF₀≈10 W m⁻²) the system is dominated by the long‑wave term; a_LW is large and negative, giving τ≈2–3 years. As ΔF₀ increases, summer ice thins, and a_IA grows sharply. Near ΔF₀≈19.8 W m⁻² the destabilizing albedo feedback nearly cancels the stabilizing long‑wave term, driving γ toward zero. This “critical slowing down” lengthens τ to about 5 years, a hallmark of an approaching bifurcation.

Further warming (ΔF₀≈22 W m⁻²) eliminates summer ice entirely; the ice never regrows in winter because the albedo feedback remains positive throughout the year. Consequently γ becomes positive, the periodic solution disappears, and the system undergoes a saddle‑node bifurcation to a seasonally ice‑free or permanently ice‑free state. The critical ΔF₀ values (≈19.8 W m⁻² for the perennial‑to‑seasonal transition and ≈22.2 W m⁻² for the seasonal‑to‑ice‑free transition) are identified from the zero‑crossings of γ.

The analysis highlights the interplay of two out‑of‑phase mechanisms: (i) long‑wave radiative cooling that stabilizes winter ice through conductive heat loss, and (ii) the ice‑albedo feedback that destabilizes summer ice. Their competition produces a non‑monotonic dependence of system stability on greenhouse forcing. The model also shows sensitivity to the characteristic extinction thickness h_α (≈0.5 m); smaller values would unrealistically amplify the albedo feedback.

By reducing the complex Arctic system to a tractable set of equations, the paper provides a clear mechanistic explanation for observed rapid Arctic sea‑ice loss and the associated increase in intrinsic response times. The derived γ‑framework offers a quantitative diagnostic that can be applied to higher‑resolution climate models and observational datasets to detect early warning signals of impending Arctic regime shifts.