Earth Science · Cryosphere
June 2026 · 12 min read · Sea Ice · Albedo Feedback · Tipping Points

Arctic Sea Ice: Physics of Freezing and the Albedo Feedback

Sea ice covers an area of the Arctic Ocean that swells and shrinks with the seasons, peaking near 15 million square kilometres each March and bottoming out each September. It is far more than a frozen lid: it is a thermostat for the planet. A thin, bright skin of ice reflects most of the sunlight that strikes it, while the dark ocean beneath absorbs almost all of it — a contrast that makes sea ice one of the most powerful feedback elements in the climate system. This article works through the physics of how sea ice freezes and grows, why it rejects salt, how the ice-albedo feedback amplifies warming, the crucial difference between first-year and multi-year ice, and why the Arctic may be approaching a tipping point.

1. Stefan's Law: How Ice Grows

When the air above open ocean drops below the freezing point of seawater (about −1.8 °C for typical salinity), ice begins to form at the surface. As the ice thickens, it insulates the water below from the cold air, so growth slows. Josef Stefan derived the governing relationship in 1891 by balancing the latent heat released at the freezing front against heat conducted up through the existing ice.

Heat conducted through ice of thickness h: Q = k · (T_f − T_a) / h (Fourier conduction) This heat must remove the latent heat of fusion as new ice of thickness dh forms: ρ · L · (dh/dt) = k · (T_f − T_a) / h Integrating from h = 0 gives Stefan's law: h(t) = √( 2·k·(T_f − T_a)·t / (ρ·L) ) where k = thermal conductivity of ice (~2.2 W/m·K) T_f = freezing temperature, T_a = air temperature ρ = ice density (~917 kg/m³) L = latent heat of fusion (~3.34×10⁵ J/kg) t = time

The key consequence is the square-root growth: ice thickens quickly at first but ever more slowly as it deepens, because thicker ice insulates better. This is often expressed via freezing degree days (the accumulated product of sub-freezing temperature and time), giving the convenient field rule h ≈ √(FDD) in appropriate units. Stefan's law explains why undeformed first-year ice rarely exceeds about 2 metres in a single Arctic winter — conduction simply cannot remove heat fast enough through ice that thick.

2. Brine Rejection and Salinity

Seawater is salty, but the crystal lattice of pure ice cannot accommodate salt ions. As water freezes, salt is excluded from the forming ice and concentrated into pockets of dense, super-salty liquid — brine — trapped between the crystals. Much of this brine drains downward through channels, a process called brine rejection.

Brine rejection has two profound consequences:

Why old ice is drinkable: Arctic explorers learned that multi-year ice, having shed most of its brine over successive summers, is nearly fresh and can be melted for drinking water — a direct, practical signature of the brine-rejection process.

3. The Ice-Albedo Feedback

Albedo is the fraction of incoming sunlight a surface reflects. The albedo contrast between sea ice and open ocean is enormous, and it sets up one of the most important positive feedbacks in the climate system.

Typical shortwave albedos: Fresh snow-covered ice α ≈ 0.80–0.85 (reflects ~80%) Bare sea ice α ≈ 0.50–0.65 Melt ponds on ice α ≈ 0.20–0.40 Open ocean α ≈ 0.06 (absorbs ~94%) Absorbed solar flux: F_abs = (1 − α) · S_incoming

The ice-albedo feedback works as a vicious circle: warming melts some ice, exposing dark ocean that absorbs far more sunlight, which warms the water further, which melts more ice. The feedback is positive — it amplifies any initial change. A region that loses its bright ice cover absorbs roughly an order of magnitude more solar energy, so even small reductions in ice extent translate into large extra heat uptake. Melt ponds add a sub-cycle of their own: as ponds darken the ice surface in summer, they accelerate melt and lower albedo even before the ice disappears.

4. Multi-Year versus First-Year Ice

Not all sea ice is equal. First-year ice (FYI) forms and melts within a single annual cycle. It is thinner (typically under 2 m), saltier, more uniform, and mechanically weaker. Multi-year ice (MYI) has survived at least one melt season — often many — and is thicker (3–5 m or more where ridged), nearly fresh, harder, and far more resistant to melting.

The distinction matters enormously for climate. Multi-year ice is the Arctic's structural backbone and its memory: thick, durable, and slow to melt. Over recent decades the Arctic has shifted dramatically from a MYI-dominated to a FYI-dominated state. The oldest, thickest ice has shrunk to a fraction of its former area, leaving a thinner, younger, more fragile ice pack that is far more vulnerable to a single warm summer or a strong storm.

Trends (satellite era, 1979–present): September minimum extent declining ~13% per decade Volume declining even faster than area (ice is also thinning) Multi-year ice fraction sharply reduced → pack now mostly first-year ice

5. Polar Amplification

The Arctic is warming several times faster than the global average — a phenomenon called polar amplification, or more specifically Arctic amplification. The ice-albedo feedback is a leading cause, alongside changes in atmospheric and ocean heat transport, the trapping of heat by water vapour and clouds, and the thin, stable Arctic atmosphere that confines surface warming to a shallow layer.

As the highly reflective ice and snow retreat, the newly exposed ocean stores summer heat that delays autumn freeze-up, thins the following winter's ice, and feeds back into still more melt. Polar amplification means the Arctic is both a sensitive early indicator of global warming and an active amplifier of it — changes there ripple outward through the jet stream and ocean circulation to influence weather far beyond the polar regions.

6. The Sea-Ice Tipping Point

Because the ice-albedo feedback is positive, scientists have long asked whether Arctic sea ice could pass a tipping point — a threshold beyond which loss becomes self-sustaining and effectively irreversible. The simplest energy-balance models do show the possibility of an abrupt jump to an ice-free state once summer ice thins past a critical point.

Energy balance for ice extent (schematic): dE/dt = (1 − α(E))·S − (outgoing longwave + ocean heat) Because α depends strongly on E (ice vs ocean), the feedback can make the steady-state curve fold → multiple equilibria → abrupt loss.

Encouragingly, more detailed models suggest the summer sea-ice loss is largely reversible and tracks global temperature without a hard cliff, because the strong seasonal cycle and longwave cooling of open water provide a stabilising negative feedback in winter. The consensus is that the first ice-free Arctic September — likely within the coming decades under continued warming — is driven primarily by rising temperatures rather than an irreversible runaway. Even so, the disappearance of summer sea ice would be a dramatic, planet-scale transformation: a darker Arctic absorbing far more heat, with consequences for ecosystems, weather patterns, ocean circulation, and the carbon cycle that we are only beginning to understand.

🧊
Arctic Ice Simulator
Grow and melt sea ice under different forcings and watch the albedo feedback run
⚠️
Climate Tipping Points Explorer
Test whether sea ice and other elements cross irreversible thresholds
🌊
Ocean Current Simulator
See how brine rejection and polar cooling drive the global overturning circulation