The retreat of sea ice has been found to be very
significant in the Arctic under global warming. It is projected to continue
and will have great impacts on navigation. Perspectives on the changes in
sea ice and navigability are crucial to the circulation pattern and future
of the Arctic. In this investigation, the decadal changes in sea ice
parameters were evaluated by the multi-model from the Coupled Model
Inter-comparison Project Phase 6, and Arctic navigability was assessed under
two shared socioeconomic pathways (SSPs) and two vessel classes with the
Arctic transportation accessibility model. The sea ice extent shows a high
possibility of decreasing along SSP5-8.5 under current emissions and climate
change. The decadal rate of decreasing sea ice extent will increase in March
but decrease in September until 2060, when the oldest ice will have
completely disappeared and the sea ice will reach an irreversible tipping
point. Sea ice thickness is expected to decrease and transit in certain
parts, declining by
The Arctic has experienced significant warming since the 1970s (Connolly et al., 2017). Along with the increasing surface air temperature, Arctic communities have experienced unprecedented changes, such as reduction of sea ice extent and thickness, loss of the Greenland ice sheet, decrease in snow coverage, and thawing of permafrost (Biskaborn et al., 2019; Box et al., 2019; Brown et al., 2017; Loomis et al., 2019). The sea ice extent has declined at a rate of approximately 3.8 % per decade. In comparison, perennial ice had a higher proportion of loss of approximately 11.5 % per decade during the period from 1979 to 2012 (Comiso and Hall, 2014). The average ice thickness near the end of the melt season decreased by 2.0 m or 66 % between the pre-1990 submarine period (1958–1976) and the CryoSat-2 period (2011–2018) (Kwok, 2018). Continued declines in sea ice have been projected by the Coupled Model Inter-comparison Project Phase 5 in the Arctic through the end of the century (Meredith et al., 2019).
Sea ice reflects a significant fraction of the solar radiation because it has a high albedo. It also reduces the heat transfer between the ocean and the atmosphere as it acts as an insulator (Screen and Simmonds, 2010). With the retreat of sea ice, thermohaline circulation has changed (Jourdain et al., 2017), and global warming has intensified (Abe et al., 2016). However, climate change has led to prolonged open water conditions and large-scale Arctic shipping that will involve ice channels (Barnhart et al., 2015; Huang et al., 2020a). The Northern Sea Route (NSR) extends along the northern coast of Eurasia from Iceland to the Bering Strait, which shortens the transit distance by approximately 15 %–50 % relative to the southern routes through the Suez Canal (Buixadé Farré et al., 2014). It is navigable for approximately 3 months per year for ice-strengthened ships at the end of summer and the beginning of autumn (Yu et al., 2021). The end of shipping season for open water (OW) vessels has reached 24 October since 2010 (Chen et al., 2019). However, navigability is still affected by the ice regime, such as ice thickness and concentration, around the Severnaya Zemlya islands, the Novosibirsk islands, and the East Siberian Sea. The Northwest Passage (NWP) follows the northern coast of North America and crosses the Canadian Arctic archipelago. Compared to the traditional Panama Canal route from western Europe to the Far East, the NWP shortens the transit distance by 9000 km (Howell and Yackel, 2004). The shortest navigable period was up to 69 d during 2006–2015 (Liu et al., 2017), and the first time of its being completely free of ice was reported to occur in September 2007 (Cressey, 2007). Geographical and political factors also pose some challenges to the navigability of passages and choice of routes (Ryan et al., 2020). The straits along the NWP are at times narrow and shallow, which are easily clogged by free floating ice. NSR is greater than NWP in terms of geography, while it still has several choke points where ships must pass through shallow straits between islands and the Russian mainland (Ostreng et al., 2013). Apart from the geographical factor, the various organizations and groups formed between the surrounding Arctic nations, as well as the disputes and agreements, give impetuses for adopting the NSR. Russia has committed several large infrastructure projects to support the NSR, such as the Yamal-Nenets railway and emergency rescue centers (Serova and Serova, 2019). China, which is characterized as a near-Arctic state, also outlined the plans to build a Polar Silk Road by building infrastructure and conducting trial voyages (Tillman et al., 2019). For the development of socioeconomics and marine transportation, future projections of ice conditions and Arctic passages are increasingly important, for which climatic changes should be considered (Gascard et al., 2017). Smith and Stephenson (2013) investigated the potential of Arctic passages under representative concentration pathway (RCP) 4.5 and RCP 8.5 and found that OW ships and Polar Class 6 (PC6) ships (Table 1) will be able to cross NSR and NWP in September by the mid-century, respectively. The areas of the Arctic accessible to PC3, PC6, and OW ships would rise to 95 %, 78 %, and 49 %, respectively, of the circumpolar International Marine Organization Guidelines Boundary area by the late 21st century (Stephenson et al., 2013). Melia et al. (2017) suggested that the Arctic passages from Europe to Asia would be 10 d faster than conventional routes by the mid-century and 13 d faster by the late century. Recent research has shown that NSR might be accessible earlier for OW ships in September 2021–2025, and the navigable window would extend to August–October during 2026–2050 under shared socioeconomic pathways (SSPs) 2–4.5 (Chen et al., 2020). However, evaluating sea ice conditions and Arctic navigability by a single climate model, even one with a higher resolution, is insufficient.
This prospective study was designed to obtain further insight into the future changes in sea ice in the Arctic and the navigability of the Arctic during this century with up-to-date ensemble climate models in the Coupled Model Inter-comparison Project Phase 6 (CMIP6). To reduce uncertainties of a single high-resolution model and multi-model average, models were filtered by comparing the historical simulations and observations of sea ice extent, and the possible SSPs were investigated with the average of multiple models. The distributions of the linear trend of sea ice extent, concentration, and thickness were explored in three stages (2021–2040, 2041–2060, and 2061–2100). In addition, the changes in sea ice volume and age were analyzed. The accessibility of the Arctic and the navigable area were evaluated with the Arctic Transportation Accessibility Model (ATAM) from the Arctic Ice Regime Shipping System (AIRSS) for OW ships and PC6 ships under SSP2–45 and SSP5–85 in 2021–2030 and 2045–2055.
The new scenario framework SSP in CMIP6 was designed to carry out research on climate change impacts and adaption by combining pathways of future radiative forcing and climate changes with socioeconomic development (O'Neill et al., 2014). SSP1 indicates a sustainable development, which proceeds at a reasonably high pace. Technological change is rapid, and inequalities are lessened and directed toward environmentally friendly processes. Unmitigated emissions are high in SSP3. It is due to a rapidly growing population, moderate economic growth, and slow technological change in the energy sector. SSP2 is an intermediate case between SSP1 and SSP3. SSP5 occurs in the absence of climate policies, energy demand is high, and most of this demand is met with carbon-based fuels.
Compared with CMIP5 models, the CMIP6 multi-model ensemble mean provides a
more realistic estimate of the Arctic sea ice extent (SIMIP Community,
2020), but the biases of the models are still large (Shu et al., 2020). This
study selected models by comparing the historical trend of Arctic sea ice
extent in simulations with remote sensing observations during 1979–2012. The
observation data come from the Sea Ice Index of the National Snow and Ice Data
Center. The selected models are those that have a correlation coefficient
between the original simulations and observations greater than 0.8 (0.7 for
March). Five-point moving averages of the simulated and observed sea ice
extent are displayed in Fig. 1. The models passing the test are CESM2,
MPI-ESM1-2-HR, MPI-ESM1-2-LR, NorESM2-LM, NorESM2-MM, ACCESS-ESM1-5,
AWI-CM-1-1-MR, and AWI-ESM-1-1-LR in September and CESM2, MPI-ESM1-2-LR,
ACCESS-ESM1-5, AWI-CM-1-1-MR, INM-CM5-0, MPI-ESM-1-2-HAM, and AWI-ESM-1-1-LR
in March. The mean of the selected models corresponds well with the
observations, and the correlation coefficients are 0.884 and 0.817 in
September and March, respectively. However, sea ice datasets in SSP1-2.6,
SSP2-4.5, SSP3-7.0, and SSP5-8.5 after 2020 have not been released for
CESM2, MPI-ESM-1-2-HAM, and AWI-ESM-1-1-LR until now. In addition,
AWI-CM-1-1-MR was excluded from analyzing the navigability of the Arctic in
the absence of sea ice concentration. The spatial resolution of monthly sea
ice concentration and thickness was normalized to
The observations and five-point moving averages of sea ice extent in March and September during 1979–2012.
Safety and pollution are two of the opposite factors considered in
developing regulatory transport standards. AIRSS was designed to minimize
the risk of pollution in the Arctic due to damage to vessels by ice
(Transport Canada, 1998). ATAM, developed by AIRSS, is commonly used to
quantify the temporal and spatial accessibility in the Arctic, in which
the ice number (IN) represents the ability of a ship to enter ice-covered
water:
Vessel classes versus operating ice thickness.
The extent and area are the most reliable products of sea ice from satellite
retrieval (Comiso, 2012; Notz, 2014). Therefore, the sea ice extent was taken
as an indicator to evaluate models and future scenarios. As shown in Fig. 2, the observation trend was made with least square regression of sea ice
extent from 1979 to 2019, in which sea ice might completely disappear in
September after 2073. In addition to the classical pathways, such as
SSP1-2.6, SSP2-4.5, and SSP5-8.5, CMIP6 provides a variety of new
selections. However, SSP1-1.9, SSP4-34, and SSP4-6.0 were not discussed in
the multi-scenario evaluation for the less common models. According to
historical development and scenarios, sea ice will retreat in the future
with a more significant decreasing trend in September. The difference
between SSPs and observation trends is greater in March than in September,
while both have large dispersions among pathways after 2050. Compared with
others, SSP5-8.5 has the greatest correlation coefficients, which are 0.784
and 0.712 in September and March, respectively, with the observation trend;
SSP2-4.5 comes second. This suggests that Arctic sea ice might be the worst
scenario in the future under the current emission and climate change trends.
The Arctic is regarded as “ice free” when the sea ice area is less than 1 million km
Sea ice extent under multiple scenarios and observation trends in March and September.
“Ice free” was taken as one of the tipping points of climate change with
significant irreversible effects (Lenton et al., 2019). Three stages were
extracted for the changes in sea ice extent in Fig. 3. Decadal linear
trends and probability distributions with an interval of 0.4 million km
Linear trends and probability distributions (PDs) of Arctic sea ice extent (SIE) in March and September.
In addition to the extent and area, thickness, concentration, volume, and
age are important indicators of changes in sea ice in the future. Figures 4
and 5 show the linear trends of ice thickness and concentration and the
changes in sea ice volume and age, respectively, under SSP5-85 in
2021–2100. Ice thickness has a negative trend within the Arctic
Archipelago, in coastal water, and in the sector to the north of the Arctic
Archipelago and Greenland in September, while the other parts will slightly
increase in the next 20 years. The trend is reversed in the Arctic Ocean,
and the decreasing area near the shore will extend to the north in
2041–2060, after which almost all sea ice will be reduced with an average
trend of
Linear trends of sea ice thickness and concentration under SSP5-85 in September.
The changes in sea ice volume and age under SSP5-85.
With the retreat of sea ice, the possibility for navigation is rising in the Arctic. The opening of passages will be profitable for ocean shipping companies (Chang et al., 2015). The most likely navigable window is in September. Figure 6 shows Arctic accessibility for the OW ships under SSP5-8.5 in September. Panel (a) indicates that the probability of crossing NSR and NWP is low in the next 10 years. The impassable areas for NSR are mainly in the East Siberian Sea and northwestern Laptev Sea, but nearshore waters might be navigable for vessels with shallow drafts. Four crucial straits, the Vilkitsky Strait, Shokalsky Strait, Dmitry Laptev Strait, and Sannikov Strait, are accessible for OW ships. NWP is impassable in the sectors west of Banks Island and Queen Elizabeth Island, as well as the M'Clure Strait, Viscount Melville Sound, Barrow Strait, and Lancaster Strait within the Parry Channel. All routes provided in the Arctic marine shipping assessment report (AMSA, 2009) are under restrictions for OW ships. By the mid-century, both NSR and NWP will open for OW ships under SSP5-8.5 in September.
INs for OW ships under SSP5-8.5 in September.
The opening of the Arctic passages mainly depends on the connectivity among
grid cells, during which the overall navigable potential in a region can be
measured by the percentage of accessible grid cells with total grid cells.
Figure 7 displays the Arctic navigable grid cells for OW ships and PC6 ships
under SSP2-4.5 and SSP5-8.5 in 2021–2030 and 2045–2055. It is the
percentage of grid cells in which INs are greater than 0. The totally navigable
percentage for OW ships is shown as a unimodal curve in both stages, with
the peak in September and the valley in April and March. It is an irregular
curve for PC6 ships with the minimum value in June. The maximum values are
shown in October 2021–2030, while they will have a range in November and December by
the mid-century. Actually, the Arctic would be navigable for PC6 ships from
October to December. It is very strange that an abnormal decrease occurs in
September in 2045–2055. The navigable grid cells within every 5
The percentage of totally navigable grid cells for OW ships and PC6 ships under SSP2-4.5 and SSP5-8.5.
The percentage of navigable grid cells for OW ships and PC6 ships under SSP2-4.5 and SSP5-8.5 within different latitudes.
The Arctic warming rate is more than double the global average, and it has
had great impacts on the Arctic and globe (Cohen et al., 2020). This paper
investigated the future changes in sea ice and navigability of passages in
the Arctic under two kinds of shared socioeconomic pathways. It provides a
vision of the earth's future and has great significance for navigation
planning. The following results were found.
The changes in sea ice would occur along SSP5-8.5 with a higher
possibility under the current trend. “Ice free” might appear in September
2060, and sea ice would completely disappear by the end of the century. The retreat of sea ice is more significant in September before 2060,
after which the decline is mainly shown in March. The decadal rate of sea
ice extent will increase under SSP5-8.5 in March, while it will decrease in
September. The decrease in sea ice thickness will transit from the Arctic Ocean
north of the Arctic Archipelago and Greenland to the seas along Russia and
North America and will totally decline with an average decadal trend of
Sea ice volume will decrease at a higher decadal rate in March than in
September. The oldest ice might eventually disappear by approximately the
mid-century. First-year ice dominates the sea ice cover. It increases mainly
before 2060 and remains stable until 2090, after which it starts to
decrease. The probability for OW ships crossing NSR and NWP is low in 2021–2030,
while it is high in August–October 2045–2055, with maximum and minimum
navigable grid cells in September and March, respectively. The passages along the coast and crossing the Arctic might open for PC6
ships during October–December and September–October 2021–2030,
respectively, with maximum navigable grid cells in October. The open
window would extend to August–January and October–January in 2045–2055,
respectively, and the maximum navigable grid cells have a range in November and
December.
The navigable window for OW ships and PC6 ships along the NSR were
investigated in our previous work (Chen et al., 2020), but it is insufficient
to evaluate Arctic navigability by a single climate model, even with a high
resolution. This study serves as a reference for future changes in sea ice
and navigability in the Arctic, including NSR, NWP, and central passage.
However, the uncertainty of the models might have affected the results and
their reliability in this research. Approximated physical processes and
unreal parameters in models are inevitable problems in the geosciences.
Differences still existed even when the models were filtered by comparing
the historical simulations with the observations of sea ice extent. The
abnormal decrease in navigable area at high latitudes (80–90
All the data used in this paper are
available online. The simulations of sea ice can be accessed from the CMIP6
(
JinC and SK developed the concept and investigated the methods of this paper. JinC and WD analyzed the data and wrote the original draft. JG, MX, YZ, XZ, WZ, and JizC reviewed and edited the manuscript.
The contact author has declared that neither they nor their co-authors have any competing interests.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We are thankful for the data from CMIP6 and NSIDC. Our cordial gratitude should be extended to the reviewers and the editors for their professional and pertinent comments on this paper.
This work was financially supported by the International Partnership Program of Chinese Academy of Sciences (grant no. 131B62KYSB20180003), the Key Research Program of Frontier Sciences, CAS (grant no. QYZDY-SSW-DQC021), the Frontier Science Key Project of CAS (grant no. QYZDJ-SSW-DQC039), the China National Key Research and Development Program (grant no. 2020YFA0608500), and the State Key Laboratory of Cryospheric Science (grant no. SKLCS-ZZ-2021).
This paper was edited by Yevgeny Aksenov and reviewed by Bjørn Åge Hjøllo, Hajime Yamaguchi, Luofeng Huang, and one anonymous referee.