Time series observation of sea ice draft and velocity from Nares Strait between 2003 and 2012 provides new insights on the statistical properties of sea ice leaving the Arctic for the Atlantic Oceans. Median ice draft is 0.8 m, but it varies annually from 1.5 m in 2007–2008 to 0.5 m in 2008–2009. Probability density distributions of sea ice draft depend on location across the channel with thicker ice near Canada and thinner ice near Greenland. Nevertheless, sea ice motion stops seasonally due to arching land‐fast ice that spans the 30–40 km wide channel for up to 190 days per year such as during the 2011–2012 winter. In contrast, the 2006–2010 period exhibits a single ice arch lasting 47 days in April/May 2008. Hence sea ice statistics are weighted by space not by time, using sea ice velocities estimated from colocated velocity observations. Multiyear sea ice with drafts exceeding 5 m constitute between 9% (2003–2004) and 16% (2007–2008) of the observed sea ice. The probability g(D) of this thick, ridged, multiyear ice decays exponentially with draft D at an e‐folding scale D0 of 3.0 ± 0.2 m. The trend of D0 with time is statistically indistinguishable from zero. This observation suggests a steady export of multiyear sea ice at decadal time scales. We speculate that our observations document the draining of the last reservoir of thick ice from the Arctic Ocean found to the north of Ellesmere Island and Greenland.

No abstract

No abstract

A semianalytic two‐dimensional model is used to analyze the interplay between the different forces acting on density‐driven flow in high‐latitude channels. In particular, the balance between wind stress, viscous forces, baroclinicity, and sea surface slope adjustments under specified flux conditions is examined. Weak winds are found not to change flow patterns appreciably, with minimal (<7%) adjustments to horizontal velocity maxima. In low‐viscosity regimes, strong winds change the flow significantly, especially at the surface, by either strengthening the dual‐jet pattern, established without wind, by a factor of 2–3 or initiating return flow at the surface. A nonzero flux does not result in the addition of a uniform velocity throughout the channel cross section, but modifies both along‐channel and cross‐channel velocities to become more symmetric, dominated by a down‐channel jet centered in the domain and counter‐clockwise lateral flow. We also consider formulations of the model that allow adjustments of the net flux in response to the wind. Flow patterns change, beyond uniform intensification or weakening, only for strong winds and high Ekman number. Comparisons of the model results to observational data collected in Nares Strait in the Canadian Archipelago in the summer of 2007 show rough agreement, but the model misses the upstream surface jet on the east side of the strait and propagates bathymetric effects too strongly in the vertical for this moderately high eddy viscosity. Nonetheless, the broad strokes of the observed high‐latitude flow are reproduced.