Risk assessment of ice-class-based navigation in Arctic: a case study in the Vilkitsky Strait

Over the past four decades, the decrease in Arctic sea ice has driven significant growth in vessel traffic through the Arctic passages. A precise and quantitative sea ice risk assessment would be the cornerstone of route planning for Arctic ships. Taking the chokepoint of Arctic Northeast Passage, the Vilkitsky Strait, as an example, the temporal and spatial characteristics of ice conditions in the strait were analyzed based on simulated ice thickness and observed sea ice concentration data from 2012 to 2021. Additionally, navigation risk in the strait was assessed based on the Polar Operational Limit Assessment Risk Indexing System (POLARIS). The results showed that the strait experienced 100% sea ice coverage from December to May, peaking in thickness of nearly 2 m in May, receding starting in June, and presenting ice-free passages by August and September. A significant interannual variability is evident in the timing of sea ice melting and freezing. Moreover, the average navigability of the strait was 365 days for vessels with an ice class of PC3, 72 days for PC6 and fewer than 64 days for those below B1. Remarkably, in 2013, 2014, and 2021, vessels below PC5 had less than 30 navigable days in the strait.


Introduction
The Arctic has experienced intensified warming and extensive sea ice retreat in the recent years [1,2].Since November 1978, a series of microwave remote sensing satellite has been used to monitor the Arctic sea ice concentration and its spatial extent.The satellite dataset reveals that the ice coverage in every month over the past 40 years has shown a pronounced negative trend [3].Taking the month with the smallest ice extent, September, as an example: the 15 lowest September ice extents recorded from 1979-2021 all occurred in the recent 15 years.The average September sea ice extent has decreased from 7.06 × 106 km 2 in 1981-1990 to 4.57 × 106 km 2 in 2011-2020, with an average decrease rate of -0.81± 0.13 × 105 km 2 /a [4].
The decrease trend of Arctic sea ice has rendered the frozen Arctic shipping routes commercially navigable.With the ever-increasing demand for global maritime trade and the capacity limits of traditional routes [5], the Arctic Passages, offering shorter travel distances, have emerged as a new maritime option linking Asian, European, and American markets.Typically, these Arctic routes are categorized into the Northeast Passage, the Northwest Passage, and the Central Route.The Northeast Passage refers to the Arctic route situated along Russia's coast, spanning westwardly through areas such as the Bering Sea, Chukchi Sea, De Long Strait, East Siberian Sea, New Siberian Islands, Laptev Sea, Vilkitsky Strait, Kara Sea, and the Norwegian Sea.Notably, the Vilkitsky Strait, positioned between the Severnaya Zemlya archipelago and the Taymyr Peninsula, serves as the shortest connector between the Laptev Sea and the Kara Sea.The strait stretch approximately 148 km along the 78° latitude, with a minimal depth of 37 meters, which is often termed the "throat" of the Northeast Passage.
In the Vilkitsky Strait, the water was covered by ice in the most time of one year and the average ice thickness is around 1.0 to 1.5 meters between January and April [6,7].Many studies indicate that the ice conditions in the Vilkitsky Strait are intricate, marking it as a key region determining the navigability of the Northeast Passage.Ji et al. utilized multi-source satellite remote sensing data for cross-calibration research [8], subsequently inferring the sea ice concentration in the Northeast Passage.By employing a 50% concentration threshold method (meaning regions with less than 50% concentration are deemed navigable), they analyzed the navigability of the Northeast Passage from 2002-2016.Their findings underscored substantial annual variations in the ice conditions of the Vilkitsky Strait, emphasizing the complexity therein and stressing the need for heightened attention to this area in future navigability studies of the Arctic route.Li et al. used multi-source sea ice concentration products released by the University of Bremen to study the connectivity of the Vilkitsky Strait from 2002-2013.Their results similarly highlighted significant interannual variations in the ice conditions of the strait.Yu et al. employed multi-source satellite data to analyze the spatial variations of ice conditions for four typical routes in the Northeast Passage from 1979-2019 [1].Their analysis revealed that the navigability of different route sections varied in sensitivity to parameters like air temperature and wind direction, with wind direction being the paramount factor influencing the navigability of the Vilkitsky Strait, whereas the Kara Strait was more sensitive to temperature.
From both geography and ice condition, the Vilkitsky Strait stands as the linchpin in determining the navigation risks of the Northeast Passage.This paper proposes to utilize the high-resolution thermodynamic model HIGHTSI to simulate the sea ice thickness in and around the Vilkitsky Strait.Combined with AMSR-2 sea ice concentration data, the navigation risks for various ship types in the region from 2012-2021 will be assessed, following the Polar Operational Limit Assessment Risk Index System (POLARIS) standards.This comprehensive approach will synergize remote sensing data with model data, offering a quantitative and visual exploration of the navigational risk indices for the Vilkitsky Strait across different seasons and years.This endeavor aims to provide technical support for China's Arctic navigation planning and serve as a reference for establishing a quantitative risk assessment system for the Arctic route in the future.

Study area
Northeast Passage (Figure 1).The area encompasses the Severnaya Zemlya archipelago and its surrounding waters, with a longitude range of 80°-120°E and a latitude range of 75°-82°N.The Vilkitsky Strait separates the Severnaya Zemlya archipelago from the Eurasian mainland.It represents a pivotal strait connecting the Kara Sea and the Laptev Sea, providing the shortest maritime route through the aforementioned waters.However, it remains free from sea ice only briefly during the summer.To elucidate the navigational risks and their variations within the Vilkitsky Strait, this paper further subdivides the area into three smaller regions, as shown in Figure 1: (1) the western area of the strait (77°-78°N, 94°-99.5°E),(2) the strait region itself (77.5°-78.5°N,99.5°-105°E), and (3) the eastern area of the strait (77.5°-78.5°N,105°-110.5°E).This paper emphasizes the navigational risk assessment for the strait from 2012 to 2021.

Sea Ice Concentration Dataset
In this paper, we use the high-resolution AMSR-2 sea ice concentration data developed by the University of Bremen, Germany.It was generated using the ASI algorithm, with a spatial resolution of 6.25 km and covers the time span from July 2012 to December 2021 with a time resolution of 1 day [9].Sea ice concentration, a widely recognized variable for representing sea ice conditions, denotes the proportion of sea ice within a unit area.It's typically represented by values ranging between 0 and 1, where 0 indicates no ice and 1 signifies complete ice coverage.

HIGHTSI model HIGHTSI (High Resolution Thermodynamic Snow and Ice Model
) is a one-dimensional thermodynamic model that simulates the evolution of snow and sea ice [10].Previous studies have demonstrated its effectiveness in modeling the seasonal variations of Arctic sea ice temperature and thickness [11,12].The model incorporates the downward shortwave and longwave radiation parameterization schemes from Shine [13] and Prata [14], respectively, while considering parameters like cloud cover, air temperature, and humidity.The surface albedo parameterization employs the scheme by Briegleb et al. [15], while the penetration effect of solar radiation in snow and ice is based on the approach by Grenfell and Maykut [16].Furthermore, this model incorporates atmospheric stratification effects when calculating sensible and latent heat fluxes using input wind speed, air temperature, humidity, and modeled ice surface temperature.When simulating sea ice thickness, the atmospheric forcing data is sourced from the European Centre for Medium-Range Weather Forecasts (ECMWF) global fifth-generation atmospheric reanalysis daily data (ERA5).It's worth noting that HIGHTSI computes the thermodynamic growth of theoretical ice thickness without accounting for regional ice dynamic processes.

POLARIS
The Polar Operational Limit Assessment Risk Indexing System (POLARIS) is a methodology developed by the International Maritime Organization (IMO), combining expertise from various Arctic marine traffic management institutions, including the Arctic Ice Regime Shipping System (AIRSS) of Canada and the issuance criteria for the Russian Arctic Passage's Ice Certificate.POLARIS is designed to evaluate navigational risks in icy regions [17].
Evaluations within POLARIS are based on ice conditions and ice classes, quantifying risks via the Risk Index Outcome (RIO).The RIO formula is given as: RIO = ∑ (  ×   )  =1 (1) Where n is the number of sea ice types in the region, Ci is the concentration of the i-th type of ice, and RIVi is the Risk Index Value (RIV) of sea ice of type i, which is determined by the sea ice type and the ice class of the vessels.In this paper, the sea ice type will be determined by the simulated sea ice thickness.
Depending on the RIO value and the vessel's ice class, POLARIS classifies vessels into three risk categories as shown in Table 1.For risk level 1, vessels should reduce their speed accordingly, with Table 2 providing the IMO recommended speeds for each ice class.For risk level 2, navigational planning should ideally avoid such conditions.If encountered, vessels must adhere to the Polar Waters Operational Manual (PWOM) procedures, reconsidering their routes, reducing speed, and conducting other risk-mitigating measures [18].

Ice class
Recommended speed limit (knots) Below PC7 3 In this paper, we utilize the sea ice thickness results from the thermodynamic simulation, determining the ice types according to the World Meteorological Organization (WMO) naming rules for sea ice [19].Thereafter, the study employs the IMO definition for RIV [17] to establish the RIV values for each ice condition and class, combined with concentration data to compute the RIO values, thereby assessing the navigational risks within the study area.

Changes in Sea Ice Concentration
According to the AMSR2 sea ice concentration, the distribution of monthly sea ice concentration from 2012 to 2021 around the Vilkitsky Strait was calculated.As depicted in Figure 2, there is a clear seasonal variation in the sea ice concentration in the Vilkitsky Strait and its surrounding areas.The study region is primarily covered by sea ice from November to the subsequent April, with a concentration above 0.8, and the entire area's ice concentration nears 1.0 from January to April.Starting from May to June, the ice near the Severnaya Zemlya archipelago and the eastern region of the Vilkitsky Strait (marked as area (3) in Figure 1) begins to melt, with the melt zone progressively expanding.In July, for the first time, the sea ice concentration in the eastern Vilkitsky Strait drops below 0.15, indicating the theoretical appearance of open water.In August, the open water area accounts for about 53% of the total study area, while in September, it reaches its maximum, approximately 77%.In October, the concentration rapidly increases from north to south on the eastern side of the strait, and by November, the Vilkitsky Strait and both its flanks are blanketed with dense ice, with concentrations above 0.7.By December, the sea ice concentration in the study area escalates to above 0.75.` Figure 3 shows the seasonal cycle of spatial mean sea-ice concentration for study area during 2012-2021.Due to the prevalent ice cover in the winter, there's not much change in sea ice concentration between winter and spring.From December to the subsequent May of each year, the spatial average fluctuates between 0.68 to 0.8.The maximum monthly average concentration appears from January to March, approximately 0.77.The study area enters the melting period from May to June, and the concentration rapidly decreases, reaching a minimum of about 0.09 in September.On the other hand, the variation trend of sea ice concentration in winter is relatively consistent among different years, with a standard deviation below 0.02 from January to April.The standard deviation is higher in summer and autumn, with maximum values appearing in October and November, approximately 0.16 and 0.12, respectively.This is because the sea ice in this area generally begins to freeze rapidly in October, but the start time and freezing speed vary from year to year, leading to larger standard deviations in October and November.Overall, the ice formation period in the study area is significantly shorter than the melting period.Figure 4 illustrates the linear trends of monthly sea-ice concentration during 2012-2021.As this area is predominantly covered by sea ice in winter, from December to the following April, there is no noticeable trend in sea ice concentration.Starting from May to June, the sea ice begins to melt in the study area.In June, the eastern region of Severnaya Zemlya shows a significant decreasing trend, with a rate of approximately -0.04/a.Notably, as mentioned earlier, the eastern part of the Vilkitsky Strait (Area (3) in Figure 1) is the location where the sea ice begins to melt earliest in the study area, and its concentration changes differently from nearby areas without a significant decrease trend.From June to September, the Kara Sea to the west of Severnaya Zemlya generally shows an increase trend, with an average increase rate of about 0.016/a, while the Laptev Sea to the east of Severnaya Zemlya shows a decrease trend with an average decrease rate of about -0.019/a.In October, the entire study area exhibits a significant decreasing trend with an average decrease rate of about -0.026/a.In November, the western part of Severnaya Zemlya has an average change rate of about 0.016/a, while the eastern part is perennially covered by sea ice, and the sea ice concentration no longer shows a significant changing trend.

Changes in Sea Ice Thickness
Figures 5 and 6 show the distribution of HIGHTSI simulated monthly sea ice thickness and its spatial average in the study area for 2012 to 2021.The sea ice thickness reached its maximum value of 1.97 meters in May.From May to June, the ice began to melt, as evidenced by Figure 5(f), where the ice first started to melt in the eastern and southwestern parts of the Vilkitsky Strait, consistent with the changes of ice concentration shown in Figure 2. The study area's average sea ice thickness diminished to its minimum value of 0.19 meters in September, after which the ice began a new growth cycle from October to the following April, consistent with the trends in concentration.Additionally, the sea ice thickness in different years followed a similar trend in winter, with the smallest standard deviation of 0.02 meters in January, while the largest differences occurred in summer, with the highest standard deviation of 0.33 meters in July.Overall, unlike the rapid growth in concentration during the freezing period, the increase in ice thickness was more gradual, taking longer time to grow thicker.The monthly trend of sea ice thickness in the study area from 2012 to 2021was calculated (Figure 7).The results show no significant change from January to March, but a noticeable decline trend began near the eastern coast of the North Archipelago in April, and intensified in May, reaching a low of approximately -0.14 meters per annum.From June to December, the overall thickness of sea ice in the study area declined, with the strongest average decrease trend in June and July, around -0.05 meters per annum.
Consistent with concentration data, the thickness simulation results also indicate a high-value area in the waters west of Bolshevik Island.To investigate the possible reason for the formation of this highvalue area, we calculated the study area's Freezing Degree Days (FDD), which measures the algebraic sum of the daily average temperature and baseline temperature differences over a specific period, assessing the coldness and duration of the cold condition.The distribution of monthly FDD is shown in Figure 8.A higher value of FDD represents a longer duration of low temperatures.It can be seen that a high-value area of FDD perennially exists in the area west of Bolshevik Island, with an annual average of 440.76°C, higher than the whole study area's 344.93°C.The high-value area's monthly average is higher than the whole area's average, with the ratio of the anomaly to the mean value being higher from May to October, peaking in July at 817.87%, while the ratio is lower than 40% in both winter and spring, with the lowest being 13.35% in December.The existence of this low-temperature area is a direct cause of the high-value area of sea ice concentration and thickness in this region.IOP Publishing doi:10.1088/1742-6596/2718/1/0120409 data and the simulated sea ice thickness data.Figure 9 shows the distribution of RIO for the CCS Ice Class B1 vessels (ice class of 'Yongsheng' -referred to as B1 class henceforth) in the study area from early July to late November 2021.In July 2021, the Vilkitsky Strait and Kara Sea regions were unsuitable for B1 class vessels to navigate without the escort of an icebreaker.Among these areas, the most severe ice conditions are found in the western part of the Vilkitsky Strait, which, as mentioned earlier, due to the high ice thickness and concentration, is challenging for vessels to pass through here safely.This high-risk area had an RIO value still reaching -5 in late July.The high-risk zone began to shrink in scope starting in August, and by early September, B1 class vessels could bypass this zone to pass through the Vilkitsky Strait.In September, the Vilkitsky Strait area was low-risk until early October when ice began to grow, and the high-risk zone in the western part of the strait reappeared.To enter the Kara Sea through the western part of the Vilkitsky Strait, B1 class vessels would have to require an icebreaker escort.The RIO values were averaged spatially for these three areas to understand the overall navigation risk of the Vilkitsky Strait, as shown in Figure 10.From 2012 to 2021, PC3 class vessels had RIO values greater than 0 throughout the year in these three areas, whereas B1 class vessels, influenced by the high sea ice concentration and thickness in the western Vilkitsky Strait, had higher navigation risks in Areas 1 and 2. They might need the escort of an icebreaker to safety navigate.However, in Area 3, during summer years with lighter ice conditions, there's a lower-risk navigable period.

Figure 1 .
Figure 1.Map of the Arctic marine area and the study area.

Figure 2 .
Figure 2. Distribution of monthly sea ice concentration during 2012-2021.The black lines are 0.15 concentration contour lines.Areas with concentrations below 0.15 represent open water in theory.

Figure 3 .
Figure 3. Seasonal cycle of spatial mean sea-ice concentration for study area during 2012-2021.The shaded regions show standard deviation.

Figure 5 .
Figure 5. Distribution of monthly sea ice thickness during 2012-2021.The black lines are the sea-ice concentration isolines with a value of 0.15.

Figure 6 .
Figure 6.Seasonal cycle of spatial mean sea ice thickness for study area during 2012-2021.The shaded regions show standard deviation.

3. 3 .
Changes in the Risk Index (RIO) Using the methods introduced in Section 2.3, the daily spatial distribution of RIO values for different ice classes in the study area from 2012 to 2021 were calculated and analyzed, based on the concentration ICFOST-2023 Journal of Physics: Conference Series 2718 (2024) 012040

Figure 9 .
Figure 9. Distribution of RIO index for CCS Ice Class B1 vessels from July to November in 2021.To clearly represent the navigational risk and its changes in the Vilkitsky Strait area, we divided the Vilkitsky Strait region into three smaller areas, as shown in the right side of Figure 1: west of the strait (77°-78°N, 94°-99.5°E),which covers the high concentration area mentioned earlier; the strait area (77.5°-78.5°N,99.5°-105°E); and east of the strait (77.5°-78.5°N,105°-110.5°E).For these regions, the commercial vessel 'Yongsheng' and the research vessel 'Xuelong 2' were used as examples.According to the ice class classifications used by POLARIS, PC3 ('Xuelong 2' class) and B1 class ('Yongsheng' class) were selected for risk index calculations.The RIO values were averaged spatially for these three areas to understand the overall navigation risk of the Vilkitsky Strait, as shown in Figure 10.From 2012 to 2021, PC3 class vessels had RIO values

Figure 10 .
Figure 10.Time series of RIO index for two classes vessels over three regions around the Vilkitsky Strait during 2012-2021.To further determine the navigability of the strait, the days of vessels can bypass high-risk areas were calculated.Whether the vessels of a specific ice class can safely navigate through the Vilkitsky Strait during a given period was determined by assessing whether the points with RIO less than 0 (i.e., safe navigation areas) can be connected.Four types of vessels: PC3 ('Xuelong 2' class), PC6 ('Xuelong' class), B1 ('Yongsheng' class), and non-ice strengthened vessel according to the IMO classification (≤ CCS Ice Class B, a common ice class for Chinese commercial vessels) were analyzed here.The results were shown in Figure 11.Icebreakers of class PC3 and above can navigate the Vilkitsky Strait almost all year, with navigable days exceeding 360 each year.For vessels with ice classes below PC3, navigable days decrease as the ice class decreases.Vessels with ice classes of PC6 and below can only navigate the Vilkitsky Strait in summer.In 2013, 2014, and 2021, when ice conditions were particularly severe, vessels with ice class below PC5 had less than 30 navigable days in the Vilkitsky Strait.In 2013, B1 and non-ice strengthened vessels couldn't navigate the Vilkitsky Strait at all without the escort of an icebreaker.

Figure 11 .
Figure 11.The number of navigable days for four classes vessels in Vilkitsky Strait from 2012 to 2021.

Table 2 .
Recommended speed limits for elevated risk operations