Ships are projected to navigate whole year-round along the North Sea route by 2100

The Arctic is highly sensitive to climate change, with sea ice extent, thickness, and concentration decreasing, especially in summer and autumn. An ice-free Arctic in September is foreseeable under all scenarios, potentially opening Arctic passages. Arctic routes could reduce voyage distance by ~40% and time by ~30% between Europe and northwestern Asia, cutting fuel consumption and emissions and improving global transport efficiency. Given the large Europe–northwestern Asia trade share, opening Arctic passages could change global trade patterns and the international division of labor. Previous work often assessed navigability using factors like sea ice thickness (SIT) and concentration (SIC), weather, and water depth, and sometimes highlighted sea ice motion (SIM) and drift as key risks. However, most long-term projections used open-water (OW) or Polar Class (PC) 6 ships, leaving two gaps: (1) navigability assessments typically considered only SIT and SIC, and (2) the widely used thin-strengthened PC7 class has been seldom evaluated. This study fills these gaps by analyzing temporal and spatial changes of sea ice and navigability of Arctic passages (2023–2100), validating CMIP6 simulations with observations, focusing on PC7 and OW ships, and assessing navigable days (ND), navigable periods (NP), navigable areas, and optimal route distributions considering SIC, SIT, and sea ice dynamics.

Existing research on Arctic maritime transport feasibility has largely focused on environmental impacts on navigation, including meteorological and hydrological risks such as strong winds, low visibility, low temperatures, and shallow waters. Water depth has been identified as a key constraint. Regulatory and assessment frameworks like POLARIS and AIRSS typically incorporate SIT and SIC. Some studies emphasize sea surface temperature and sea ice motion/drift as critical to feasibility, given their influence on ship–ice interactions and heading, especially for large carriers. Recent efforts link long-term climate change to changes in sea ice and Arctic navigability, often using OW and PC6 ships to project accessibility and optimal routes; however, these often omit SIM in risk calculations and rarely evaluate PC7 ships, which are widely used. This paper responds to these gaps by integrating SIM alongside SIT and SIC and by explicitly assessing PC7 navigability.

Study area included the three recognized Arctic passages: the Northern Sea Route (NSR), Northwest Passage (NWP), and Transpolar Sea Route (TSR). Data selection and validation: CMIP6 models (2015–2100) providing SIT, SIC, SIM, and water depth were evaluated against observations (NSIDC SIC and SIM; AWI CS2SMOS SIT) at 25×25 km resolution for 2015–2023. CMIP6 outputs were interpolated to 25×25 km using inverse distance weighting. Model performance was quantified with a Distance between Indices of Simulation and Observation (DISO) metric combining, for SIT and SIC, correlation coefficients and RMSE, and for SIM, RMSE (speed) and MAE (angle), then normalized and aggregated. SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5 were compared; SSP2-4.5 had the lowest overall DISO and was selected for detailed projections. Temporal partitions were defined based on SIT/SIC trends: 2023–2040, 2041–2064, 2065–2100; monthly analyses emphasized May–July, Aug–Oct, and Nov–Dec. Navigability modeling: The POLARIS framework was used to compute risk index outcomes (RIO) from ice regimes, combining ice-type SIC (in tenths) with risk index values (RIV) determined by SIT bands and vessel class for PC7 and OW ships without icebreaker escort. To incorporate dynamic ice hazards, a comprehensive risk index (COMRIV) integrated RIO and SIM, applying a SIM threshold of 0.15 m/s; if SIM exceeded the threshold, a normalized negative risk (−1 to 0) was applied irrespective of vessel type. COMRIV combined normalized RIO and SIM with weights 0.75 and 0.25, respectively. Navigability was defined when COMRIV was non-negative. Route optimization: An A* least-cost path algorithm in a gridded GIS network computed Optimal Arctic Shipping Routes (OASRs) between Rotterdam (51°5′N, 4°30′E) and the Bering Strait (65°38′36″N, 169°11′42″W) using travel time as cost on navigable cells (non-negative COMRIV). Navigable Days (ND) were days with an OASR present; Navigable Periods (NP) were sequences of continuous NDs. Route density was summarized monthly/seasonally, and correlations with SIC were assessed. Emission scenarios SSP1-2.6, SSP2-4.5, and SSP5-8.5 were used to compare ND trajectories through 2100, with primary emphasis on SSP2-4.5.

- Model validation and scenario choice: Among CMIP6 scenarios, SSP2-4.5 had the lowest DISO (best agreement) overall (SIT, SIC, SIM) and was selected for projections. SIT peaks in May and is lowest in October, lagging SIC; both SIT and SIC decline from 2023 to 2100 with fluctuations. - Effect of sea ice motion (SIM): SIM substantially increases navigational risk, reducing navigability. Navigable grid proportions across four cases (PC7/OW × without/with SIM) were approximately 70.33%, 41.51%, 13.20%, and 0.01%, respectively, indicating drastic reductions when SIM is considered, especially for OW ships. - Spatial navigability (PC7): Greenland and Barents Seas are navigable May–Dec across periods. 2023–2040: Aug–Oct, most of the Arctic Ocean is navigable except eastern Central Arctic; Nov–Dec, coasts of Chukchi, Beaufort, Baffin Bay, and Kara Seas are navigable; May–Jul largely non-navigable with most of NSR closed. 2041–2064: May–Jul mostly non-navigable; Aug–Dec coastal areas largely risk-free; western Central Arctic remains non-navigable. 2065–2100: May–Jul non-navigable except Barents Sea; Aug–Dec, the entire Arctic Ocean becomes navigable. - Spatial navigability (OW): Smaller navigable areas than PC7. 2065–2100 Aug–Oct sees expansion including the Kara Sea and most of Baffin Bay. NWP remains persistently risky, though high-risk coverage decreases from 2023 to 2100. NSR shows higher likelihood of navigability than NWP; some coastal areas hover just below safe thresholds (−0.2 to 0), suggesting potential opening post-2100. - Navigable days (ND) and trends: Under SSP2-4.5, average annual ND rises for PC7 from 199 (2023) to 301 (2100), +1.31 days/year, and for OW from 195 to 247, +0.67 days/year. By 2100, average ND for PC7 is ~240 under SSP1-2.6 and ~283 under SSP5-8.5; for OW ~210 (SSP1-2.6) and ~248 (SSP5-8.5). - Optimal Arctic Shipping Routes (OASRs): For 2065–2100, OASRs concentrate in the TSR from Aug–Oct, with 14 distinct routes. September shows greatest centralization toward the Central Arctic when SIC is smallest; route density correlates strongly and negatively with SIC (r = −0.95, p < 0.01). OASRs align more with coastal areas in July and November. NSR supports OASRs July–November; NWP supports OASRs only in July and November. - Overall projection: PC7 ships achieve stable, safe summer–autumn navigation with potential year-round navigation from 2065 onward; by 2100, ND ~301 (PC7) and ~247 (OW) under SSP2-4.5. NSR holds a comparative advantage over NWP, particularly for OW vessels.

The study links projected Arctic sea ice decline to navigability and routing for PC7 and OW ships through 2100. Incorporating sea ice motion (SIM) reveals notably higher risks and reduced navigable areas than assessments based on SIT and SIC alone, directly addressing a key research gap. The findings indicate that, under SSP2-4.5, Arctic maritime access will expand seasonally and spatially, with PC7 vessels potentially operating year-round after 2065, and OW ships gaining but still facing constraints, especially along the NWP. The resulting optimal routes increasingly utilize central Arctic corridors (particularly in September when SIC is minimal) but remain coastal at the shoulders (July, November). The NSR emerges as more consistently feasible than the NWP, implying a reshaping of global shipping networks by enabling shorter trans-Arctic connections between Asia and Europe. These outcomes are significant for logistics planning, trade competitiveness, and policy initiatives such as the "Polar Silk Road," and they suggest that progress toward transport-related sustainability goals may be achievable, albeit with environmental and risk management caveats.

This paper validated CMIP6 outputs against observations using a composite DISO metric, selected SSP2-4.5 as the best-performing scenario, and projected Arctic navigability for PC7 and OW vessels from 2023 to 2100 using POLARIS augmented with sea ice motion and an A* routing framework. It closes gaps in prior work by: (1) accounting for SIM in navigability risk, and (2) assessing the widely used PC7 class. Key contributions include quantified ND growth (PC7 to ~301 days by 2100; OW to ~247 days), detailed spatial-temporal patterns of navigability (notably Aug–Dec expansion), and OASR distributions focused on TSR in late summer/early autumn with strong SIC dependence. The NSR exhibits a comparative advantage over the NWP, especially for OW ships. These results suggest substantial impacts on global shipping networks and trade by late century. Future research should integrate environmental impact assessments (pollution, spills, accidents), vessel design and ice-class cost tradeoffs, icebreaker escort strategies, and geopolitical uncertainties to holistically evaluate Arctic shipping viability and sustainability.

- When SIM is included, navigable grid proportions become very low for OW vessels (as low as ~0.01%), limiting the analysis of ND/NP/OASRs under dynamic-ice-constrained conditions. - Results depend on CMIP6 model fidelity and the DISO-based selection (SSP2-4.5); model biases and scenario uncertainties remain. - The POLARIS-based risk indices and SIM threshold (0.15 m/s) introduce methodological assumptions; effects of alternative thresholds or weighting schemes were not explored. - Economic factors, vessel-specific design/capacity trade-offs, and the potential role and cost of icebreaker escort were not modeled. - Non-environmental uncertainties (e.g., geopolitical events) and broader ecosystem impacts were acknowledged but not explicitly incorporated. - Route optimization used a specific origin–destination pair (Rotterdam–Bering Strait); alternate OD pairs and operational constraints (e.g., port access, regulations) were not assessed.