In those parts of the Arctic where the sea ice decrease is fastest, the sea ice retreat has in the recent years been nearly two months earlier and subsequent advance more than a month later than in 1979-1980, resulting in a three-month longer summer ice-free season (Stammerjohn et al., 2012). From the point of view of the atmosphere, another important variable is the length of the melt season, when the snow/ice surface temperature is fixed to 0◦C (with occasional slightly lower values). The spring onset of snow melt on sea ice reduces the snow albedo and initiates the positive feedback effect. An early melt onset tends to result in an early generation of open water areas, which favors heat accumulation in the upper ocean. Maksimovich and Vihma (2012) concluded that in the period of 1989-2008, local 20-year snow melt onset trends averaged over the central Arctic Ocean were towards earlier melt by 9 days per decade. The earlier melt onset has been due to higher air temperatures in spring (Belchansky et al. 2004) and, above all, increased downward longwave radiation (Maksimovich and Vihma, 2012). In the autumn, ice cannot form over open water areas until the ocean mixed layer loses its heat. As a result, the freeze-up starts later, which contributes to sea ice thinning in the following year (Laxon et al., 2003; Notz, 2009). The autumn freeze-up has become later everywhere in the Arctic (except in the Sea of Okhotsk), but the change has been smaller than in the spring melt onset. For autumn, the largest changes have been observed in the Laptev/East Siberian Seas and Chukchi/Beaufort Seas with trends between 6.9 and 8.4 days per decade (Markus et al. 2009).
Reasons for the Arctic sea ice decline have been addressed in numerous studies, with Meier et al. (2011), Stroeve et al. (2012) and Polyakov et al. (2012) providing up-to-date summaries. During the recent decades the temporal and spatial changes in the atmospheric circulation have been essential factors directly affecting the sea ice decline (Zhang et al., 2008, Overland and Wang, 2010). Overland and Wang (2010) concluded that a shift to a more meridional atmospheric pattern, the Arctic Dipole (AD), over the last decade contributed to recent reductions in summer Arctic sea ice extent. According to Polyakov et al. (2012), particularly the atmospheric thermodynamic forcing has played an increasingly important role. The advection of moist air masses from lower latitudes to the Arctic, the associated increase in downward longwave radiation, and its dependency on the cloud fraction, base height, and phase (liquid water or ice) have been addressed by, e.g., Francis and Hunter (2007) and Kapsch et al. (2013). As the ice thickness has decreased, the ice pack has become increasingly sensitive to the ice-albedo feedback (Perovich et al., 2008; Serreze et al., 2009). This is seen, among others, in more negative inter-annual trends in sea ice extent for later summer, when the albedo feedback lasts for a longer period, than for early summer (Stroeve et al., 2012). Furthermore, Screen and Simmonds (2012) concluded that the decline in summer snowfall has likely contributed to the thinning of sea ice over recent decades. The change from snowfall to rain has strongly increased the fraction of bare sea ice, which has a much lower albedo than snowcovered sea ice.
In addition to the decreasing trend in September ice extent, its inter-annual variability has strongly increased during recent years. Stroeve et al. (2012) argue that this is partly due to the thinning of the ice, as thin ice is more sensitive to variations in the atmospheric conditions in summer (Holland et al. 2008). Besides thermodynamic conditions, thinner ice is more sensitive to wind forcing, seen as increased ice drift speeds (Spreen et al., 2011; Vihma et al., 2012). Ice export out of the Arctic was large in late 1980s and early 1990s (Rigor and Wallace, 2004), but according to Polyakov et al. (2012) it is unclear if it has played a significant role in the (multiyear) ice mass budget over the past decade.
Numerous studies have addressed the reasons for the record minimum September sea ice extents in 2007 and 2012. In 2007, a key driver of the record was an atmospheric pattern allowing large heat transport from the Pacific sector (Kay et al. 2008; L’Heureux et al. 2008; Stroeve et al. 2008; Perovich et al. 2008; Schweiger et al. 2008; Ogi et al. 2010; Graversen et al., 2011), but also an increased oceanic heat transport via the Bering Strait played a role (Woodgate et al., 2010). Kay et al. (2008) stressed the importance of large shortwave radiation in summer, whereas Graversen et al. (2011) concluded that the increased downward longwave radiation in spring was a major factor. The extent and area of Arctic sea ice reached a new record in September 2012 (Parkinson and Comiso, 2013). This was partly due to a record-strong storm that occurred over the Beaufort Sea in early August (Simmonds and Rudeva, 2012). The storm caused disintegration of a large part of the central Arctic sea ice pack. This part entirely melted, and the disintegration enhanced the melt of the remaining main ice pack (Parkinson and Comiso, 2013). According to a modeling study by Zhang et al. (2013), the strong melt was largely due to a quadrupling in bottom melt, caused by storm-driven enhanced mixing in the ocean boundary layer. Zhang et al. (2013) argued that a record minimum ice extent would have been reached in 2012 even without the storm, as by early August the ice volume had already decreased ~40% from the 2007-2011 mean.
The fast climate warming in the Arctic has accelerated the sea ice decline, but simultaneously the sea ice decline has amplified the warming (e.g., Kumar et al., 2010). Among the many processes that have been considered responsible for the Arctic amplification of climate warming, the sea ice decline is associated with the snow/ice albedo feedback effect (Flanner et al., 2011; Graversen and Wang, 2009; Serreze and Barry, 2011), increased heat loss from the ocean (Screen and Simmonds, 2010a,b), the water-vapour and cloud radiative feedbacks (Sedlar et al., 2011), and the small heat capacity of the shallow stably stratified boundary layer (Esau and Zilitinkevich, 2010). It should be noted that there is no general consensus on the relative importance of various factors responsible for the Arctic amplification and sea ice decline. According to model experiments, the amplification occurs also in a world without sea ice (Alexeev et al., 2005).
