Evidence collected from both newly developed and long-established environmental observing systems is used to develop a coherent picture of the fluctuations in the atmosphere and ocean, which have tended to shift the Beaufort Sea climate away from once familiar seasonal patterns. We utilize in situ and satellite-derived ocean temperature measurements, sea-ice concentration and motion, wind and other meteorological variables extracted from NNR, Mackenzie River run-off and daily radiation climatology.

PMEL Wave Gliders were deployed in the Arctic for the first time in the summer of 2011. On 2 August an 11°C surface layer was detected over the continental shelf break, along with a sharp temperature contrast exceeding 8°C between the surface and 6 m depth (Fig. 3a). A review of MODIS SST and true-colour satellite imagery showed that the Wave Gliders had encountered the boundary of a large Mackenzie River discharge plume. In situ temperature observations from the Wave Gliders are consistent with satellite-derived SSTs from MODIS, but more importantly they provide uninterrupted high-resolution monitoring of an undisturbed surface layer at depths not seen by satellite, and above the reach of typical oceanographic moorings. Together the two sources indicate a 6–12°C surface layer associated with the plume that was at least 6 m thick and approximately 74 000 km² in extent. A hydrographic transect along 140°W obtained on 26–27 July from the Canadian ice-breaker CCGS Louis S. St.-Laurent confirmed that the 6°C isotherm was generally ≤8 m deep and that it coincided with the strong salinity stratification that defines the base of the plume and summer polar mixed layer (Fig. 3b).

By the end of July 2011, an excess heat content of 2.0–2.5×10¹⁹ J had already accumulated in the surface layer associated with the Mackenzie plume (relative to a baseline temperature of −1.8°C). A well-stratified surface layer in the Beaufort Sea can warm very rapidly, especially on cloud-free days in late June and July when direct radiation can exceed 300 W m⁻². A striking example is provided by a rare 23-day sequence of MODIS images from July 2008 that shows heating rates in excess of 0.5°C day⁻¹ (see the Supplementary File or go directly to the animation). This rate also implies that heating was distributed over an approximately 12-m-thick surface layer, consistent with plume depths observed along 140°W in 2011.

Entrainment of the Mackenzie River discharge under moving pack ice early in the summer can also help set the stage for rapid warming offshore later in the season. MODIS imagery from early June 2011 shows warm and discoloured water flowing under the westward-drifting ice pack. Given that freely moving sea ice amplifies the effect of wind stress on the underlying water layer (Pite et al. 1995), it is certain that river water was carried along under the ice. Buoy data indicate that the ice pack was moving westward at a mean rate of 15.1 cm s⁻¹ (maximum 28.6 cm s⁻¹; Rigor 2002). Counting ca. 55 km³ of fresh water impounded under coastal fast ice (Macdonald & Carmack 1991; Dean et al. 1994; Dunton et al. 2006), half of the total summer discharge (129 km³ of 246 km³ in 2011) had already entered the Beaufort Sea by mid-July. Although sea-ice melt obviously contributes fresh water to the surface layer, there was enough Mackenzie River water discharged under the ice cover of the Beaufort Sea early in the season to create a plume of the dimensions described here of an average thickness of 6–8 m and salinity of 20 psu.

Heat supplied directly by the Mackenzie River around break-up in late May and June melts and mobilizes the sea ice near the delta (Macdonald 2000), but as more area opens up offshore due to the combined effect of early melt and wind, solar radiation provides most of the heat in the surface layer. With about 80% more ice-free area compared to the 1981–2010 climatology, the Beaufort Sea south of 76°N has been absorbing much more shortwave radiation than in the past—an average of 5.1×10²⁰ J (2007–11) compared to 2.8×10²⁰ J during the reference period. This is roughly equivalent to the amount of heat advected through the Bering Strait annually (Woodgate et al. 2010), and is enough to melt about 1.5×10⁶ km² of 1-m-thick sea ice each year. How some of this heat is stored and later released in the autumn is modulated by the Mackenzie plume, as can be inferred from the patterns in the MODIS SST composites in Fig. 1. This warm water contributes directly to the formation of large mean surface temperature anomalies in the Beaufort Sea around freeze-up in September and October (Fig. 4).

Large fluctuations in the summer wind field over the Beaufort Sea provide a common thread linking sea-ice distribution (and ice-albedo feedback), surface stratification amplified by the westward advection of the Mackenzie River plume, higher ocean temperatures and the formation of persistent temperature anomalies that last well into the freeze-up period in the autumn. Easterly winds have been stronger and more frequent in this region since 2007 (Fig. 2a, c). These winds are linked to anomalous anticyclonic circulation around an intensified Beaufort High (Ogi & Wallace 2012; Fig. 2d), and the pattern is present throughout the entire depth of the troposphere, reaching a maximum at 300 hPa (Fig. 5). Thus, even though the maximum wind anomaly appears over the Beaufort Sea, it is more than a regional or atmospheric boundary layer phenomenon.

The perception that a new normal climate is emerging in the Beaufort Sea is rooted in the combined impacts of large-scale anthropogenic warming, represented most clearly by the diminishing Arctic ice pack, and, in the Beaufort Sea, the immediate effects of anomalous easterly winds that promote additional sea-ice loss and rapid warming of the surface ocean, especially as amplified by the presence of a well-stratified surface layer produced by an extensive Mackenzie River plume. Relatively strong easterly winds have occurred on occasion over the Beaufort Sea in the past, notably in 1997–98, but there is no period in NNR equivalent to the most recent run of years. The easterly wind anomaly that has occurred lately is part of a deep Arctic-wide pattern of circulation around an intensified Beaufort High.

In the absence of easterly winds and a thinner sea-ice cover, the conditions for rapid surface warming of the Beaufort Sea are not established, and the sea and surrounding area remain relatively cool. When winds are westerly (as in 2002 and 2003, for example, and more often in the past) the ice pack stays closer to the coast, and the Mackenzie River plume is constrained to the continental shelf area, or drifts towards the east into the Canadian Arctic Archipelago (Macdonald & Yu 2006; Yamamoto-Kawai et al. 2009), and as shown by the example in Fig. 1c. In former times, when the ice pack was heavier, it would have been less mobile and less prone to blowing out, and would likely have further limited the movement of Mackenzie water off the continental shelf and to the west. While our historical knowledge of sea-ice conditions in the Beaufort Sea is incomplete, no evidence of a major reduction like 2007 and 2012 in the past 100–150 years has yet come to light (Mahoney et al. 2011; DMI & NSIDC 2012).

The summer of 2011 set the record for the seasonal mean NNR zonal wind anomaly over the Beaufort Sea (−4.4 m s⁻¹/ − 2.6 SD at 925 hPa). These winds caused the advection of both the sea ice and the Mackenzie River plume to the west. The large ice-free area that was created warmed rapidly due to positive ice-albedo feedback supported by strong stratification of the surface layer, in this case due to the plume, as recorded by the PMEL Wave Gliders and other observing systems. Anomalous heat content persisted into the autumn and contributed to the formation of very large surface temperature anomalies around freeze-up; a common occurrence over the past six years here and in other Arctic marginal seas.

It is still unclear how the additional heat input into the Arctic Ocean in recent years is ultimately partitioned (Perovich et al. 2007). Some of the heat absorbed in summer is stored in the ocean and later enters the deepening winter mixed layer where it can continue to melt sea ice during the winter (Steele et al. 2011; Jackson et al. 2012). Another fraction of this heat is released in the autumn, when it may perturb the large-scale atmospheric circulation. Extreme weather events at mid-latitudes associated with more persistent weather patterns may be set up in this way (Overland & Wang 2010; Francis & Vavrus 2012).

Variants of the fast regional-scale processes that have contributed to the appearance of the “new normal” climate in the Beaufort Sea are probably typical in other Arctic marginal seas, and this may help explain why current climate models under-project the observed rate of sea-ice loss and environmental change in the Arctic. It is likely that gradual large-scale warming associated with anthropogenic forcing is relatively well represented (Mahlstein & Knutti 2012), but rapidly evolving regional interactions between the atmosphere, sea ice and ocean of the type described here may not yet be accounted for properly even by the latest models. In particular, the wind patterns seen recently appear to be well-suited to both remove multiyear ice from the Canadian Arctic Archipelago, where it has been historically concentrated into a warmer Beaufort Sea, as shown by the supplementary animation and consistent with early CryoSat-2 results reported by Laxon et al. (2013), and to accelerate the transport of sea ice from the western Arctic towards Fram Strait (e.g., Ogi & Wallace 2012).

The recent run of anomalous easterlies has undoubtedly contributed to the acceleration of sea-ice loss and surface warming in the Beaufort Sea and is likely to have significantly impacted the broader ecosystem, but can the arrival of a “new normal” climate be declared? The reason the summer Beaufort High has strengthened is largely unexplained (Ogi & Wallace 2012), and as we have shown, this is intimately linked to what has happened in the Beaufort Sea over the past few years. Overland et al. (2012) suggest that the anomalous easterlies are part of a larger climate pattern over North America previously recognized as the Arctic Dipole. However, until this question is answered uncertainty remains over the possibility of continued persistence. Nevertheless, it is clear that the overall decline of the Arctic sea ice has left the Beaufort Sea more susceptible to rapid warming than it has been in the past, and this has also magnified the potential for additional regional and far-field impacts that are only now becoming apparent. If the wind patterns seen lately are reinforced by changes in long-term climate forcing then the Beaufort Sea, already warm and mostly ice-free by late summer, will tend to remain that way.