Melting summer sea ice uses energy because the ice must change phase from solid to liquid. It also exposes dark water in place of a bright surface. The water absorbs more of the sunlight that follows, and that extra energy can warm the upper ocean or melt more ice. The ice–albedo feedback is this self-reinforcing change in reflection, absorption, and melting.

From reflectivity to absorbed sunlight

Albedo, $\alpha$, is the fraction of incoming shortwave radiation that a surface reflects. It ranges from 0 when all the radiation is absorbed to 1 when all of it is reflected. The absorbed fraction is $1-\alpha$:

$$S_{\mathrm{abs}}=(1-\alpha)S,$$

where $S$ is incoming sunlight at the surface.

Representative Arctic values used here are:

Using the same illustrative input of $458\ \mathrm{W/m^2}$ for every surface shows the size of the albedo contrast. This calculation holds the incoming flux fixed; it does not mean that every surface actually receives that value:

With this common input, open ocean absorbs more than six times as much sunlight as fresh snow and more than twice as much as bare ice. The exact albedo changes with snow age, wetness, pond depth, ice type, and solar angle, but fresh snow generally reflects the most and open water the least.

How the loop closes

A small warming from any source melts some snow or ice, which lowers the average surface albedo. The darker surface absorbs a larger fraction of the next incoming sunlight. That extra energy warms the upper ocean or melts more ice, so it strengthens the change that began the sequence.

The ice–albedo feedback

• Fresh snow: 0.85
• Bare sea ice: 0.60
• Melt pond: 0.25
• Open ocean: 0.06

This is a positive feedback because the effects reinforce the initial change. The process can also reinforce cooling: persistent snow or ice keeps the albedo high, reduces solar absorption, and favours further cooling and ice survival. Positive does not mean beneficial, and it does not mean that the feedback grows without limit.

Melting still requires latent heat. While ice at its melting point remains, part of the net energy changes its phase instead of raising its temperature. Once the reflective ice is gone, water absorbs the same sunlight much more efficiently. The complete surface energy balance determines which effect dominates at a given place and time.

The shortwave part of the feedback stops during polar night. With no incoming sunlight, a change in albedo cannot change solar absorption. Summer absorption can lead to warmer conditions in autumn, but albedo is not affecting the energy balance in darkness.

Put the albedo contrast on an energy scale

An idealized June-to-August comparison at 75°N applies the two surface albedos to top-of-atmosphere solar input and ignores losses in the atmosphere. In this calculation, open water absorbs about 1,800 MJ/m² more shortwave energy than an ice surface. Melting ice requires about $306\ \mathrm{MJ/m^3}$, so the energy difference is equivalent to the latent heat needed to melt

$$\frac{1{,}800\ \mathrm{MJ/m^2}} {306\ \mathrm{MJ/m^3}} \approx5.9\ \mathrm{m}.$$

This comparison sets an energy scale; it does not predict 5.9 m of melt. It uses fixed albedos and incident summer energy. Clouds and atmospheric absorption reduce the sunlight that reaches the surface, while longwave, sensible, latent, and oceanic fluxes continue. The extra energy absorbed by open water also cannot simply be assigned to an ice column that is no longer there. The calculation compares the reflectivity-driven energy difference with latent heat; it does not give an observed melt depth.

Feedback gain is not the observed warming ratio

A simple linear model gives a precise meaning to feedback gain. Suppose an initial response is $\Delta T_0$ and each pass through an idealized feedback adds a fraction $f$ of the response from the preceding pass. For $0\le f<1$,

$$\Delta T =\Delta T_0(1+f+f^2+\cdots) =\frac{\Delta T_0}{1-f}.$$

The multiplier $1/(1-f)$ is the feedback gain. In this teaching model, $f=0.5$ makes the total response twice the initial response, while $f=0.75$ makes it four times as large. A larger gain also magnifies uncertainty in both the initial response and the feedback strength. These are illustrative loop gains. A “pass” is not a calendar year, and the equation does not say that temperature multiplies again each year.

This gain is different from the observed Arctic amplification ratio in §1. That ratio compares Arctic and global temperature trends over a specified period. It includes several feedbacks and energy transports rather than measuring one value of $f$ for the ice–albedo feedback.

The seasonal delay: absorb first, release later

The main parts of the seasonal cycle peak at different times:

• incoming solar energy peaks around June;
• open-water area is greatest near the September sea-ice minimum;
• ocean-to-atmosphere heat release strongly affects near-surface air in October and November.

The ocean loses energy during summer through longwave radiation and sensible and latent heat exchange. At the same time, open water absorbs shortwave sunlight. Its upper layers warm only when the absorbed energy is greater than the simultaneous losses, and they store the difference.

As sunlight weakens in autumn, the energy balance reverses. Cold air moves over water that still holds some of the summer heat. The ocean transfers energy upward through longwave radiation and turbulent sensible and latent heat fluxes. Before ice can form, the upper ocean must cool to its salinity-dependent freezing point. Freezing then releases latent heat, which must also leave the water. A larger store of summer heat therefore tends to delay freeze-up.

Later freeze-up leaves less time for ice growth during winter. The ice that forms is thinner, so more thin first-year ice enters the next melt season. Thin ice can grow quickly during cold winter conditions, but less energy is needed to melt it in spring and summer. This sequence carries the effect from one season into the next without multiplying it by a fixed amount each year.

Arctic near-surface warming can peak in autumn even though albedo is no longer changing solar absorption. The extra shortwave energy entered the ocean earlier. In autumn, sunlight is weak or absent, the air cools, open water releases the stored ocean heat upward, and the missing ice no longer blocks that transfer.

Melt onset and melt ponds

The surface starts to darken before an ice floe disappears. As snow warms, its grains become wetter and coarser. This reduces the multiple scattering that makes fresh snow highly reflective. An earlier melt onset adds more low-albedo days during the part of the year with the strongest sunlight.

Meltwater collects in depressions on intact ice and forms melt ponds. A ponded surface can have an albedo near 0.25, compared with roughly 0.60 for bare ice. Shallow ponds are brighter than deep ponds because more light reaches and reflects from the ice below. Ponds increase absorption even while satellite images still show continuous ice cover.

Pond coverage often peaks in July and can help indicate how much ice will survive until September. It marks an early stage of the surface change: snow-covered ice → wet snow and bare ice → ponded ice → broken floes and open water.

Where the feedback leaves the clearest signal

The local albedo change is largest where the surface is changing from ice to water. In the central pack, ice that remains covered does not complete that transition. Farther south, water that is already open has no remaining bright ice to lose. Along the retreating ice edge, floes, ponds, leads, and open water occur together and shift in area, so the average albedo can change quickly.

This process produces several observable patterns:

• satellite observations should show lower summer surface albedo where ice retreats;
• the upper ocean should gain heat in newly ice-free regions;
• autumn upward heat flux and near-surface warming should be largest where ice would previously have insulated the ocean;
• freeze-up should occur later after earlier retreat and greater summer ocean heat uptake, although winds and ocean transport can modify the relationship.

In the Barents Sea and other marginal seas, a retreating ice edge lowers the surface albedo while ocean currents also deliver heat. The albedo change operates alongside this transported ocean heat, not by itself.

The same change in reflectivity occurs when seasonal snow melts on land and when Greenland's ice surface darkens. The effect on sea level is different. Greenland ice rests on land, so lost ice mass enters the ocean. Floating sea ice already displaces seawater, so its melting has almost no direct effect on sea level even though it still affects climate.

Surface albedo is one part of Arctic amplification, not its complete cause. Clouds, water vapour, the vertical pattern of atmospheric warming, poleward energy transport, and the loss of sea ice's insulating effect all interact with it. Replacing bright snow or ice with darker water increases summer shortwave absorption, adds heat to the upper ocean, and can delay the following freeze-up.