Decadal to seasonal variability of Arctic sea ice albedo

A controlling factor in the seasonal and climatological evolution of the sea ice cover is its albedo $\alpha$. Here we analyze Arctic data from the Advanced Very High Resolution Radiometer (AVHRR) Polar Pathfinder and assess the seasonality and variability of broadband albedo from a 23 year daily record. We produce a histogram of daily albedo over ice covered regions in which the principal albedo transitions are seen; high albedo in late winter and spring, the onset of snow melt and melt pond formation in the summer, and fall freeze up. The bimodal late summer distribution demonstrates the combination of the poleward progression of the onset of melt with the coexistence of perennial bare ice with melt ponds and open water, which then merge to a broad peak at $\alpha \gtrsim $ 0.5. We find the interannual variability to be dominated by the low end of the $\alpha$ distribution, highlighting the controlling influence of the ice thickness distribution and large-scale ice edge dynamics. The statistics obtained provide a simple framework for model studies of albedo parameterizations and sensitivities.

💡 Research Summary

This paper investigates the seasonal and interannual variability of broadband albedo (α) over Arctic sea‑ice using a 23‑year daily record (1982–2004) from the Advanced Very High Resolution Radiometer (AVHRR) Polar Pathfinder (APP) dataset. The authors first retrieve apparent hemispheric albedo for each 5 km × 5 km pixel, applying a four‑step correction (solar‑zenith normalization, conversion to TOA broadband reflectance, anisotropy correction, and clear‑sky surface albedo conversion). Sea‑ice pixels are identified with the Surface Type Mask and the NASA Team Sea‑Ice Algorithm, which distinguishes first‑year ice (FYI) from multi‑year ice (MYI).

A novel aspect of the analysis is the introduction of a “time‑threshold” (τth) that specifies the minimum number of years a pixel must have contained ice to be included in a given histogram. Six thresholds are used (τth = 23, 22, 17, 12, 8, 1 yr). For each day and each τth, the authors compute the mean albedo of all qualifying pixels and construct histograms of pixel counts versus albedo bins (0.2–1.0). This approach separates the contributions of long‑standing thick MYI (high τth) from more transient FYI and open‑water/melt‑pond areas (low τth).

The seasonal evolution of the albedo histograms reveals five distinct phases:

1. Late winter–spring: A single, narrow peak near α ≈ 0.75–0.80, representing snow‑covered ice. The high‑albedo side drops sharply, while the low‑albedo tail reflects mixed pixels near the ice edge and occasional open water.

2. Early melt (early June): The peak shifts downward to α ≈ 0.70–0.65 as snow melts and wet ice appears. Low‑τth histograms show an expanding low‑albedo tail, indicating the emergence of melt ponds and thin FYI.

3. Mid‑summer (June–July): Histograms become bimodal. One mode remains near α ≈ 0.70 (remaining snow‑covered MYI), while a second mode appears around α ≈ 0.45–0.55, corresponding to melt‑ponds and bare thin ice. The bimodality reflects the poleward progression of melt onset, which can span several hundred kilometres and take about a week to move from 73° N to 82° N.

4. Late summer (July–August): The two modes merge into a broad peak centered near α ≈ 0.5–0.6, indicating a mixed surface of perennial ice, melt ponds, and open water.

5. Fall freeze‑up (September–October): The low‑albedo component recedes, the peak shifts slightly rightward to α ≈ 0.6–0.65, and the surface re‑solidifies.

Mean albedo values increase monotonically with τth, confirming that longer‑lived, thicker ice has higher reflectivity. Conversely, low‑τth histograms display greater variability, especially on the low‑albedo side, driven by fluctuations in ice‑type fractions and the position of the ice edge. The authors report an annual standard deviation of 0.05–0.07, dominated by the α < 0.5 region.

The paper also discusses the role of cloud cover. Seasonal cloud fraction varies from ~20 % in winter to ~70 % in summer. During the period of increasing cloud cover (April–June), the τth dependence of mean albedo is minimal, which the authors attribute to the surface approaching the bulk freezing point, thereby reducing albedo differences among ice types and water. In contrast, during autumn, when cloud cover declines, τth dependence strengthens, reflecting the growing influence of open water and melt‑ponds on the albedo distribution.

Methodological uncertainties are acknowledged: (i) the retrieval is less accurate when snow is wet or when melt ponds are present, because liquid water alters the emission characteristics; (ii) near the ice edge, mixed pixels of ice and open water introduce additional error; (iii) cloud‑mask imperfections can affect the apparent albedo. Nevertheless, the authors argue that the statistical framework they present is robust enough to serve as a benchmark for model development.

In the conclusions, the authors emphasize that:

• The histogram‑based approach captures the principal albedo transitions (high‑albedo snow, melt onset, pond formation, and freeze‑up).
• The late‑summer bimodality is a clear signature of simultaneous melt‑onset progression and coexistence of bare ice and melt ponds.
• Interannual variability is dominated by the low‑albedo tail, implying that changes in ice thickness distribution and ice‑edge dynamics are the primary drivers.
• Because albedo depends on ice thickness (e.g., Eisenman & Wettlaufer 2009), the observed shift toward lower albedo in recent years (with increasing FYI) underscores the need for climate models to represent thickness‑dependent albedo more realistically.

The authors suggest that modelers can directly incorporate the provided histograms or their moments as input, and use the τth‑dependent variances as a baseline for stochastic albedo parameterizations. This could improve the representation of the ice‑albedo feedback in simple energy‑balance models as well as in fully coupled GCMs, which currently under‑project Arctic sea‑ice loss partly due to overly simplistic albedo schemes.

Overall, the study delivers a comprehensive, observationally grounded statistical description of Arctic sea‑ice albedo variability across multiple time scales, offering a valuable resource for the refinement of albedo parameterizations in climate modeling.