Arctic maritime operations have significantly increased in recent decades, driven by various activities with complex spatial and temporal mobility patterns. The harsh conditions—sea ice, hazardous weather, remote infrastructure, and limited communication—pose substantial risks to human safety and the environment. To mitigate these, the International Code for Ships Operating in Polar Waters (Polar Code) was implemented in 2017. The Polar Code categorizes vessels based on anticipated sea ice conditions and sets a design temperature threshold, defining hazardous conditions narrowly by temperature and sea ice. However, it overlooks other crucial metocean parameters like wind, waves, sea-spray icing, and visibility, which are vital for safe Arctic operations. The Polar Code primarily relies on climatological data for risk assessment, neglecting the importance of readily available real-time environmental information. This study leverages a decade-long Automatic Identification System (AIS) dataset integrated with weather and sea-ice data to analyze Arctic shipping activities and trends from 2013 to 2022. The objectives are to (1) provide a pan-Arctic overview of shipping activities and trends, and (2) assess where ships encountered hazardous conditions as defined by the Polar Code, identifying its limitations. The study aims to highlight the inadequacy of the Polar Code's definition of hazardous conditions and propose improvements by incorporating real-time environmental information systems.

Existing literature documents the rise in Arctic shipping and its associated challenges. Studies quantify Arctic shipping activity using satellite monitoring (AIS data), highlighting the complexity of maritime operations and their diverse purposes. Research emphasizes the need for risk mitigation strategies, considering the harsh environment and the limited infrastructure. The Polar Code's implementation has been examined, focusing on safety requirements, manning, and training. However, there's a gap in research regarding the actual effectiveness of the Polar Code in mitigating risks given the increasing frequency of operations in challenging conditions, particularly during winter. The lack of detailed analysis of the correlation between shipping activity and diverse hazardous conditions (beyond sea ice and temperature) is a significant gap in existing literature. This study aims to fill that gap by combining AIS data with extensive environmental data to analyze shipping activity in the context of various hazardous conditions, evaluating the Polar Code's efficacy.

This study uses a decade-long (2013-2022) dataset of Automatic Identification System (AIS) data from the Arctic Ship Traffic Data (ASTD) provided by the Arctic Council's Protection of the Arctic Marine Environment (PAME) working group. The ASTD Level 2 data comprises millions of messages with information on ship location, time, type, flag, etc., from thousands of ships. The data was quality-controlled to create consistent ship tracks, extracting information on daily mean position, ship type, and size. Ship types were categorized into seven distinct groups. Sea ice concentration data was obtained from a daily product provided by the University of Bremen using AMSR-2 passive microwave sensors (6.25 km resolution). Atmospheric data (10-meter wind speed and 2-meter temperature) were sourced from the ERA5 atmospheric reanalysis (T639 resolution, ~36 km). Daily mean sea surface temperature, daily minimum 2-meter temperature, and maximum 10-meter wind speed were used. Daily AIS ship positions were co-located with sea ice concentration and atmospheric reanalysis data. Shipping activity was analyzed across seven regions: pan-Arctic, Northern Sea Route (NSR), Svalbard, East Greenland, West Greenland, Northwest Passage, and Chukchi-Bering Sea. The main metric was shipping days per month, representing the time ships were exposed to specific conditions. Linear regression was used to determine trends, considering only trends with a 98% confidence level (p-values <0.02). Sea-spray icing was calculated using a simplified algorithm from Overland (1990), based on wind speed, air temperature, and sea surface temperature, classifying icing into light, moderate, heavy, and extreme categories. The Polar Service Temperature (PST) was calculated using the mean daily minimum temperature (MDLT) from a ten-year climatology. The study considered uncertainties related to AIS data limitations, warm bias in ERA5 over sea ice, spatial resolution of the sea ice product, and limitations of the simplified sea spray icing algorithm.

Arctic shipping activity significantly increased from 2013 to 2022, with an annual growth rate of 7% in shipping days. Excluding the fishing sector, the growth rate increased to 12%. The NSR and the Chukchi-Bering Sea showed the highest shipping activity. Shipping patterns shifted from seasonal to year-round, especially along the NSR due to LNG and oil projects, with winter-spring activity tripling. The number of shipping days in close ice (sea ice concentration >80%) increased from 150 to 500 per month on average and even more dramatically during winter. In 2021, over 6300 shipping days occurred in close ice, primarily by bulk carriers, other vessels and crude oil tankers. Shipping activities under low temperatures (-20°C and below) similarly increased from 50 to 140 days per month on average, with winter peaks reaching 500-800 days. This increase is primarily due to NSR winter sailings. Analysis of sea-spray icing showed that nearly all hazardous icing conditions in the Polar Code region occurred in the Barents and Bering Seas, impacting fishing vessels disproportionately. A fishing vessel capsizing incident highlights the real-world consequences of severe sea-spray icing. The study found that the Polar Code’s reliance on a climatological approach for risk assessment is insufficient, as illustrated by a case study of numerous ships becoming trapped in unexpected sea ice along the NSR. This event, which occurred despite new ice-class requirements, underscores the limitation of climatological data in predicting short-term variations in sea ice.

The findings reveal a considerable increase in Arctic shipping activity and exposure to hazardous conditions, exceeding the Polar Code's current limitations. The reliance on climatological data for risk assessment proves inadequate for predicting rapid changes in Arctic weather and sea-ice conditions. The substantial increase in winter shipping and incidents like the October 2021 NSR incident demonstrates the need for incorporating real-time monitoring and forecasting data. The study highlights the need to expand the definition of hazardous conditions beyond sea ice and temperature to include wind, waves, sea-spray icing, and visibility, considering their significant impact on Arctic navigation. The integration of existing international maritime information services (WWMIWS) with the Polar Code could improve access to and utilization of essential weather and sea ice information. The study’s findings emphasize the urgent need for a more comprehensive and flexible regulatory framework that accounts for the complexity of the Arctic environment and the evolving nature of shipping operations.

This study demonstrates a significant increase in Arctic shipping activity and its exposure to hazardous conditions, exceeding the scope of the current Polar Code's definition of hazards. The study underscores the inadequacy of solely relying on climatological data for risk assessment. Recommendations for refining the Polar Code include incorporating more comprehensive hazardous conditions (wind, waves, visibility, sea spray icing), mandating the use of real-time monitoring and forecasting systems, and integrating existing international information services. Future research could focus on developing more precise algorithms for predicting hazardous conditions, investigating the usability of updated forecast models, and creating user-oriented support services to address specific end-user needs in different Arctic regions and seasons.

The study acknowledges limitations inherent to the data used. AIS data may underestimate the total ship traffic, particularly smaller vessels. The ERA5 reanalysis has a known warm bias over sea ice, potentially underestimating exposure to low temperatures. The spatial resolution of the sea ice product might lead to misclassification of ships near the sea ice edge. The simplified sea spray icing algorithm doesn’t account for wave conditions or ship characteristics, providing only estimates of icing intensity. Future research should address these limitations by incorporating more comprehensive datasets and refining the prediction models.