Search

The global shipping industry faces increasingly complex safety challenges due to the rapid growth of international maritime trade. This study develops a novel framework that combines spatial density analysis and machine learning (i.e., extreme gradient boosting model) to investigate the evolutionary patterns of global maritime accidents during 2001–2020 from both spatial and temporal dimensions, and then identifies key environmental risk factors affecting maritime safety. The results show that the number of global maritime accidents exhibits fluctuations between 2001 and 2019, with a significant decrease observed in 2020. Furthermore, the distribution of global maritime accidents shows significant spatial variation over different time periods. Denmark's sea areas have high accident rates between 2001 and 2005, while concentrated accidents are observed in the seas around the United Kingdom, Denmark, and China between 2006 and 2010. From 2011 to 2015, Europe's accident-prone areas increase, but fewer accidents are reported along China's east coast. The Strait of Malacca is also an accident-prone area from 2016 to 2020. In addition, wave height, sea surface temperature, wind speed, water depth, and precipitation are identified as key environmental risk factors affecting maritime safety. These findings can inform strategies and mitigation plans to improve navigational safety in the global shipping industry. • A framework combining spatial density analysis and machine learning is developed. • The evolutional patterns of global maritime accidents during 2001–2020 are analyzed. • Global maritime accidents show spatial variation across different time periods. • The key environmental risk factors are identified based on the importance scores. [ABSTRACT FROM AUTHOR]