Long-term variation of sea ice and its response to thermodynamic factors in the Northwest Passage of the Canadian Arctic Archipelago

Sea ice conditions in the Canadian Arctic Archipelago (CAA) play a key role in the navigation of the Northwest 10 Passage (NWP). Based on the observed and simulated sea ice concentration and thickness data, we studied the temporal and spatial characteristics of sea ice from 1979 to 2017 in the NWP of the CAA and evaluated the sea ice conditions along the southern and northern routes of the NWP. Against the background of the rapid retreat of Arctic sea ice, the 39-year observed sea ice concentration of the NWP exhibited a relatively large decreasing trend in summer and fall, while heavy sea ice conditions were maintained in winter and spring, with a slight increasing trend. Consistent with Arctic sea ice, the sea ice 15 extent in the NWP displayed a decreasing trend of -2.34%/10a, with its minimum occurring in 2012. The sea ice thickness in most subregions of the NWP showed a decreasing trend, with the exception of Lancaster Sound. The decreasing trend of sea ice thickness in the NWP was estimated to -0.16 m/10a. Based on the sea ice concentration and thickness, however, the sea ice conditions were heavier along the northern route than the southern route. This study considered both of these routes, and we selected and evaluated more specific pathways. The correlation results between the sea ice and atmospheric and oceanic 20 thermodynamic factors in the NWP suggested that the thermodynamic factors had a greater impact on sea ice in the summer and fall, and the variations of sea ice concentration were more closely correlated with the thermodynamic factors than sea ice thickness. The sea surface temperature (SST) had a higher correlation with sea ice concentration than surface air temperature (SAT), while SAT exhibited a higher correlation with sea ice thickness than SST. The residual amount of sea ice concentration and thickness in the fall, associated with the fall SAT and SST, contributed to the formation of sea ice in the following winter 25 and spring.


85
The simulated sea ice thickness data were from the Arctic Ocean-Finite Volume Community Ocean Model (AO-FVCOM) (Chen et al., 2009;Chen et al., 2016;Gao et al., 2011;Zhang et al., 2016a;Zhang et al., 2016b) for the period 1979-2017, with a horizontal resolution as high as 1 km in the CAA. These high-resolution data could be used to study the spatial distribution, as well as the seasonal and long-term variation characteristics, of sea ice thickness in the NWP. The AO-FVCOM was configured with a nonoverlapped triangular grid that could better reproduce complex shorelines and topography in the 90 NWP. In the vertical resolution, we used a hybrid terrain-conforming coordinate with 45 layers. The driving forces included tidal forcing with eight major constituents (M2, S2, N2, K2, K1, P1, O1, and Q1), surface wind stress, net heat flux at the surface plus shortwave irradiance in the water column, surface air pressure gradients, precipitation (P) minus evaporation (E), and river discharges. We conducted a detailed validation and comparison of the AO-FVCOM sea ice thickness data with the Arctic multisource sea ice thickness data, including ICESat-2 satellite data, field drill-hole observations, airborne electromagnetic 95 observations, and sea ice station data (Zhang et al., 2016b). The results revealed that the sea ice thickness data of the AO-FVCOM well captured the spatial distribution, as well as seasonal and interannual variation characteristics, of Arctic sea ice thickness data. In addition, in a comprehensive comparison with six other sea ice models, the AO-FVCOM achieved better sea ice thickness results in terms of smaller data bias and higher correlation with the multisource sea ice thickness observational data.

Atmospheric and oceanic thermodynamic data
In this study, we used the monthly mean SAT and SST reanalysis data from the European Centre for Medium-Range Weather Forecasts (ECMWF, https://cds.climate.copernicus.eu/) (Hersbach et al., 2019) to explore the impacts of atmospheric and oceanic thermodynamics on sea ice conditions in the NWP. The time period of the SAT and SST data was 1979-2017.

Spatial distribution of sea ice concentration
From 1979 to 2017, the sea ice concentration over the entire NWP displayed significant spatial distribution differences in different months ( Figure 2). The NWP was covered by high sea ice concentrations during spring and winter and lower concentrations in summer and fall. The domain was covered completely by a high sea ice concentration close to 1 (i.e., 100% 110 coverage) from December to April, when both routes were closed by sea ice and would not be navigable by ships lacking ice- The sea ice concentration varied in the four seasons. During winter, most subregions in the NWP displayed a slight increasing trend. During spring, the sea ice concentration in most subregions of the NWP had a slight increasing trend, which was similar to winter, with the exceptions of the Amundsen Gulf, M'Clure Strait, and Lancaster Sound gateways. In these subregions, the sea ice concentration exhibited decreasing trends, the maximum of which was -1.17%/10a in the Amundsen Gulf.
In contrast, the sea ice concentration strongly decreased in summer and fall, particularly in the southern routes. In summer, the 130 largest variation was located in the Lancaster Sound (-11.26%/10a) and the smallest rate was in the Prince of Wales Strait (-2.41%/10a). In fall, the variation peak shifted to the Amundsen Gulf (-10.44%/10a), and the smallest rate was still located in the Prince of Wales Strait (-2.98%/10a).

Temporal variation of sea ice extent
To further understand the sea ice conditions in the NWP, we also studied the variation of the sea ice extent. The yearly mean 135 sea ice extent exhibited significant differences among the 10 subregions of the NWP (Figure 4). The mean sea ice extent in the NWP was 4.99 × 10 5 km 2 (94.41% of the total area) from 1979 to 2017. The sea ice extent displayed a significant decreasing trend during this period, at a rate of -2.34%/10a. The minimum sea ice extent occurred in 2012 (4.53 ×10 5 km 2 , 85.64% of the total area), which was the same year that the minimum sea ice extent occurred in the entire Arctic Ocean. The seasonal cycle the NWP from June to October. In addition, the different impacts of SAT and SST on sea ice conditions in the subregions were examined and discussed by analyzing the seasonal correlations and comparing the interannual variations.
The seasonal spatial distributions of sea ice concentration varied significantly. In winter and spring, the entire NWP was fully 335 covered by a sea ice concentration that was close to 1. During summer and fall, the sea ice concentration decreased differently in different subregions. In general, the sea ice concentration was greater along the northern route than the southern routes. The 39-year observed sea ice concentration of the NWP exhibited relatively large decreasing trends in summer and fall, whereas the heavy sea ice conditions were maintained in winter and spring, with a slight increasing trend, which differed from the decreasing sea ice concentration trend in all four seasons across the entire Arctic. The sea ice extent in the NWP displayed a 340 significant decreasing trend (-2.34%/10a), with the minimum occurring in 2012 (4.53 × 10 5 km 2 , 85.64% of the total area), a pattern that was consistent with the sea ice extent over the entire Arctic.
From 1979-2017, the interannual variation of sea ice thickness decreased at a rate of -0.16 m/10a, reaching its minimum in 2016 (0.93 m). The multiyear mean seasonal sea ice thickness in the NWP increased from October to April, with its maximum of 2.26 m occurring in April and its minimum of 0.26 m occurring in September. In general, the sea ice thickness along the 345 northern route was greater than the thicknesses along the southern routes, while the decreasing trend of the northern route was stronger. In the most areas of the NWP, with the exception of Lancaster Sound, the sea ice thickness exhibited a decreasing trend, which was larger in summer and fall and smaller in winter and spring.
Based on the sea ice concentration and thickness distribution from June-October and the probability of light sea ice, which was defined as the percentage of lighter sea ice conditions compared with the climatological means, we evaluated additional 350 specific pathways along the southern routes and the northern route. According to sea ice concentration, the more specific pathway for the southern routes was the center of the Amundsen Gulf-Coronation Gulf-south side of the Queen Maud Gulfcenter of Prince Regent Inlet-south side of Lancaster Sound. For the northern route, the more specific pathway was the center of M'Clure Strait-north side of Viscount Melville Sound-center of Barrow Strait-south side of Lancaster Sound. Considering sea ice thickness, the more specific pathway for the southern routes was the north side of the Amundsen Gulf-north side of 355 the Coronation Gulf-south side of the Queen Maud Gulf-north side of Prince Regent Inlet-north side of Lancaster Sound. For the northern route, the more specific pathway was to travel along the north side from M'Clure Strait to Lancaster Sound.
As for the impacts of atmospheric and oceanic thermodynamic factors on sea ice conditions, the correlations between sea ice and thermodynamic factors were higher in summer and fall, and the sea ice concentration generally exhibited a higher correlation than the sea ice thickness. The SST had a higher correlation with sea ice concentration than SAT, whereas the SAT 360 had a higher correlation with sea ice thickness than SST. The sea ice thickness in the winter and spring was dominant by the residual of sea ice concentration and thickness, which was affected by SAT and SST in the previous fall.