Albedo evolution of seasonal Arctic sea ice



[1] There is an ongoing shift in the Arctic sea ice cover from multiyear ice to seasonal ice. Here we examine the impact of this shift on sea ice albedo. Our analysis of observations from four years of field experiments indicates that seasonal ice undergoes an albedo evolution with seven phases; cold snow, melting snow, pond formation, pond drainage, pond evolution, open water, and freezeup. Once surface ice melt begins, seasonal ice albedos are consistently less than albedos for multiyear ice resulting in more solar heat absorbed in the ice and transmitted to the ocean. The shift from a multiyear to seasonal ice cover has significant implications for the heat and mass budget of the ice and for primary productivity in the upper ocean. There will be enhanced melting of the ice cover and an increase in the amount of sunlight available in the upper ocean.

1. Introduction

[2] The decline of the Arctic sea ice cover is well established. Observations show a reduction in summer ice extent [Comiso et al., 2008; Serreze et al., 2007; Stroeve et al., 2007], a decrease in ice thickness [Rothrock et al., 2008; Giles et al., 2008; Haas et al., 2008; Kwok and Rothrock, 2009], and a lengthening of the summer melt season [Markus et al., 2009]. There has also been a fundamental shift in the character of the ice cover from a primarily multiyear ice pack to mainly seasonal ice [Rigor and Wallace, 2004; Nghiem et al., 2007; Maslanik et al., 2007, 2011; Comiso, 2012].

[3] This shift in the age of the ice has consequences for the morphologic, thermodynamic, and dynamic properties and processes of the ice cover. The treatment of many sea ice properties and processes in models needs to be reconsidered and adapted in response to the new ice conditions. One key element is the parameterization of sea ice albedo. Correctly measuring and predicting albedo is important not only because solar radiation is a large term in the surface heat budget, but also because the ice albedo feedback can amplify ongoing changes in the ice cover.

[4] While sea ice albedo can be estimated from satellites [Lindsay and Rothrock, 1994], these observations are infrequent during summer melt because of the pervasive presence of summer clouds. As a result, the treatment of sea ice albedo in models is based primarily on in situ observations of multiyear ice. We know, however, that the seasonal evolution of seasonal ice albedo differs in several substantial ways from that of multiyear ice [Hanesiak et al., 2001; Grenfell and Perovich, 2004].

[5] In this paper, we examine field observations capturing the evolution of wavelength-integrated albedo of seasonal shorefast sea ice and compare this evolution to that of multiyear ice. Results from four years of melt season observations are presented and synthesized into a general, simplified framework. Finally, the impact of a shift from multiyear to seasonal ice on the partitioning of solar energy is explored for a particular case.

2. Observations

[6] We have conducted several field experiments near Barrow Alaska examining the optical and morphological properties of shorefast, seasonal ice [Grenfell and Perovich, 2004; Polashenski et al., 2012]. In these experiments, a 200 m long measurement line was installed in undeformed seasonal ice. Surface conditions were characterized and spectral and wavelength integrated albedos were measured every 2.5 m along the line, providing both a qualitative and quantitative examination of the evolution of snow and ice physical properties and albedo.

[7] Since approximately two-thirds to three quarters of solar energy is visible light, visually examining how the surfaces of these ice types evolve provides considerable insight into their albedo. Photographs of seasonal ice and multiyear ice at different stages of melt are presented inFigure 1. Before the onset of melt, both seasonal and multiyear ice are covered by snow. Multiyear ice tends to have a deeper snow cover than seasonal [Sturm et al., 2002], but the snow on seasonal ice is also optically thick (greater than ∼0.1 m) and the albedos for snow covered seasonal and multiyear ice are the same (Figure 1a). When melt begins, the thinner snow cover on seasonal ice melts in less time, resulting in more rapid transition to bare ice and a smaller albedo (Figure 1b). Furthermore, the thinner bare seasonal ice contains a much lower bubble fraction in its upper layers, scatters light less effectively, has a smaller freeboard, and has an albedo about 0.1 lower than multiyear ice (∼0.65 vs. ∼0.55). As the snow and surface ice melts, melt ponds form rapidly on both seasonal and multiyear ice. The initial stages of pond formation, however, are extremely different for seasonal and multiyear ice. Level, undeformed seasonal ice can reach pond fractions greater than 0.7 [Polashenski et al., 2012], as meltwater, unconstrained by topographic features, spreads in shallow ponds of great spatial coverage (Figure 1c). In contrast, multiyear ice, with its undulating topography, typically has peak pond fractions of only 0.3–0.4 [Fetterer and Untersteiner, 1998; Perovich et al., 2002a; Tschudi et al., 1997] because meltwater is collected into deeper ponds with less area. This initial stage of maximum pond coverage lasts only a few days on both seasonal and multiyear ice, after which increased surface melt water drainage reduces the area covered by ponds. The reduction occurs on both ice types, but is markedly greater on seasonal ice. After these dynamic initial stages of melt pond formation, pond coverage slowly increases through the rest of the melt season on both ice types (Figure 1d). At some point the seasonal ice completely melts and there is an open water period. There is a tremendous difference between the initial growth of seasonal ice and multiyear ice during fall freezeup (Figure 1e). Multiyear ice albedos increase rapidly as the as an ice layer quickly freezes on the surface of melt ponds and the first snowfall occurs. Albedo increases on seasonal ice later and more slowly. Seasonal ice forms from open water in the fall, a process which may take several weeks if the upper ocean warmed significantly in summer. Only when the ice is thick enough to support a snow cover [Perovich, 1991] will there be a significant rise in albedo.

Figure 1.

Sequence of photographs showing the evolution of albedo for seasonal ice (left column) and multiyear ice (right column). Seasonal ice photographs are from Barrow Alaska and multiyear ice are from SHEBA. The sequence is: (a) cold snow, (b) melting snow, (c) initial melt pond formation, (d) melt pond evolution, and (e) fall freezeup.

[8] This qualitative description demonstrates that there are similarities and differences in the evolution of albedo for seasonal and multiyear ice. We now quantitatively examine the evolution of seasonal ice albedo using four years of wavelength–integrated albedos (300–3000 nm) measured on shorefast seasonal ice near Barrow, Alaska (Figure 2). A Kipp and Zonen albedometer was used to measure albedos every 2.5 m along a 200-m-long line, which were then averaged, giving an areal mean albedo. The interannual variability, evident in the four albedo time series, is caused primarily by variations in the timing of melt onset and intensity of melt. Despite this variability, albedos from all four years follow a similar pattern. First, there is cold optically thick snow with an albedo of ∼0.85. As temperatures increase, the snow cover begins to warm and melt, thereby decreasing the snow albedo to about 0.7. Next, there is a rapid, large decrease in albedo as melt ponds form and proliferate to cover as much as 70% of the surface. Albedos decrease from about 0.7 to as low as 0.2 during this period. There is then a rapid rebound of albedo to as high as 0.6 as surface water drainage occurs and the pond fraction decreases to as little as 10%. Albedos then decrease over time as the melt ponds gradually widen and the ice underlying the ponds thins [Polashenski et al., 2012].

Figure 2.

Four years of albedo time series measured in shorefast ice near Barrow Alaska. The albedos are values averaged over a 200-m-long line.

[9] Though the time-series of observations presented here do not continue far enough into the melt season to demonstrate this, isolated shipborne observations we have made suggest that, as melt progresses into its latest stages, the un-ponded ice may thin to the point where it no longer has sufficient freeboard to maintain a surface scattering layer of drained ice crystals. As the unponded ice fraction becomes very thin (less than about 0.5 m), the steady decline in albedo shown inFigure 2, would greatly accelerate [Frey et al., 2011]. We represent this rapid change as a step decrease in albedo to 0.07 (open ocean) occurring once albedo drops below 0.2. A seasonal ice albedo of 0.2 denotes thin, highly porous ice that has been observed to completely melt in just a few days.

[10] The results in Figures 1 and 2 provide insights into the evolution of seasonal ice albedo. We can use these insights to formulate a generalized albedo evolution sequence for seasonal ice (Figure 3). In this particular example, melt is defined to start on May 29 and freezeup on August 13. Figure 3 presents a simplified albedo evolution that can be described in seven steps.

Figure 3.

Time series of the evolution of seasonal ice albedo. Seven phases of melt are illustrated. The evolution of multiyear ice albedo [Perovich et al., 2007] is plotted in blue and seasonal ice albedo is in red. These particular time series assume melt onset on 29 May and freezeup on 13 August.

[11] 1. Cold snow: Before melt onset the snow albedo is 0.85.

[12] 2. Melting snow: Starting with the onset of snow melt, there is a linear decrease to 0.6 in 7 days.

[13] 3. Pond formation: For the next 7 days, decrease from 0.6 to 0.32.

[14] 4. Pond drainage: For next 7 days, increase from 0.32 to 0.54.

[15] 5. Pond evolution: Decreases by 0.0083 d−1 until albedo reaches 0.2. Then the ice is assumed to rapidly melt and albedo drops in a single day to 0.07.

[16] 6. Open water: Albedo of 0.07

[17] 7. Freezeup: Once new seasonal ice begins to form from open water the albedo increases by 0.0082 d−1 to a maximum of 0.85

[18] There are a few caveats concerning this approach. It is only for undeformed seasonal ice. Deformed seasonal ice will have a different melt pond evolution. There are nuances that are not included in this simplified treatment, such as changes in snow albedo due to metamorphism and daily summer albedo fluctuations due to weather (e.g. a midsummer snowfall). Integrated over an entire summer melt cycle these omissions will have modest impact [Perovich et al., 2002a]. There will also be variations in albedo evolution from place to place and from year to year. These variations will primarily impact the rate of albedo decline in the pond evolution stage. Larger atmospheric fluxes will increase the rate of the albedo decrease during pond evolution. However, the stages of the albedo evolution and thus the fundamental behavior will be similar. The final stages of pond evolution and the transition to open water were inferred from qualitative field observations. This seven-step sequence can be used with observed onset dates of melt and freezeup [Markus et al., 2009] to generate large-scale estimates of the albedo evolution of seasonal ice.

[19] A multiyear albedo evolution, derived from Perovich et al. [2002b, 2007], is plotted for comparison. The seasonal snow cover is not as deep, so it melts away faster. As ponds form, albedo drops much more rapidly on seasonal ice, because the flat topography encourages more extensive pond formation. During pond evolution, the thinner seasonal ice still displays smaller albedos than multiyear ice, since the albedo of seasonal ponds is typically less than multiyear ponds, and because ponds typically remain more extensive on seasonal than multiyear ice during this stage [Grenfell and Perovich, 2004]. Eventually, after fall freezeup, the albedos of seasonal and multiyear ice converge, as both types of ice become covered by an optically thick snow layer.

3. Discussion

[20] The decreased albedo of seasonal ice results in greatly increased solar input. We illustrate this by considering a particular example. During the summer of 1998, measurements of incident, reflected, and transmitted solar radiation of multiyear ice were made as part of the Surface Heat Budget of the Arctic Ocean field experiment (SHEBA) [Persson et al., 2002; Uttal et al., 2002]. The ice drifted from 76° to 80°N along 165°W longitude in a region that was predominantly multiyear ice. This region is now mainly seasonal ice.

[21] Observed daily averaged values of incident solar irradiance (Fr) from SHEBA are plotted in Figure 4a. An annual solar cycle modulated by cloud cover variations is evident. During the SHEBA experiment, surface melt started on 29 May and ended on 17 August [Perovich et al., 2003]. A multiyear ice albedo evolution sequence based on SHEBA observations is plotted in Figure 4b [Perovich, 2002], along with a seasonal ice albedo evolution governed by the SHEBA onset dates.

Figure 4.

Time series of solar partitioning: (a) daily average incident solar radiation observed at SHEBA 1998, (b) albedo evolution from seasonal (red) and multiyear ice (blue), (c) daily solar heat input to the snow and sea ice, and (d) cumulative solar heat input to the snow and sea ice.

[22] The time series of incident solar radiation is combined with the albedos to determine the daily incident sunlight deposited to the ice Qs:

display math

where Δt = one day. From March through late May, there is no difference in heat input between seasonal and multiyear ice. Once melt begins, differences rapidly arise. Peak differences are in mid-June when the seasonal ice reaches its maximum solar input, which is a factor of two greater than multiyear ice input. The magnitude of this seasonal peak is due to a combination of large values of incident sunlight and a small albedo caused by surface flooding of melt ponds. The peak solar heat input to the multiyear ice does not occur until a couple of weeks later. Throughout the summer melt season the daily solar heat input to seasonal ice is consistently greater than to multiyear ice.

[23] The cumulative solar heat input is plotted in Figure 4d. The annual cycle of albedo is evident in the plot. Even though the incident solar irradiance is large in May, solar heat to the ice increases slowly due to the large albedo of the snow-covered surface. In June, as melt begins and the albedo decreases, solar heat input rapidly increases. The increase is particularly acute for seasonal ice. The solar heat input tapers off by August as the incident solar heat decreases and the albedo increases. For both seasonal and multiyear ice, approximately two-thirds of the total solar heat input to the ice occurred in June and July. From 1 March to 1 October, the total solar heat input to the multiyear ice was 893 MJ m−2 compared to 1235 MJ m−2 to the seasonal ice. Keeping the incident solar and the onset dates of melt and freezeup the same; the shift from multiyear to seasonal ice increased the solar heat input by 342 MJ m−2, a 38% increase and enough heat to potentially thin the ice by 1.02 m. This added heat will enhance internal melting and further thinning of the ice. The combination of more solar heat input to seasonal ice and thinner ice will also result in an increase of solar heat input through the ice into the upper ocean [Frey et al., 2011]. This increased sunlight will be available for warming the upper ocean and melting at the ice bottom, as well as for primary productivity.

4. Conclusions

[24] A general sequence of albedo evolution for seasonal ice was formulated from four years of field observations. The albedo of seasonal, shorefast ice evolves through seven stages; cold snow, melting snow, pond formation, pond drainage, pond evolution, open water, and freezeup. The details of the albedo evolution depend primarily on the timing of melt onset and freezeup. This albedo evolution provides a simplified method of estimating the changing albedo of seasonal ice during melt and the amount of solar heat input to the ice. During the melt season the albedo of seasonal ice is consistently smaller than multiyear ice. Thus the ongoing shift from multiyear ice to seasonal ice will increase the total solar heat input to the ice cover, enhance summer melting, and increase the amount of sunlight transmitted through the ice into the upper ocean. This transmitted sunlight will be available for warming the ocean, melting the bottom of the ice cover, and increasing the photosynthetically available radiation for primary productivity.


[25] Financial support was provided by the National Science Foundation Arctic Observing Network and Arctic System Science Programs. We thank T. C. Grenfell and Z. Courville for their assistance with the albedo measurements.

[26] The Editor thanks Mark Tschudi and Marcel Nicolaus their assistance in evaluating this paper.