The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLVI-4/W2-2021
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-4/W2-2021, 71–76, 2021
https://doi.org/10.5194/isprs-archives-XLVI-4-W2-2021-71-2021
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-4/W2-2021, 71–76, 2021
https://doi.org/10.5194/isprs-archives-XLVI-4-W2-2021-71-2021

  19 Aug 2021

19 Aug 2021

MARITIME BIG DATA ANALYSIS WITH ARLAS

W. Gautier, S. Falquier, and S. Gaudan W. Gautier et al.
  • Data Science Department, Gisaïa, 2 avenue de l’Escadrille Normandie Niemen, 31700 Blagnac, France

Keywords: Maritime Data, Automatic Identification System (AIS), Vessels Traffic, Origin Destination, Big Data, Open Source, Data Visualization, Machine Learning

Abstract. The maritime industry has become a major part of globalization. Political and economic actors are meeting challenges regarding shipping and people transport. The Automatic Identification System (AIS) records and broadcasts the location of numerous vessels and delivers a huge amount of information that can be used to analyze fluxes and behaviors. However, the exploitation of these numerous messages requires tools based on Big Data principles.

Acknowledgement of origin, destination, travel duration and distance of each vessel can help transporters to manage their fleet and ports to analyze fluxes and focus their investigations on some containers based on their previous locations. Thanks to the historical AIS messages provided by the Danish Maritime Authority and ARLAS PROC/ML, an open source and scalable processing platform based on Apache SPARK, we are able to apply our pipeline of processes and extract this information from millions of AIS messages. We use a Hidden Markov Model (HMM) to identify when a vessel is still or moving and we create “courses”, embodying the travel of the vessel. Then we derive the travel indicators.

The visualization of results is made possible by ARLAS Exploration, an open source and scalable tool to explore geolocated data. This carto-centered application allows users to navigate into the huge amount of enriched data and helps to take benefits of these new origin and destination indicators. This tool can also be used to help in the creation of Machine Learning algorithms in order to deal with many maritime transportation challenges.