C23B-01 INVITED
Snow on Arctic sea ice
Snow is a pervasive presence on Arctic sea ice for more than nine months of the year. Average snow depths on multiyear ice are about 0.3 – 0.4 m, while first year values are approximately half that. Snow begins to accumulate in summer, continues to deepen during the fall, is fairly stable throughout the winter, and then melts in late spring and early summer. There is considerable spatial variability in snow depth due to wind drifting, with depths ranging from bare ice to more than a meter. The snow depth is often closely tied to the ice type through the interaction of wind drifting and ice roughness. The snow cover and its variability have a profound impact on the thermal and optical properties of the sea ice cover. The influence of snow is enhanced by the contrast in physical properties between snow and ice. Because of its low density snow has a thermal conductivity roughly one-tenth that of sea ice. Deep snow can severely impede winter ice growth. Because of its abundant air-ice interfaces, snow has an optical extinction coefficient 10 times greater than sea ice. Light transmission through snow covered sea ice is small (less than 1%) and due to variations in snow depth can vary by orders of magnitude over horizontal distances of tens of meters. As snow melts there is a decrease in the albedo of the ice cover and an increase in the transmitted light. Changes in the timing of snow melt onset affect both the heat budget of the ice and biological activity in and under the ice. Earlier snow melt onset causes increases in ice ablation and in levels of photosynthetically available radiation under the ice
C23B-02 INVITED
Snow on Arctic sea ice: Updating the climatology and assessing its impact on ice ablation
The Arctic sea-ice cover is undergoing a major transformation with perennial ice extent decreasing at a rate of more than 8% per decade since 1979 and onset of seasonal ice formation delayed over much of the Arctic shelf seas. In this presentation we will (1) discuss the role of snow and its variability in summer surface ice melt, and (2) examine whether any significant changes are detectable in the depth or characteristics of the snow cover on Arctic sea ice based on measurements compiled for the time period from the 1960s onward. The importance of snow as a thermal insulator that constrains conductive heat flux and hence ice growth is well recognized. Less understood is its role in controlling surface melt and evolution of ice albedo. Studies of the evolution of melting landfast and drifting sea ice indicate that snow depth is a key variable that drives the initial evolution of surface melt ponds during summer. The latter in turn controls ice albedo over much of the melt season. Studies of ice melt indicate three primary controlling mechanisms, i.e. the role of snow in retarding ice surface warming, concealing surface melt water accumulation and in governing the formation of superimposed ice which in turn is critical to retention of melt water at the ice surface. Long-term studies at Barrow, Alaska suggest that through these mechanisms snow-depth variations are responsible for significant interannual variability in ponding and ice albedo. Based on field studies in the Alaskan Arctic and analysis of snow depth data compiled from icebreaker cruises, drifting stations and Russian "Sever" aircraft landings, snow depth patterns in the Arctic are examined, in particular in comparison with the climatology derived by Warren et al. (1999) from Russian drifting stations. Regional and temporal variations are discussed in the context of sea ice and precipitation changes as well as their potential impact on ice ablation patterns.
C23B-03
Fluid and electrical transport in sea ice
Brine flow through sea ice facilitates key geophysical and biological processes, such as snow-ice formation and nutrient replenishment, and is controlled by the fluid permeability of the ice. We shall discuss recent results in modeling and measuring the permeability of sea ice, and the related problem of investigating its electrical transport properties.
C23B-04 INVITED
Snow thickness over Antarctic sea ice from in situ measurements, aircraft and satellite data
The snow cover on sea ice is an important climate variable that affects the surface albedo, ocean-atmosphere heat flux, and fresh water flux to the ocean. Its thickness and distribution are affected by precipitation, wind, ice topography and the formation of snow ice resulting from snow loading and flooding. The latter is a process unique to the Antarctic that complicates efforts to measure or monitor the snow cover thickness on Antarctic sea ice. Recent results from Antarctica suggests that widespread flooding can occur over relatively short times scales, affecting surface roughness and passive microwave retrievals of snow thickness. In situ snow thickness measurements coupled with textural analysis of aerial photography indicate values for snow thickness over rough sea ice that are approximately 2-3 times larger than the AMSR-E snow thickness product over the same region. Recent data from sled and aircraft-based radar over Antarctic sea ice will also be presented.
C23B-05
Antarctic Sea Ice Thickness from Passive Microwave Retrievals of Snow Depth
Antarctic sea ice thickness retrievals from satellite altimeters are critically dependent on accurate estimates of snow depth, particularly so because of the relatively high snow loads and thin ice. We assess the accuracy of passive microwave estimates of snow depth and their utility for ice thickness determination by comparison with shipboard observations. Using a simple model of snow depth evolution, the predicted occurrence of snow-to-ice conversion is used as a proxy indicator for near-zero freeboard, allowing the possibility of sea-ice thickness estimation directly from satellite snow depth retrievals. The satellite data provide a good indicator of snow depth over broad scales in most cases and for all areas except for the East Antarctic sector. Using a modified algorithm for satellite snow-depth retrieval, we show that satellite snow depth can provide reasonable estimates of regionally-averaged ice thicknesses. Moreover, expected errors are likely to be less than those for altimetric methods. However, there is significant variability in the passive microwave snow depth signatures, both spatially and temporally, such that the uncertainty in ice thickness determination is comparable to the observed regional and interannual variability. This highlights the need for assimilation of models of snow-cover evolution to reduce uncertainty in satellite snow-depth retrievals.
C23B-06
Sea Ice Freeboards and Snow Depths in the Arctic Based on Satellite Laser and Radar Altimetry
Recent techniques make it possible to monitor the sea ice freeboards in the Arctic Ocean from satellite altimetry, and by use of models under assumption of isostatic equilibrium the freeboard can be converted into thicknesses. This is however, a crude assumption as the conversion factor depends on local sea ice and snow properties, e.g. densities and the depth of the snow cover. The main unknown factor so far is the snow depth. The laser detect the surface of the snow cover, whereas conventional radar altimetry is believed to penetrate down to the snow-ice interface, depending on the snow properties. In this study we present the methods for estimation of sea ice freeboards obtained from ICESat laser altimetry from a lowest-level technique, which on Arctic Ocean basin wide scale shows good correlation with the distribution of thick multi year sea ice extracted from QuikSCAT backscatter maps. In addition we will present the results of near coincident high-resolution airborne laser scanner measurements and radar altimetry from ENVISAT in the Fram Strait and north of Alaska to estimate the differences in freeboard heights and by that obtain an estimate of the snow depths.
C23B-07
A tentative climatology of the snow load on Arctic sea ice based on satellite
Having a firm grasp of the sea ice extent carries over to the understanding of poleward energy transport, atmospheric heat exchange and high-latitude ocean dynamics at large. One reason to investigate the snow load is the insulation against exchange of heat. Another, regarding the intrinsic value of remote sensing, is that snow constitutes the greatest unknown in sea ice altimetry. The properties of snow can modify how deeply into the snow-ice system the altimeter signal penetrates. While Cryosat views to the ice surface, Icesat views to the snow surface. The freeboard cannot be measured and converted to ice thickness properly without compensation for the thickness and density of the snow cover. To identify the satellite channels with most information on the scenery, we made the standard assumption that the inversion of measured brightness temperature to physical parameters is sufficiently linear to converge for Gauss-Newtonian iteration. An optimal estimation scheme has been adopted and the information content in the averaging kernel matrix scrutinized for the parameters at stake. The a priori covariance and initial guess on parameters was computed by feeding the snow-ice model Memls with ERA40 atmospheric reanalysis over a range of locations, winters, and type of ice as having grown from either scratch (first-year) or not (multiyear). Each of the currently flown passive sounders under consideration, the Advanced Microwave Scanning Radiometer (AMSR), the Advanced Microwave Sounding Unit (AMSU), and the Microwave Humidity Sounder (MHS), is modelled with a measurement error taken as the sum of sensitivity and accuracy prior to launch. Covariance between the channels has been neglected. Simulation of the actual measurement discretizes the snow pack into ten numerical layers to resolve the steep temperature gradient and applies the model Rttov to represent the air column. Snow is taken to be fresh and dry, a valid assumption until melt sets in, and the density of multi-year ice is imposed a fixed decrease above the waterline. The correlation length in ice that governs scattering shifts from water content (brine) to air bubbles after the first year. The optimal set of satellite channels has been chosen, in part, by minimizing the number of platforms involved and the jumps in frequency between them. These channels provide the basis on which we intend to retrieve a snow climatology that spans the past few years. Construction requires iteration against the assumption that either type of ice alone was covering the surface pixel and then engagement with a lookup table to meet with the brightness temperature observed. Comparison of the seasonal and regional variability is made to reanalysis and in situ measurement.
C23B-08
Monitoring Snow on ice as Critical Habitat for Ringed Seals
Ringed seals are the primary prey of polar bears, and they are found in all seasonally ice covered seas of the northern hemisphere as well as in several freshwater lakes. The presence of snow covered sea ice is essential for successful ringed seal reproduction. Ringed seals abrade holes in the ice allowing them to surface and breathe under the snow cover. Where snow accumulates to sufficient depths, ringed seals excavate subnivean lairs above breathing holes. They rest, give birth, and nurse their young in those lairs. Temperatures within the lairs remain within a few degrees of freezing, well within the zone of thermal neutrality for newborn ringed seals, even at ambient temperatures of -30° C. High rates of seal mortality have been recorded when early snow melt caused lairs to collapse exposing newborn seals to predators and to subsequent extreme cold events. As melt onset dates come earlier in the Arctic Ocean, ringed seal populations (and the polar bears that depend upon them) will be increasingly challenged. We determined dates of lair abandonment by ringed seals fitted with radio transmitters in the Beaufort Sea (n = 60). We compared abandonment dates to melt onset dates measured in the field, as well as estimated dates derived from active (Ku-band backscatter) and passive (SSM/I) microwave satellite imagery. Date of snow melt significantly improved models of environmental influences on the timing of lair abandonment. We used an algorithm based on multi-channel means and variances of passive microwave data to detect melt onset dates. Those melt onset dates predicted the date of lair abandonment ± 3 days (r 2 = 0.982, p = 0.001). The predictive power of passive microwave proxies combined with their historical record suggest they could serve to monitor critical changes to ringed seal habitat.